Home |
Search |
Today's Posts |
#122
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"chung" wrote in message
... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "Harry Lavo" wrote in message ... "Chung" wrote in message ... jonrkc wrote: snip I also recorded it through my DTI Pro / Proceed DAC, so I was using my system DAC, not the internal Marantz DAC. So it was an excellent recording setup, but not one too expensive (via eBay) for the average audiophile. Now here's the question to you: Why does the majority of critical listeners prefer digital? snip Before I get inundated, I made a mistake here and gave you my CD recording setup, which is taken from the digital out of the Proceed after passing from CD player to DTI-Pro to DAC. For SACD, I used analog out on the Sony, into my preamp, and through the tape outs to the Marantz analog ins. Usually I listen to the Marantz through its analog output, which is very fine. I can also switch the digital out into the DTI Pro/Proceed Combo but in the case of the Marantz, it adds very little. Well, maybe you are setting yourself up to get inundated again. You are saying that when you copy from CD to a blank CD, you take the digital out of the CD player to DTI-Pro to DAC? What exactly are you talking about? You have these expensive gear and can't afford a CD burner, which is standard now in just about any PC sold? I have two PC's with three burners. I use them to copy CD's or DVD's when I just want a casual copy....they are not in my audio system. I burn CD's IN my audio system when I want the best quality...from the player to the DTI Pro in real time which noise shapes them to 20 bits and de-jitters them, then feeds the signal back out from the Proceed via the sp/dif connection to the Marantz CD burner (which uses older, slower, but higher quality CD-R blanks). You know, Harry, this speaks volumes about the mentality of some subjectivists. I know that you are not alone. The goal is to make a copy of a CD, or certain tracks on a CD, onto a blank CD-R. Normal people use a computer to rip the tracks, and use software, like EAC, to insure that the data is ripped as accurately as possible. Then they use mastering software, such as Nero, to insure that the data is written correctly (verifiably so) to the CD blank. (Nero can write to slow speed CD blanks, too, and there are many burners that do a great job of burning at 4X speeds.) They get a bit-accurate copy of the CD tracks. Some subjectivists go through a DAC conversion, then digital filtering/signal processing, and then further digital manipulations and/or ADC to copy a CD or the tracks of a CD. So they add unnecessary steps, and more importantly signal-to-noise degradations due to the conversions and digital processing, to get the copies. Moreover, they have to do this in real time (a 70 minute CD takes at least 70 minutes to copy using Harry's high-end method), have no permanant copy of the tracks on storage, and cannot make compilations easily. But they *think* they get a *better* copy that way, although they do not get a bit accurate copy. Accuracy to the original CD must not be the most important thing. Oh I see, trekking to the basement or bedroom with CD in hand, burning it to hard disk, burning it again to blank...is not an "extra step". But putting it in the CD player, pushing the "Auto Mode" button on the CD burner in the system, and sitting back to listen to the disk while it burns IS an extra step. I get it. Not so quick, Harry. The extra step(s) I referred to was the additional DAC/digital-filtering/ADC that you said your CD went through in the duplicating process. What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? I guess if you take music out of the equation it makes sense. NOT! And the amazing thing is that you believe all those extra steps somehow gives you a better copy, despite the fact that you are not getting a bit-accurate copy as most of us could do easily. Except to improve the perceived noise level in the audible midrange? Well, Harry, if you believe that sending the data from a CD through a DTI-Pro where certain noise shaping is applied and then back to the CD recorder to be recorded always improve the "noise level in the audible mid-range", then perhaps you should do that to every CD you own. You know, make a copy of every CD through your patented method. The Perceived Noise Floor in Audible Midrange gets better! The copy is better than the original! If you think about it more, you are still limited to the 16 bits when you send your data to your CD recorder. Those 20 bit noise shaping does not do you any good. You still have to follow the 16 bit/sample 44.1KHz sampling standard when you make that CD copy. You are *NOT* using the DAC of your vaulted DAC! I think you are wrong, here. The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine, . But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent. OTOH, you seem to believe that when you copy a CD, you should not copy the exact data as it was recorded in the original. Somehow massaging it makes a CD "better" than the original...Your 16 bit, 44.1KHz samples somehow sound so much better than the original . Not if it sounds better doing it my way. And I think it does. Even if I am wrong, no harm done. |
#123
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"Stewart Pinkerton" schrieb im Newsbeitrag
... Actually, it rapidly turns out that he's selling his services, and making *extremely* dubious claims about 'digital flaws' and 'natural timbre'. Caveat emptor! :-) Selling what? Could You please explain what is "extremely" dubious on my claims? I think You are rapidly turning out from the discussion lacking own arguments, when the other offers practical evidence for his claims. It is the false time to exit from the game. I would appreciate to see _Your_ practical evidences, however I am afraid You can not. To discuss in the air only for the discussion itself is not enchanting enough for me. It is much more stirring if everybody undertakes responsibility for what he clames. Try to follow me if You can, but prove it too! I have the cards in my hand, my offer is on the table! It is Your turn now :-))))) Best regards -- Johann Spischak SDG, Spischak Digital GmbH +49-911-965-7319 http://sdg-master.com |
#124
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Johann Spischak wrote:
"Stewart Pinkerton" schrieb im Newsbeitrag ... Actually, it rapidly turns out that he's selling his services, and making *extremely* dubious claims about 'digital flaws' and 'natural timbre'. Caveat emptor! :-) Selling what? Could You please explain what is "extremely" dubious on my claims? This one, particularly: " The real problem with digital is, that the natural timbre of human voices and instruments will lost. This is even between the most hearable middle and lowest frequency range on the whole dynamic range." To the extent such problems exist, it is far more likely to be due to inadequacies of the transducers at either end of the signal chain (microphones and loudspeakers) , and their setup, than to *digital* anything. Please present evidence that timbre distortion is an inherent flaw of digital. -- -S "If men were angels, no government would be necessary." - James Madison (1788) |
#125
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Harry Lavo wrote:
"chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "Harry Lavo" wrote in message ... "Chung" wrote in message ... jonrkc wrote: snip I also recorded it through my DTI Pro / Proceed DAC, so I was using my system DAC, not the internal Marantz DAC. So it was an excellent recording setup, but not one too expensive (via eBay) for the average audiophile. Now here's the question to you: Why does the majority of critical listeners prefer digital? snip Before I get inundated, I made a mistake here and gave you my CD recording setup, which is taken from the digital out of the Proceed after passing from CD player to DTI-Pro to DAC. For SACD, I used analog out on the Sony, into my preamp, and through the tape outs to the Marantz analog ins. Usually I listen to the Marantz through its analog output, which is very fine. I can also switch the digital out into the DTI Pro/Proceed Combo but in the case of the Marantz, it adds very little. Well, maybe you are setting yourself up to get inundated again. You are saying that when you copy from CD to a blank CD, you take the digital out of the CD player to DTI-Pro to DAC? What exactly are you talking about? You have these expensive gear and can't afford a CD burner, which is standard now in just about any PC sold? I have two PC's with three burners. I use them to copy CD's or DVD's when I just want a casual copy....they are not in my audio system. I burn CD's IN my audio system when I want the best quality...from the player to the DTI Pro in real time which noise shapes them to 20 bits and de-jitters them, then feeds the signal back out from the Proceed via the sp/dif connection to the Marantz CD burner (which uses older, slower, but higher quality CD-R blanks). You know, Harry, this speaks volumes about the mentality of some subjectivists. I know that you are not alone. The goal is to make a copy of a CD, or certain tracks on a CD, onto a blank CD-R. Normal people use a computer to rip the tracks, and use software, like EAC, to insure that the data is ripped as accurately as possible. Then they use mastering software, such as Nero, to insure that the data is written correctly (verifiably so) to the CD blank. (Nero can write to slow speed CD blanks, too, and there are many burners that do a great job of burning at 4X speeds.) They get a bit-accurate copy of the CD tracks. Some subjectivists go through a DAC conversion, then digital filtering/signal processing, and then further digital manipulations and/or ADC to copy a CD or the tracks of a CD. So they add unnecessary steps, and more importantly signal-to-noise degradations due to the conversions and digital processing, to get the copies. Moreover, they have to do this in real time (a 70 minute CD takes at least 70 minutes to copy using Harry's high-end method), have no permanant copy of the tracks on storage, and cannot make compilations easily. But they *think* they get a *better* copy that way, although they do not get a bit accurate copy. Accuracy to the original CD must not be the most important thing. Oh I see, trekking to the basement or bedroom with CD in hand, burning it to hard disk, burning it again to blank...is not an "extra step". But putting it in the CD player, pushing the "Auto Mode" button on the CD burner in the system, and sitting back to listen to the disk while it burns IS an extra step. I get it. Not so quick, Harry. The extra step(s) I referred to was the additional DAC/digital-filtering/ADC that you said your CD went through in the duplicating process. What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? I guess if you take music out of the equation it makes sense. NOT! And the amazing thing is that you believe all those extra steps somehow gives you a better copy, despite the fact that you are not getting a bit-accurate copy as most of us could do easily. Except to improve the perceived noise level in the audible midrange? Well, Harry, if you believe that sending the data from a CD through a DTI-Pro where certain noise shaping is applied and then back to the CD recorder to be recorded always improve the "noise level in the audible mid-range", then perhaps you should do that to every CD you own. You know, make a copy of every CD through your patented method. The Perceived Noise Floor in Audible Midrange gets better! The copy is better than the original! If you think about it more, you are still limited to the 16 bits when you send your data to your CD recorder. Those 20 bit noise shaping does not do you any good. You still have to follow the 16 bit/sample 44.1KHz sampling standard when you make that CD copy. You are *NOT* using the DAC of your vaulted DAC! I think you are wrong, here. The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine, . But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent. Harry, do you understand that redbook CD, which is what your semi-pro Marantz recorder attempts to adhere to, is 16 bit/44.1 KHz? No exception? There is no 18 bit equivalent or 20 bit equivalent that can be stored on a CD and to be played as a CD. I would have thought that a seasoned audiophile as yourself will know the basics about CD's. You are saying that sampling a 16 bit CD at 20 bits and then downsampling back to 16 bits improve the "Perceived Audio Mid-Range". Think about it this way: you are not using the DAC as a DAC. So how can it possibly help? You are simply trying to make a copy of the original CD. How can you do better than try to make a bit-acurate copy? If your way is better, then the professionals would have done it. They could have easily sample the 16 bit to 24 bits and back to 16 bits again, right? Heck, you should do it to all your CD's. OTOH, you seem to believe that when you copy a CD, you should not copy the exact data as it was recorded in the original. Somehow massaging it makes a CD "better" than the original...Your 16 bit, 44.1KHz samples somehow sound so much better than the original . Not if it sounds better doing it my way. And I think it does. Even if I am wrong, no harm done. Only harm done is that to your credibility as someone who is familar with audio. In particular, high-end audio. |
#126
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
I agree with Steven on his assessment of timbre distortion.
Try using some software to eq some of your CD's and burn them this way. Just bump say 200hz to 1500hz by a dB. Apparent imaging will shift. The sense of timbre will definitely shift. Probably more than you would believe. Bump it both up and down. Try other things. I think anyone who hasn't done this will be surprised how it can alter your perception of the musical gestalt with these minor changes in a system you are familiar with. I use Cool Edit (now Adobe Audition) for this kind of filtering. Audacity is a free piece of software that can do many of the same things. Though it isn't quite so nicely laid out. Grab a copy of Audacity, dump a CD track on the hard drive as a wav file, and under Effects in Audacity choose FFT filter. Let it filter the track, burn it to a CD-RW and play around with such things. This can be an eye opener for lots of people. Besides for someone interested in audio this kind of thing is lots of fun. You might fix some of your most liked, but barely listenable music if you practice with it a bit. Dennis "Steven Sullivan" wrote in message ... Johann Spischak wrote: "Stewart Pinkerton" schrieb im Newsbeitrag ... Actually, it rapidly turns out that he's selling his services, and making *extremely* dubious claims about 'digital flaws' and 'natural timbre'. Caveat emptor! :-) Selling what? Could You please explain what is "extremely" dubious on my claims? This one, particularly: " The real problem with digital is, that the natural timbre of human voices and instruments will lost. This is even between the most hearable middle and lowest frequency range on the whole dynamic range." To the extent such problems exist, it is far more likely to be due to inadequacies of the transducers at either end of the signal chain (microphones and loudspeakers) , and their setup, than to *digital* anything. Please present evidence that timbre distortion is an inherent flaw of digital. -- -S "If men were angels, no government would be necessary." - James Madison (1788) |
#127
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"Steven Sullivan" schrieb im Newsbeitrag
... Johann Spischak wrote: This one, particularly: " The real problem with digital is, that the natural timbre of human voices and instruments will lost. This is even between the most hearable middle and lowest frequency range on the whole dynamic range." To the extent such problems exist, it is far more likely to be due to inadequacies of the transducers at either end of the signal chain (microphones and loudspeakers), and their setup, than to *digital* anything. Perhaps it would make sense for You to compare the sound with and without the digital part between these transducers? Of course You can claim, that there is no difference, but it would be as far from the truth as Stewarts claim to make a CD--R from an LP whithout hearable differences. Please present evidence that timbre distortion is an inherent flaw of digital. Whishes the gentleman a.) simply understandable explanation? (this version looks to be the adequate for You, since the question shows a very high grade of naivity) b.) a comprehensive scientific analysis about the attributes, advantages and problems of digital audio? (could be too complex for the start) c.) a practical demonstration which explains itself? (since I have already offered it and You answered with another question, I doubt You will like it.) Please let me know Your choice. Just as first sample would I recommend to take a look on these studies: http://www.theimann.com/Analog/A77/A77vsPCM/index.html http://www.dcsltd.co.uk/technical_papers/effects.pdf Of course any evidences as underpinninng Your statement would be appreciated! Best regards -- Johann Spischak SDG, Spischak Digital GmbH +49-911-965-7319 http://sdg-master.com |
#128
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"chung" wrote in message
... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "Harry Lavo" wrote in message ... "Chung" wrote in message ... jonrkc wrote: snip I also recorded it through my DTI Pro / Proceed DAC, so I was using my system DAC, not the internal Marantz DAC. So it was an excellent recording setup, but not one too expensive (via eBay) for the average audiophile. Now here's the question to you: Why does the majority of critical listeners prefer digital? snip Before I get inundated, I made a mistake here and gave you my CD recording setup, which is taken from the digital out of the Proceed after passing from CD player to DTI-Pro to DAC. For SACD, I used analog out on the Sony, into my preamp, and through the tape outs to the Marantz analog ins. Usually I listen to the Marantz through its analog output, which is very fine. I can also switch the digital out into the DTI Pro/Proceed Combo but in the case of the Marantz, it adds very little. Well, maybe you are setting yourself up to get inundated again. You are saying that when you copy from CD to a blank CD, you take the digital out of the CD player to DTI-Pro to DAC? What exactly are you talking about? You have these expensive gear and can't afford a CD burner, which is standard now in just about any PC sold? I have two PC's with three burners. I use them to copy CD's or DVD's when I just want a casual copy....they are not in my audio system. I burn CD's IN my audio system when I want the best quality...from the player to the DTI Pro in real time which noise shapes them to 20 bits and de-jitters them, then feeds the signal back out from the Proceed via the sp/dif connection to the Marantz CD burner (which uses older, slower, but higher quality CD-R blanks). You know, Harry, this speaks volumes about the mentality of some subjectivists. I know that you are not alone. The goal is to make a copy of a CD, or certain tracks on a CD, onto a blank CD-R. Normal people use a computer to rip the tracks, and use software, like EAC, to insure that the data is ripped as accurately as possible. Then they use mastering software, such as Nero, to insure that the data is written correctly (verifiably so) to the CD blank. (Nero can write to slow speed CD blanks, too, and there are many burners that do a great job of burning at 4X speeds.) They get a bit-accurate copy of the CD tracks. Some subjectivists go through a DAC conversion, then digital filtering/signal processing, and then further digital manipulations and/or ADC to copy a CD or the tracks of a CD. So they add unnecessary steps, and more importantly signal-to-noise degradations due to the conversions and digital processing, to get the copies. Moreover, they have to do this in real time (a 70 minute CD takes at least 70 minutes to copy using Harry's high-end method), have no permanant copy of the tracks on storage, and cannot make compilations easily. But they *think* they get a *better* copy that way, although they do not get a bit accurate copy. Accuracy to the original CD must not be the most important thing. Oh I see, trekking to the basement or bedroom with CD in hand, burning it to hard disk, burning it again to blank...is not an "extra step". But putting it in the CD player, pushing the "Auto Mode" button on the CD burner in the system, and sitting back to listen to the disk while it burns IS an extra step. I get it. Not so quick, Harry. The extra step(s) I referred to was the additional DAC/digital-filtering/ADC that you said your CD went through in the duplicating process. What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? I guess if you take music out of the equation it makes sense. NOT! And the amazing thing is that you believe all those extra steps somehow gives you a better copy, despite the fact that you are not getting a bit-accurate copy as most of us could do easily. Except to improve the perceived noise level in the audible midrange? Well, Harry, if you believe that sending the data from a CD through a DTI-Pro where certain noise shaping is applied and then back to the CD recorder to be recorded always improve the "noise level in the audible mid-range", then perhaps you should do that to every CD you own. You know, make a copy of every CD through your patented method. The Perceived Noise Floor in Audible Midrange gets better! The copy is better than the original! If you think about it more, you are still limited to the 16 bits when you send your data to your CD recorder. Those 20 bit noise shaping does not do you any good. You still have to follow the 16 bit/sample 44.1KHz sampling standard when you make that CD copy. You are *NOT* using the DAC of your vaulted DAC! I think you are wrong, here. The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine, . But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent. Harry, do you understand that redbook CD, which is what your semi-pro Marantz recorder attempts to adhere to, is 16 bit/44.1 KHz? No exception? There is no 18 bit equivalent or 20 bit equivalent that can be stored on a CD and to be played as a CD. You've never heard of super bit mapping? I would have thought that a seasoned audiophile as yourself will know the basics about CD's. You are saying that sampling a 16 bit CD at 20 bits and then downsampling back to 16 bits improve the "Perceived Audio Mid-Range". And I would think an AA such as yourself would know something about noise shaping. Think about it this way: you are not using the DAC as a DAC. So how can it possibly help? You are simply trying to make a copy of the original CD. How can you do better than try to make a bit-acurate copy? If your way is better, then the professionals would have done it. They could have easily sample the 16 bit to 24 bits and back to 16 bits again, right? Heck, you should do it to all your CD's. With noise shaping as part of the process, Sony has. It's called Super Bit Mapping. OTOH, you seem to believe that when you copy a CD, you should not copy the exact data as it was recorded in the original. Somehow massaging it makes a CD "better" than the original...Your 16 bit, 44.1KHz samples somehow sound so much better than the original . Not if it sounds better doing it my way. And I think it does. Even if I am wrong, no harm done. Only harm done is that to your credibility as someone who is familar with audio. In particular, high-end audio. I acknowledged I might be wrong. Does that make me somehow unworthy? |
#129
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
On 18 Feb 2006 19:41:22 GMT, "Johann Spischak"
wrote: "Stewart Pinkerton" schrieb im Newsbeitrag ... Actually, it rapidly turns out that he's selling his services, and making *extremely* dubious claims about 'digital flaws' and 'natural timbre'. Caveat emptor! :-) Selling what? Johann Spischak SDG, Spischak Digital GmbH +49-911-965-7319 http://sdg-master.com You mean you don't know what you're selling at sdg-master.com? Could You please explain what is "extremely" dubious on my claims? The notion that digital audio somehow alters musical timbre. I think You are rapidly turning out from the discussion lacking own arguments, when the other offers practical evidence for his claims. But you don't, you merely make baseless claims. It is the false time to exit from the game. I would appreciate to see _Your_ practical evidences, however I am afraid You can not. To discuss in the air only for the discussion itself is not enchanting enough for me. It is much more stirring if everybody undertakes responsibility for what he clames. Try to follow me if You can, but prove it too! I have the cards in my hand, my offer is on the table! It is Your turn now :-))))) You have no cards in your hand, you have merely made some ludicrous claims about digital audio, easily refuted by anyone with access to a decent modern 24/96 AD/DA system, such as can be found in many PCs, which will sound identical to a 'straight wire' bypass. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#130
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
[Moderator's note: This is the corrected version of Harry's post. I
inadvertantly posted a earlier version that had be corrected by Harry due to my questioning the "AA" thing. -- deb ] "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: "Harry Lavo" wrote in message ... "Chung" wrote in message ... jonrkc wrote: snip I also recorded it through my DTI Pro / Proceed DAC, so I was using my system DAC, not the internal Marantz DAC. So it was an excellent recording setup, but not one too expensive (via eBay) for the average audiophile. Now here's the question to you: Why does the majority of critical listeners prefer digital? snip Before I get inundated, I made a mistake here and gave you my CD recording setup, which is taken from the digital out of the Proceed after passing from CD player to DTI-Pro to DAC. For SACD, I used analog out on the Sony, into my preamp, and through the tape outs to the Marantz analog ins. Usually I listen to the Marantz through its analog output, which is very fine. I can also switch the digital out into the DTI Pro/Proceed Combo but in the case of the Marantz, it adds very little. Well, maybe you are setting yourself up to get inundated again. You are saying that when you copy from CD to a blank CD, you take the digital out of the CD player to DTI-Pro to DAC? What exactly are you talking about? You have these expensive gear and can't afford a CD burner, which is standard now in just about any PC sold? I have two PC's with three burners. I use them to copy CD's or DVD's when I just want a casual copy....they are not in my audio system. I burn CD's IN my audio system when I want the best quality...from the player to the DTI Pro in real time which noise shapes them to 20 bits and de-jitters them, then feeds the signal back out from the Proceed via the sp/dif connection to the Marantz CD burner (which uses older, slower, but higher quality CD-R blanks). You know, Harry, this speaks volumes about the mentality of some subjectivists. I know that you are not alone. The goal is to make a copy of a CD, or certain tracks on a CD, onto a blank CD-R. Normal people use a computer to rip the tracks, and use software, like EAC, to insure that the data is ripped as accurately as possible. Then they use mastering software, such as Nero, to insure that the data is written correctly (verifiably so) to the CD blank. (Nero can write to slow speed CD blanks, too, and there are many burners that do a great job of burning at 4X speeds.) They get a bit-accurate copy of the CD tracks. Some subjectivists go through a DAC conversion, then digital filtering/signal processing, and then further digital manipulations and/or ADC to copy a CD or the tracks of a CD. So they add unnecessary steps, and more importantly signal-to-noise degradations due to the conversions and digital processing, to get the copies. Moreover, they have to do this in real time (a 70 minute CD takes at least 70 minutes to copy using Harry's high-end method), have no permanant copy of the tracks on storage, and cannot make compilations easily. But they *think* they get a *better* copy that way, although they do not get a bit accurate copy. Accuracy to the original CD must not be the most important thing. Oh I see, trekking to the basement or bedroom with CD in hand, burning it to hard disk, burning it again to blank...is not an "extra step". But putting it in the CD player, pushing the "Auto Mode" button on the CD burner in the system, and sitting back to listen to the disk while it burns IS an extra step. I get it. Not so quick, Harry. The extra step(s) I referred to was the additional DAC/digital-filtering/ADC that you said your CD went through in the duplicating process. What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? I guess if you take music out of the equation it makes sense. NOT! And the amazing thing is that you believe all those extra steps somehow gives you a better copy, despite the fact that you are not getting a bit-accurate copy as most of us could do easily. Except to improve the perceived noise level in the audible midrange? Well, Harry, if you believe that sending the data from a CD through a DTI-Pro where certain noise shaping is applied and then back to the CD recorder to be recorded always improve the "noise level in the audible mid-range", then perhaps you should do that to every CD you own. You know, make a copy of every CD through your patented method. The Perceived Noise Floor in Audible Midrange gets better! The copy is better than the original! If you think about it more, you are still limited to the 16 bits when you send your data to your CD recorder. Those 20 bit noise shaping does not do you any good. You still have to follow the 16 bit/sample 44.1KHz sampling standard when you make that CD copy. You are *NOT* using the DAC of your vaulted DAC! I think you are wrong, here. The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine, . But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent. Harry, do you understand that redbook CD, which is what your semi-pro Marantz recorder attempts to adhere to, is 16 bit/44.1 KHz? No exception? There is no 18 bit equivalent or 20 bit equivalent that can be stored on a CD and to be played as a CD. You've never heard of super bit mapping? I would have thought that a seasoned audiophile as yourself will know the basics about CD's. You are saying that sampling a 16 bit CD at 20 bits and then downsampling back to 16 bits improve the "Perceived Audio Mid-Range". And I would think an EE such as yourself would know something about noise shaping. Think about it this way: you are not using the DAC as a DAC. So how can it possibly help? You are simply trying to make a copy of the original CD. How can you do better than try to make a bit-acurate copy? If your way is better, then the professionals would have done it. They could have easily sample the 16 bit to 24 bits and back to 16 bits again, right? Heck, you should do it to all your CD's. With noise shaping as part of the process, Sony has. It's called Super Bit Mapping. OTOH, you seem to believe that when you copy a CD, you should not copy the exact data as it was recorded in the original. Somehow massaging it makes a CD "better" than the original...Your 16 bit, 44.1KHz samples somehow sound so much better than the original . Not if it sounds better doing it my way. And I think it does. Even if I am wrong, no harm done. Only harm done is that to your credibility as someone who is familar with audio. In particular, high-end audio. I acknowledged I might be wrong. Does that make me somehow unworthy? |
#131
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Johann Spischak wrote:
"Steven Sullivan" schrieb im Newsbeitrag ... Johann Spischak wrote: This one, particularly: " The real problem with digital is, that the natural timbre of human voices and instruments will lost. This is even between the most hearable middle and lowest frequency range on the whole dynamic range." To the extent such problems exist, it is far more likely to be due to inadequacies of the transducers at either end of the signal chain (microphones and loudspeakers), and their setup, than to *digital* anything. Perhaps it would make sense for You to compare the sound with and without the digital part between these transducers? You mean, compare an all-analog chain to an all-digital? I would not be at all surprised if that sounded different. I would expect the digital chain to be more *accurate*. Of course You can claim, that there is no difference, but it would be as far from the truth as Stewarts claim to make a CD--R from an LP whithout hearable differences. Based on my experience, his is a reasonable claim. Have you tried it? Now , if you can, try recording a CD to an LP. I predict you will introduce far more 'timbral distortion' than going from LP--CD. Please present evidence that timbre distortion is an inherent flaw of digital. Whishes the gentleman a.) simply understandable explanation? (this version looks to be the adequate for You, since the question shows a very high grade of naivity) b.) a comprehensive scientific analysis about the attributes, advantages and problems of digital audio? (could be too complex for the start) c.) a practical demonstration which explains itself? (since I have already offered it and You answered with another question, I doubt You will like it.) KNock yourself out, using the scientific explanation. Whatever I don't comprehend -- I'm a scientist myself, though not in a field related to audio -- I'm sure someone else here will. I will require it to be in English, though. Just as first sample would I recommend to take a look on these studies: http://www.theimann.com/Analog/A77/A77vsPCM/index.html in German http://www.dcsltd.co.uk/technical_papers/effects.pdf No blind listening tests in evidence. And these sighted results to NOT accord with comparisons done by Bob Katz and other mastering engineers who are interested in scientific evaluation of such claims. I will recommend these two books to you: Ken Pohlmann -- Principles of Digital Audio Nika Aldrich -- Digital Audio Explained for the Recording Engineer -- -S "If men were angels, no government would be necessary." - James Madison (1788) |
#132
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Johann Spischak wrote:
"Steven Sullivan" schrieb im Newsbeitrag ... Johann Spischak wrote: This one, particularly: " The real problem with digital is, that the natural timbre of human voices and instruments will lost. This is even between the most hearable middle and lowest frequency range on the whole dynamic range." To the extent such problems exist, it is far more likely to be due to inadequacies of the transducers at either end of the signal chain (microphones and loudspeakers), and their setup, than to *digital* anything. Perhaps it would make sense for You to compare the sound with and without the digital part between these transducers? Of course You can claim, that there is no difference, but it would be as far from the truth as Stewarts claim to make a CD--R from an LP whithout hearable differences. Perhaps Johann would care to read this article: http://www.bostonaudiosociety.org/ba...x_testing2.htm It is also trivially easy to show that a ADC followed by DAC generates insignificant changes in the spectrum, or the timbre, of musical signals. Unless one insists that one can hear ultrasonics, but then Johann was talking about mid-ranges and low frequencies. Arny Krueger used to have a website where he tested soundcards, and showed results after multiple loopbacks. Even $100 soundcards do an exemplary job of preserving timber, especially in the low to middle frequencies. Does Johann have any technical data showing the changes in spectrum or timbre of signals after ADC-DAC? Does Johann believe that in this case, test instruments are much more sensitive than ears? |
#133
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Harry Lavo wrote:
[Moderator's note: This is the corrected version of Harry's post. I inadvertantly posted a earlier version that had be corrected by Harry due to my questioning the "AA" thing. -- deb ] For a moment, I thought Harry knew something about me that I don't know . What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? I guess if you take music out of the equation it makes sense. NOT! And the amazing thing is that you believe all those extra steps somehow gives you a better copy, despite the fact that you are not getting a bit-accurate copy as most of us could do easily. Except to improve the perceived noise level in the audible midrange? Well, Harry, if you believe that sending the data from a CD through a DTI-Pro where certain noise shaping is applied and then back to the CD recorder to be recorded always improve the "noise level in the audible mid-range", then perhaps you should do that to every CD you own. You know, make a copy of every CD through your patented method. The Perceived Noise Floor in Audible Midrange gets better! The copy is better than the original! If you think about it more, you are still limited to the 16 bits when you send your data to your CD recorder. Those 20 bit noise shaping does not do you any good. You still have to follow the 16 bit/sample 44.1KHz sampling standard when you make that CD copy. You are *NOT* using the DAC of your vaulted DAC! I think you are wrong, here. The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine, . But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent. Harry, do you understand that redbook CD, which is what your semi-pro Marantz recorder attempts to adhere to, is 16 bit/44.1 KHz? No exception? There is no 18 bit equivalent or 20 bit equivalent that can be stored on a CD and to be played as a CD. You've never heard of super bit mapping? I heard of it, but never was intereseted. I just did some googling and got some more info about this. So do you know what it means? Need some help? I would have thought that a seasoned audiophile as yourself will know the basics about CD's. You are saying that sampling a 16 bit CD at 20 bits and then downsampling back to 16 bits improve the "Perceived Audio Mid-Range". And I would think an EE such as yourself would know something about noise shaping. Another example of your poor undestanding of digital audio noted. A lot of EE's do know something about noise shaping, but certainly not all. EE is such a broad discipline. I am not all sure what you meant by "an EE like you". I happen to work with modern radios a lot, so I do have more than a passing familiarlity with the concept. What you still do not understand is that your original CD has 16 bit samples. The information content is already set and cannot be increased. You can only degrade or lose information; you cannot gain more. You *cannot* gain more than the 16 bits of resolution that is on the original CD by re-sampling. You may *imagine* so, but that does not make it real. You need to get a clue about this basic understanding. What Sony does, and I can tell you that it's really a marketing gimmick to give it a name called "Super Bit Mapping", is to oversample the *analog* source to higher than 16 bit resolution and/or higher sample rates, and then apply digital filtering to truncate/round-off to 16 bits, while paying careful attention to dithering. In other words, they are not sampling at 16/44.1K, but at higher rates and resolution and then noise filter and resample to 16 bits. Other people simply call it over-sampling using 18, 20 or 24 bit ADC's. The big difference between that and what you think you are attempting to do, is that the source has to be analog, or digital source with higher than 16 bit resolution. If the original is already at 16/44.1 already, you cannot get a "better" copy, since there is no additional information in the original. Try reading some basic books on digital audio to get some understanding. BTW, understanding noise shaping and knowing what "Super Bit Mapping" means are two totally orthogonal issues. One can be a world-class expert on noise shaping, yet know nothing about "Super Bit Mapping". And I see once again a vain attempt at belittling the opposition when you have lost the argument. Think about it this way: you are not using the DAC as a DAC. So how can it possibly help? You are simply trying to make a copy of the original CD. How can you do better than try to make a bit-acurate copy? If your way is better, then the professionals would have done it. They could have easily sample the 16 bit to 24 bits and back to 16 bits again, right? Heck, you should do it to all your CD's. With noise shaping as part of the process, Sony has. It's called Super Bit Mapping. See above for an explanation of where you went wrong in your understanding. OTOH, you seem to believe that when you copy a CD, you should not copy the exact data as it was recorded in the original. Somehow massaging it makes a CD "better" than the original...Your 16 bit, 44.1KHz samples somehow sound so much better than the original . Not if it sounds better doing it my way. And I think it does. Even if I am wrong, no harm done. Only harm done is that to your credibility as someone who is familar with audio. In particular, high-end audio. I acknowledged I might be wrong. Does that make me somehow unworthy? Only if you say so . However, that makes your opinions, which often you stated with such apparent authority, about audio extremely suspect. Another obvious conclusion from what you posted is that if you think something should sound better, then, of course, it sounds better. |
#134
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"chung" wrote in message
... Harry Lavo wrote: [Moderator's note: This is the corrected version of Harry's post. I inadvertantly posted a earlier version that had be corrected by Harry due to my questioning the "AA" thing. -- deb ] For a moment, I thought Harry knew something about me that I don't know . I know some things, but that's not one of them :-). What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? I guess if you take music out of the equation it makes sense. NOT! And the amazing thing is that you believe all those extra steps somehow gives you a better copy, despite the fact that you are not getting a bit-accurate copy as most of us could do easily. Except to improve the perceived noise level in the audible midrange? Well, Harry, if you believe that sending the data from a CD through a DTI-Pro where certain noise shaping is applied and then back to the CD recorder to be recorded always improve the "noise level in the audible mid-range", then perhaps you should do that to every CD you own. You know, make a copy of every CD through your patented method. The Perceived Noise Floor in Audible Midrange gets better! The copy is better than the original! If you think about it more, you are still limited to the 16 bits when you send your data to your CD recorder. Those 20 bit noise shaping does not do you any good. You still have to follow the 16 bit/sample 44.1KHz sampling standard when you make that CD copy. You are *NOT* using the DAC of your vaulted DAC! I think you are wrong, here. The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine, . But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent. Harry, do you understand that redbook CD, which is what your semi-pro Marantz recorder attempts to adhere to, is 16 bit/44.1 KHz? No exception? There is no 18 bit equivalent or 20 bit equivalent that can be stored on a CD and to be played as a CD. You've never heard of super bit mapping? I heard of it, but never was intereseted. I just did some googling and got some more info about this. So do you know what it means? Need some help? Yes, I know what it means. I may or may not be interpreting correctly whether or not my system is actually accomplishing it. I would have thought that a seasoned audiophile as yourself will know the basics about CD's. You are saying that sampling a 16 bit CD at 20 bits and then downsampling back to 16 bits improve the "Perceived Audio Mid-Range". No, I am saying that noise-shaping a 16 bit signal to a 20 bit-equivalent signal of 18 bits, and then downsampling the noise-shaped signal back to 16 bits may improve the "Perceived Audio Mid-Range" by the functional equivalent of two bits. And I would think an EE such as yourself would know something about noise shaping. Another example of your poor undestanding of digital audio noted. A lot of EE's do know something about noise shaping, but certainly not all. EE is such a broad discipline. I am not all sure what you meant by "an EE like you". I happen to work with modern radios a lot, so I do have more than a passing familiarlity with the concept. I simply meant an EE with a passionate devotion to the audio hobby and to digital reproduction, that's all. What you still do not understand is that your original CD has 16 bit samples. The information content is already set and cannot be increased. You can only degrade or lose information; you cannot gain more. You *cannot* gain more than the 16 bits of resolution that is on the original CD by re-sampling. You may *imagine* so, but that does not make it real. You need to get a clue about this basic understanding. What Sony does, and I can tell you that it's really a marketing gimmick to give it a name called "Super Bit Mapping", is to oversample the *analog* source to higher than 16 bit resolution and/or higher sample rates, and then apply digital filtering to truncate/round-off to 16 bits, while paying careful attention to dithering. In other words, they are not sampling at 16/44.1K, but at higher rates and resolution and then noise filter and resample to 16 bits. Other people simply call it over-sampling using 18, 20 or 24 bit ADC's. The big difference between that and what you think you are attempting to do, is that the source has to be analog, or digital source with higher than 16 bit resolution. If the original is already at 16/44.1 already, you cannot get a "better" copy, since there is no additional information in the original. Why would not the digital noise and artifacts at the 16 bit level not be reduced when upsampled, noise shaped, and then reduced back. Try reading some basic books on digital audio to get some understanding. You may or may not know, but it seems to me relevant, that the DTI Pro does actually put out an eighteen bit, noise shaped signal to the Proceed DAC. DTI claims that this 18 bit signal is the equivalent (in noise) of a 20 bit signal in the sensitive midrange region. Keep in mind that the DTI Pro is for altering 16 bit signals (either 44.1 or 48khz), not for working from analog or high-bit-rate sources. Since the Proceed DAC is an 18 bit dac, if one is tapping its analog output, one is getting that noise-shaped eighteen bit signal (20 bit noise equivalent) translated back into analog. I am assuming here that the extra four bits is a reduction in digital noise and artifacts from the 16 bit level. On the other hand, if one is tapping the SPDIF output from the Proceed DAC, as I am doing into the Marantz, one can only get a 16 bit signal. Whether it is truncated or dithered down from eighteen bits, I do not know. But the shaping of the original resampled noise should still give a two-bit gain in digital noise in the midrange versus a straight copy of the 16 bit signal, should it not. That is my understanding. If I am wrong, please correct me with this specific example, not with a general theory please. Why does not upsampling and then noise shaping not hold when downsampled to 16 bits? If it was the case that recording at 16 bits wiped out the advantage of noise shaping, then what would be the tecnical advantage of Sony's Super Bit Mapped Disks. If it is the case that a 16 bit source can gain no advantage from upsampling and noise-shaping, why would DTI go to the expense of producing such a piece of gear? BTW, understanding noise shaping and knowing what "Super Bit Mapping" means are two totally orthogonal issues. One can be a world-class expert on noise shaping, yet know nothing about "Super Bit Mapping". I'll take your word for it. And I see once again a vain attempt at belittling the opposition when you have lost the argument. ??? What argument have I lost?. And what belittlement have I given? Up to this point I don't even know if I lost. Think about it this way: you are not using the DAC as a DAC. So how can it possibly help? You are simply trying to make a copy of the original CD. How can you do better than try to make a bit-acurate copy? If your way is better, then the professionals would have done it. They could have easily sample the 16 bit to 24 bits and back to 16 bits again, right? Heck, you should do it to all your CD's. With noise shaping as part of the process, Sony has. It's called Super Bit Mapping. See above for an explanation of where you went wrong in your understanding. I hear you, but there are some troubling questions, which I've raised above. OTOH, you seem to believe that when you copy a CD, you should not copy the exact data as it was recorded in the original. Somehow massaging it makes a CD "better" than the original...Your 16 bit, 44.1KHz samples somehow sound so much better than the original . Not if it sounds better doing it my way. And I think it does. Even if I am wrong, no harm done. Only harm done is that to your credibility as someone who is familar with audio. In particular, high-end audio. I acknowledged I might be wrong. Does that make me somehow unworthy? Only if you say so . However, that makes your opinions, which often you stated with such apparent authority, about audio extremely suspect. Another obvious conclusion from what you posted is that if you think something should sound better, then, of course, it sounds better. I know a lot about audio, with authority, built up over 50 years as a hobbyist and semi-professional recordist. That doesn't mean I can't be wrong about something as arcane as digital theory, when even our group's leading EE acknowledges that he didn't know about or understand what Super Bit Mapping was until today. Give me a break! |
#135
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Harry Lavo wrote:
"chung" wrote in message ... Harry Lavo wrote: [Moderator's note: This is the corrected version of Harry's post. I inadvertantly posted a earlier version that had be corrected by Harry due to my questioning the "AA" thing. -- deb ] For a moment, I thought Harry knew something about me that I don't know . I know some things, but that's not one of them :-). What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? I guess if you take music out of the equation it makes sense. NOT! And the amazing thing is that you believe all those extra steps somehow gives you a better copy, despite the fact that you are not getting a bit-accurate copy as most of us could do easily. Except to improve the perceived noise level in the audible midrange? Well, Harry, if you believe that sending the data from a CD through a DTI-Pro where certain noise shaping is applied and then back to the CD recorder to be recorded always improve the "noise level in the audible mid-range", then perhaps you should do that to every CD you own. You know, make a copy of every CD through your patented method. The Perceived Noise Floor in Audible Midrange gets better! The copy is better than the original! If you think about it more, you are still limited to the 16 bits when you send your data to your CD recorder. Those 20 bit noise shaping does not do you any good. You still have to follow the 16 bit/sample 44.1KHz sampling standard when you make that CD copy. You are *NOT* using the DAC of your vaulted DAC! I think you are wrong, here. The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine, . But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent. Harry, do you understand that redbook CD, which is what your semi-pro Marantz recorder attempts to adhere to, is 16 bit/44.1 KHz? No exception? There is no 18 bit equivalent or 20 bit equivalent that can be stored on a CD and to be played as a CD. You've never heard of super bit mapping? I heard of it, but never was intereseted. I just did some googling and got some more info about this. So do you know what it means? Need some help? Yes, I know what it means. I may or may not be interpreting correctly whether or not my system is actually accomplishing it. I don't think you know what it means. You may have heard of it, or even read something about it. But if you are not interpreting it correctly or even know whether you system is actually accomplishing it or not, then you simply do not know what it really means. I would have thought that a seasoned audiophile as yourself will know the basics about CD's. You are saying that sampling a 16 bit CD at 20 bits and then downsampling back to 16 bits improve the "Perceived Audio Mid-Range". No, I am saying that noise-shaping a 16 bit signal to a 20 bit-equivalent signal of 18 bits, and then downsampling the noise-shaped signal back to 16 bits may improve the "Perceived Audio Mid-Range" by the functional equivalent of two bits. You cannot improve it by "functional equivalent" of two bits by resampling it. Look at it this way: the CD's 16-bit samples are the original. You cannot get more bits out of the 16-bit samples. You cannot create something out of nothing. You are also recording 16-bit samples back to the blank CD. Like I said, the best you can hope for is a bit-accurate copy. If there is any difference in the samples, you have *lost* information and added noise and distortion. You are stepping all over yourself. How can you noise-shape a 16 bit signal to a 20 bit equivalent signal of 18 bits, and still record at 16 bits? Do you understand what you are saying? And I would think an EE such as yourself would know something about noise shaping. Another example of your poor undestanding of digital audio noted. A lot of EE's do know something about noise shaping, but certainly not all. EE is such a broad discipline. I am not all sure what you meant by "an EE like you". I happen to work with modern radios a lot, so I do have more than a passing familiarlity with the concept. I simply meant an EE with a passionate devotion to the audio hobby and to digital reproduction, that's all. I don't really have a passionate devotion to the audio hobby. I do like to point out some people's faulty understanding or ill-conceived explanations now and then. What you still do not understand is that your original CD has 16 bit samples. The information content is already set and cannot be increased. You can only degrade or lose information; you cannot gain more. You *cannot* gain more than the 16 bits of resolution that is on the original CD by re-sampling. You may *imagine* so, but that does not make it real. You need to get a clue about this basic understanding. What Sony does, and I can tell you that it's really a marketing gimmick to give it a name called "Super Bit Mapping", is to oversample the *analog* source to higher than 16 bit resolution and/or higher sample rates, and then apply digital filtering to truncate/round-off to 16 bits, while paying careful attention to dithering. In other words, they are not sampling at 16/44.1K, but at higher rates and resolution and then noise filter and resample to 16 bits. Other people simply call it over-sampling using 18, 20 or 24 bit ADC's. The big difference between that and what you think you are attempting to do, is that the source has to be analog, or digital source with higher than 16 bit resolution. If the original is already at 16/44.1 already, you cannot get a "better" copy, since there is no additional information in the original. Why would not the digital noise and artifacts at the 16 bit level not be reduced when upsampled, noise shaped, and then reduced back. Because you start out with 16 bits and you end up with 16 bits. Can you do better than preserving those 16 bits you attempt to copy? Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Try reading some basic books on digital audio to get some understanding. You may or may not know, but it seems to me relevant, that the DTI Pro does actually put out an eighteen bit, noise shaped signal to the Proceed DAC. A DAC nowadays is often the oversampling type for a lot of good reasons. But oversampling does not mean that you can get more "equivalent" bits. You still only get 16-bits of information. DTI claims that this 18 bit signal is the equivalent (in noise) of a 20 bit signal in the sensitive midrange region. Keep in mind that the DTI Pro is for altering 16 bit signals (either 44.1 or 48khz), not for working from analog or high-bit-rate sources. Since the Proceed DAC is an 18 bit dac, if one is tapping its analog output, one is getting that noise-shaped eighteen bit signal (20 bit noise equivalent) translated back into analog. I am assuming here that the extra four bits is a reduction in digital noise and artifacts from the 16 bit level. Your explanation, if it's correct, simply means that the DTI and the Proceed DAC together implement the functionality of the oversampling DAC. On the other hand, if one is tapping the SPDIF output from the Proceed DAC, as I am doing into the Marantz, one can only get a 16 bit signal. Whether it is truncated or dithered down from eighteen bits, I do not know. But the shaping of the original resampled noise should still give a two-bit gain in digital noise in the midrange versus a straight copy of the 16 bit signal, should it not. No, it would not. Because the output still is 16 bits. Because the noise is already set by the original CD samples. You can only hope to reproduce those samples as accurately as possible. You cannot hope to get 18 bit or 20 bit SNR's out of those CD samples. That is my understanding. If I am wrong, please correct me with this specific example, not with a general theory please. Why does not upsampling and then noise shaping not hold when downsampled to 16 bits? If it was the case that recording at 16 bits wiped out the advantage of noise shaping, then what would be the tecnical advantage of Sony's Super Bit Mapped Disks. You still missed the point I made. To use Super bit Mapping, you have to start with either an analog source or a higher resolution source than 16/44.1K. If it is the case that a 16 bit source can gain no advantage from upsampling and noise-shaping, why would DTI go to the expense of producing such a piece of gear? The DTI is meant to work with a DAC that has higher resolution. The combo is an oversampling DAC, and there are implementation advantages over a non-oversampling DAC. The effective noise floor may be lower, meaning it degrades less by using the high-resolution DAC to generate the analog signals. But the important thing is that the noise floor is set by the 16-bit samples when the input is a 16-bit source. All you could hope for is to not degrade the S/N of the original CD. The best way to not degrade the S/N of the original CD in the duplication process is simply to get bit-accurate copies. BTW, understanding noise shaping and knowing what "Super Bit Mapping" means are two totally orthogonal issues. One can be a world-class expert on noise shaping, yet know nothing about "Super Bit Mapping". I'll take your word for it. And I see once again a vain attempt at belittling the opposition when you have lost the argument. ??? What argument have I lost?. And what belittlement have I given? Up to this point I don't even know if I lost. The belittlement that somehow I do not know anything about noise shaping according to you? The argument that somehow you could get a better 16 bit samples than the original? Think about it this way: you are not using the DAC as a DAC. So how can it possibly help? You are simply trying to make a copy of the original CD. How can you do better than try to make a bit-acurate copy? If your way is better, then the professionals would have done it. They could have easily sample the 16 bit to 24 bits and back to 16 bits again, right? Heck, you should do it to all your CD's. With noise shaping as part of the process, Sony has. It's called Super Bit Mapping. See above for an explanation of where you went wrong in your understanding. I hear you, but there are some troubling questions, which I've raised above. OTOH, you seem to believe that when you copy a CD, you should not copy the exact data as it was recorded in the original. Somehow massaging it makes a CD "better" than the original...Your 16 bit, 44.1KHz samples somehow sound so much better than the original . Not if it sounds better doing it my way. And I think it does. Even if I am wrong, no harm done. Only harm done is that to your credibility as someone who is familar with audio. In particular, high-end audio. I acknowledged I might be wrong. Does that make me somehow unworthy? Only if you say so . However, that makes your opinions, which often you stated with such apparent authority, about audio extremely suspect. Another obvious conclusion from what you posted is that if you think something should sound better, then, of course, it sounds better. I know a lot about audio, with authority, built up over 50 years as a hobbyist and semi-professional recordist. As they say, a little knowledge can be dangerous... That doesn't mean I can't be wrong about something as arcane as digital theory, when even our group's leading EE acknowledges that he didn't know about or understand what Super Bit Mapping was until today. What does knowing about Super Bit Mapping have to do with being an EE? Give me a break! |
#136
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
On 21 Feb 2006 00:34:16 GMT, "Harry Lavo" wrote:
You may or may not know, but it seems to me relevant, that the DTI Pro does actually put out an eighteen bit, noise shaped signal to the Proceed DAC. DTI claims that this 18 bit signal is the equivalent (in noise) of a 20 bit signal in the sensitive midrange region. Keep in mind that the DTI Pro is for altering 16 bit signals (either 44.1 or 48khz), not for working from analog or high-bit-rate sources. Since the Proceed DAC is an 18 bit dac, if one is tapping its analog output, one is getting that noise-shaped eighteen bit signal (20 bit noise equivalent) translated back into analog. I am assuming here that the extra four bits is a reduction in digital noise and artifacts from the 16 bit level. Unfortunately, it isn't. Whatever artifacts are recorded on the 16-bit original, will be faithfully reproduced after the upsampling - hopefully. Same applies to the more modern players which internally upsample to 24/192. On the other hand, if one is tapping the SPDIF output from the Proceed DAC, as I am doing into the Marantz, one can only get a 16 bit signal. Whether it is truncated or dithered down from eighteen bits, I do not know. But the shaping of the original resampled noise should still give a two-bit gain in digital noise in the midrange versus a straight copy of the 16 bit signal, should it not. No, it shouldn't. What is will do is give you an exact copy of the noise floor of the original 16-bit signal - presuming it is done competently. After all, if it didn't, then it wouldn't be giving you an accurate copy of the low-level signals around the noise floor, now would it? After all, it doesn't have a magic algorithm that can tell music and low-level 'ambience' signals from random noise. That is my understanding. If I am wrong, please correct me with this specific example, not with a general theory please. Why does not upsampling and then noise shaping not hold when downsampled to 16 bits? If you want the mathematical answer, it's because you are *not* truly upsampling, since you're starting with a 16/44 recording and creating more bits out of thin air. True upsampling requires a higher *frequency* to achieve improved resolution from the same number of bits, or the same resolution from less bits, vide the very first Philips CD players which were 4x oversampled 14-bit players. Noise shaping is a device used to shift the noise spectrum, and it requires a higher sampling frequency, not more bits. If it was the case that recording at 16 bits wiped out the advantage of noise shaping, then what would be the tecnical advantage of Sony's Super Bit Mapped Disks. Their theory seems to be that they record at 20-bit resolution, and then carefully dither down to 16 bits. How this is supposed to be superior to any typical modern 24/96 master, also carefully resampled and dithered down to 16/44, is a mystery best left to Sony's marketing department. If it is the case that a 16 bit source can gain no advantage from upsampling and noise-shaping, why would DTI go to the expense of producing such a piece of gear? Because bigger numbers sell more units. Marketing 101 in digital audio. There's absolutely *no* technical benefit in doing this, as has been pointed out on numerous previous occasions. If you start with a 16-bit signal, you can *not* extract more information, or increase the dynamic range in *any* part of the audio spectrum, by upsampling. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#137
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"chung" wrote in message
... Harry Lavo wrote: "chung" wrote in message ... Harry Lavo wrote: [Moderator's note: This is the corrected version of Harry's post. I inadvertantly posted a earlier version that had be corrected by Harry due to my questioning the "AA" thing. -- deb ] For a moment, I thought Harry knew something about me that I don't know . I know some things, but that's not one of them :-). What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? I guess if you take music out of the equation it makes sense. NOT! And the amazing thing is that you believe all those extra steps somehow gives you a better copy, despite the fact that you are not getting a bit-accurate copy as most of us could do easily. Except to improve the perceived noise level in the audible midrange? Well, Harry, if you believe that sending the data from a CD through a DTI-Pro where certain noise shaping is applied and then back to the CD recorder to be recorded always improve the "noise level in the audible mid-range", then perhaps you should do that to every CD you own. You know, make a copy of every CD through your patented method. The Perceived Noise Floor in Audible Midrange gets better! The copy is better than the original! If you think about it more, you are still limited to the 16 bits when you send your data to your CD recorder. Those 20 bit noise shaping does not do you any good. You still have to follow the 16 bit/sample 44.1KHz sampling standard when you make that CD copy. You are *NOT* using the DAC of your vaulted DAC! I think you are wrong, here. The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine, . But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent. Harry, do you understand that redbook CD, which is what your semi-pro Marantz recorder attempts to adhere to, is 16 bit/44.1 KHz? No exception? There is no 18 bit equivalent or 20 bit equivalent that can be stored on a CD and to be played as a CD. You've never heard of super bit mapping? I heard of it, but never was intereseted. I just did some googling and got some more info about this. So do you know what it means? Need some help? Yes, I know what it means. I may or may not be interpreting correctly whether or not my system is actually accomplishing it. I don't think you know what it means. You may have heard of it, or even read something about it. But if you are not interpreting it correctly or even know whether you system is actually accomplishing it or not, then you simply do not know what it really means. I would have thought that a seasoned audiophile as yourself will know the basics about CD's. You are saying that sampling a 16 bit CD at 20 bits and then downsampling back to 16 bits improve the "Perceived Audio Mid-Range". No, I am saying that noise-shaping a 16 bit signal to a 20 bit-equivalent signal of 18 bits, and then downsampling the noise-shaped signal back to 16 bits may improve the "Perceived Audio Mid-Range" by the functional equivalent of two bits. You cannot improve it by "functional equivalent" of two bits by resampling it. Look at it this way: the CD's 16-bit samples are the original. You cannot get more bits out of the 16-bit samples. You cannot create something out of nothing. You are also recording 16-bit samples back to the blank CD. Like I said, the best you can hope for is a bit-accurate copy. If there is any difference in the samples, you have *lost* information and added noise and distortion. You are stepping all over yourself. How can you noise-shape a 16 bit signal to a 20 bit equivalent signal of 18 bits, and still record at 16 bits? Do you understand what you are saying? And I would think an EE such as yourself would know something about noise shaping. Another example of your poor undestanding of digital audio noted. A lot of EE's do know something about noise shaping, but certainly not all. EE is such a broad discipline. I am not all sure what you meant by "an EE like you". I happen to work with modern radios a lot, so I do have more than a passing familiarlity with the concept. I simply meant an EE with a passionate devotion to the audio hobby and to digital reproduction, that's all. I don't really have a passionate devotion to the audio hobby. I do like to point out some people's faulty understanding or ill-conceived explanations now and then. What you still do not understand is that your original CD has 16 bit samples. The information content is already set and cannot be increased. You can only degrade or lose information; you cannot gain more. You *cannot* gain more than the 16 bits of resolution that is on the original CD by re-sampling. You may *imagine* so, but that does not make it real. You need to get a clue about this basic understanding. What Sony does, and I can tell you that it's really a marketing gimmick to give it a name called "Super Bit Mapping", is to oversample the *analog* source to higher than 16 bit resolution and/or higher sample rates, and then apply digital filtering to truncate/round-off to 16 bits, while paying careful attention to dithering. In other words, they are not sampling at 16/44.1K, but at higher rates and resolution and then noise filter and resample to 16 bits. Other people simply call it over-sampling using 18, 20 or 24 bit ADC's. The big difference between that and what you think you are attempting to do, is that the source has to be analog, or digital source with higher than 16 bit resolution. If the original is already at 16/44.1 already, you cannot get a "better" copy, since there is no additional information in the original. Why would not the digital noise and artifacts at the 16 bit level not be reduced when upsampled, noise shaped, and then reduced back. Because you start out with 16 bits and you end up with 16 bits. Can you do better than preserving those 16 bits you attempt to copy? Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Yes, but can it give you a lower level of digital noise and artifacts down at the 16 bit level in the crucial midrange. Seems to me noise shaping should be able to. Try reading some basic books on digital audio to get some understanding. You may or may not know, but it seems to me relevant, that the DTI Pro does actually put out an eighteen bit, noise shaped signal to the Proceed DAC. A DAC nowadays is often the oversampling type for a lot of good reasons. But oversampling does not mean that you can get more "equivalent" bits. You still only get 16-bits of information. The DTI is not the DAC. The DTI reduces jitter, adds bits, and noise shapes...in this case to 18 bits which is what the Proceed DAC was designed with. So the proceed is receiving an upsampled and noise shapped signal to translate or downsample. DTI claims that this 18 bit signal is the equivalent (in noise) of a 20 bit signal in the sensitive midrange region. Keep in mind that the DTI Pro is for altering 16 bit signals (either 44.1 or 48khz), not for working from analog or high-bit-rate sources. Since the Proceed DAC is an 18 bit dac, if one is tapping its analog output, one is getting that noise-shaped eighteen bit signal (20 bit noise equivalent) translated back into analog. I am assuming here that the extra four bits is a reduction in digital noise and artifacts from the 16 bit level. Your explanation, if it's correct, simply means that the DTI and the Proceed DAC together implement the functionality of the oversampling DAC. Okay, and why does this not reduce noise and distortion at 16 bit level in the midrange? My understanding is that you can still maintain 16 bit resolution overall by trading off more noise and distortion in the high frequencies for greater apparent resolution in the critical midrange. This is the benefit promised by Sony, even though it is a sixteen bit disk. On the other hand, if one is tapping the SPDIF output from the Proceed DAC, as I am doing into the Marantz, one can only get a 16 bit signal. Whether it is truncated or dithered down from eighteen bits, I do not know. But the shaping of the original resampled noise should still give a two-bit gain in digital noise in the midrange versus a straight copy of the 16 bit signal, should it not. No, it would not. Because the output still is 16 bits. Because the noise is already set by the original CD samples. You can only hope to reproduce those samples as accurately as possible. You cannot hope to get 18 bit or 20 bit SNR's out of those CD samples. Even in the midrange? Then what does the noise shaping do? Why can't I have 16 bits with the noise and distortion moved up in frequency? That is my understanding. If I am wrong, please correct me with this specific example, not with a general theory please. Why does not upsampling and then noise shaping not hold when downsampled to 16 bits? If it was the case that recording at 16 bits wiped out the advantage of noise shaping, then what would be the tecnical advantage of Sony's Super Bit Mapped Disks. You still missed the point I made. To use Super bit Mapping, you have to start with either an analog source or a higher resolution source than 16/44.1K. Okay, I see your answer below. If it is the case that a 16 bit source can gain no advantage from upsampling and noise-shaping, why would DTI go to the expense of producing such a piece of gear? The DTI is meant to work with a DAC that has higher resolution. The combo is an oversampling DAC, and there are implementation advantages over a non-oversampling DAC. The effective noise floor may be lower, meaning it degrades less by using the high-resolution DAC to generate the analog signals. But the important thing is that the noise floor is set by the 16-bit samples when the input is a 16-bit source. All you could hope for is to not degrade the S/N of the original CD. The best way to not degrade the S/N of the original CD in the duplication process is simply to get bit-accurate copies. So let's see if I take in what you are saying. You are saying that you can reshape noise when digitizing to get more effective bits in the midrange while still maintaining 16 bits overall, but once you have recorded those sixteen bits, you can't "rearrange" the noise and distortion, is that it? I still am nonplussed, then, as to why the DTI goes to the trouble of noise shaping. BTW, understanding noise shaping and knowing what "Super Bit Mapping" means are two totally orthogonal issues. One can be a world-class expert on noise shaping, yet know nothing about "Super Bit Mapping". I'll take your word for it. And I see once again a vain attempt at belittling the opposition when you have lost the argument. ??? What argument have I lost?. And what belittlement have I given? Up to this point I don't even know if I lost. Well, at least now I know what argument I have lost :-) The belittlement that somehow I do not know anything about noise shaping according to you? The argument that somehow you could get a better 16 bit samples than the original? Think about it this way: you are not using the DAC as a DAC. So how can it possibly help? You are simply trying to make a copy of the original CD. How can you do better than try to make a bit-acurate copy? If your way is better, then the professionals would have done it. They could have easily sample the 16 bit to 24 bits and back to 16 bits again, right? Heck, you should do it to all your CD's. With noise shaping as part of the process, Sony has. It's called Super Bit Mapping. See above for an explanation of where you went wrong in your understanding. I hear you, but there are some troubling questions, which I've raised above. OTOH, you seem to believe that when you copy a CD, you should not copy the exact data as it was recorded in the original. Somehow massaging it makes a CD "better" than the original...Your 16 bit, 44.1KHz samples somehow sound so much better than the original . Not if it sounds better doing it my way. And I think it does. Even if I am wrong, no harm done. Only harm done is that to your credibility as someone who is familar with audio. In particular, high-end audio. I acknowledged I might be wrong. Does that make me somehow unworthy? Only if you say so . However, that makes your opinions, which often you stated with such apparent authority, about audio extremely suspect. Another obvious conclusion from what you posted is that if you think something should sound better, then, of course, it sounds better. I know a lot about audio, with authority, built up over 50 years as a hobbyist and semi-professional recordist. As they say, a little knowledge can be dangerous... Goes in reverse as well, when it comes to some of the finer points of the entire hobby..... That doesn't mean I can't be wrong about something as arcane as digital theory, when even our group's leading EE acknowledges that he didn't know about or understand what Super Bit Mapping was until today. What does knowing about Super Bit Mapping have to do with being an EE? Well, my assumption was that also you were an audio hobbiest, and that the combination of two would have zeroed you in on Super Bit Mapping like a hawk. Turns out only the North-South radio beacon was in operation. :-) Give me a break! But thanks for the clarification. |
#138
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"Stewart Pinkerton" wrote in message
... On 21 Feb 2006 00:34:16 GMT, "Harry Lavo" wrote: You may or may not know, but it seems to me relevant, that the DTI Pro does actually put out an eighteen bit, noise shaped signal to the Proceed DAC. DTI claims that this 18 bit signal is the equivalent (in noise) of a 20 bit signal in the sensitive midrange region. Keep in mind that the DTI Pro is for altering 16 bit signals (either 44.1 or 48khz), not for working from analog or high-bit-rate sources. Since the Proceed DAC is an 18 bit dac, if one is tapping its analog output, one is getting that noise-shaped eighteen bit signal (20 bit noise equivalent) translated back into analog. I am assuming here that the extra four bits is a reduction in digital noise and artifacts from the 16 bit level. Unfortunately, it isn't. Whatever artifacts are recorded on the 16-bit original, will be faithfully reproduced after the upsampling - hopefully. Same applies to the more modern players which internally upsample to 24/192. On the other hand, if one is tapping the SPDIF output from the Proceed DAC, as I am doing into the Marantz, one can only get a 16 bit signal. Whether it is truncated or dithered down from eighteen bits, I do not know. But the shaping of the original resampled noise should still give a two-bit gain in digital noise in the midrange versus a straight copy of the 16 bit signal, should it not. No, it shouldn't. What is will do is give you an exact copy of the noise floor of the original 16-bit signal - presuming it is done competently. After all, if it didn't, then it wouldn't be giving you an accurate copy of the low-level signals around the noise floor, now would it? After all, it doesn't have a magic algorithm that can tell music and low-level 'ambience' signals from random noise. That is my understanding. If I am wrong, please correct me with this specific example, not with a general theory please. Why does not upsampling and then noise shaping not hold when downsampled to 16 bits? If you want the mathematical answer, it's because you are *not* truly upsampling, since you're starting with a 16/44 recording and creating more bits out of thin air. True upsampling requires a higher *frequency* to achieve improved resolution from the same number of bits, or the same resolution from less bits, vide the very first Philips CD players which were 4x oversampled 14-bit players. Noise shaping is a device used to shift the noise spectrum, and it requires a higher sampling frequency, not more bits. If it was the case that recording at 16 bits wiped out the advantage of noise shaping, then what would be the tecnical advantage of Sony's Super Bit Mapped Disks. Their theory seems to be that they record at 20-bit resolution, and then carefully dither down to 16 bits. How this is supposed to be superior to any typical modern 24/96 master, also carefully resampled and dithered down to 16/44, is a mystery best left to Sony's marketing department. If it is the case that a 16 bit source can gain no advantage from upsampling and noise-shaping, why would DTI go to the expense of producing such a piece of gear? Because bigger numbers sell more units. Marketing 101 in digital audio. There's absolutely *no* technical benefit in doing this, as has been pointed out on numerous previous occasions. If you start with a 16-bit signal, you can *not* extract more information, or increase the dynamic range in *any* part of the audio spectrum, by upsampling. Thanks for adding to the explanation. Between you and Chung, I think I finally "get it." |
#139
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Harry Lavo wrote:
Why would not the digital noise and artifacts at the 16 bit level not be reduced when upsampled, noise shaped, and then reduced back. Because you start out with 16 bits and you end up with 16 bits. Can you do better than preserving those 16 bits you attempt to copy? Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Yes, but can it give you a lower level of digital noise and artifacts down at the 16 bit level in the crucial midrange. Seems to me noise shaping should be able to. No, because the 16 bit samples sets the noise floor at the very output, aassuming the DAC is working well. You cannot get any lower noise than what has been recorded on those 16-bit samples. Look at it another way, let's say you somehow have obtained different 16-bit samples that are different than the original. In that case, the original samples have been changed. You no longer have the original performance. What oversampling and using high resolution DAC's can help is to make the overall DAC more accurate, by minimizing implementation errors. But the best you can do is to reproduce those 16-bits samples exactly and ideally. Try reading some basic books on digital audio to get some understanding. You may or may not know, but it seems to me relevant, that the DTI Pro does actually put out an eighteen bit, noise shaped signal to the Proceed DAC. A DAC nowadays is often the oversampling type for a lot of good reasons. But oversampling does not mean that you can get more "equivalent" bits. You still only get 16-bits of information. The DTI is not the DAC. The DTI reduces jitter, adds bits, and noise shapes...in this case to 18 bits which is what the Proceed DAC was designed with. So the proceed is receiving an upsampled and noise shapped signal to translate or downsample. It certianly looks like the DTI is not doing anything useful. First of all, the concept of a de-jittering device is suspect. The DAC, in this case the Proceed, should have clean clocks and circuitry that will reduce incoming jitter. The DAC's clock controls the final jitter in the output samples. If the DAC is well designed, then certainly it does not need any de-jittering circuits in front. If the DAC is poorly designed, then it will have poor jitter regardless of how good the incoming jitter is. Second, the DAC should have noise shaping if it is the oversampling high-resolution type. It is designed to work on 16-bit samples. I don't see how feeding it 18-bit samples can help. Does the DTI change the sampling rate, too? Third, the effect of jitter is so subtle, that you are wide open to expectation bias when you compare the results with and without de-jitters. If you really believe they work, then by golly, they work and they will sound better. There is no measurement test that I know of that shows having de-jitters improve the jitters of any competently designed DAC. Also be aware of lsiteners' low sensitivity to jitter. It has to be pretty bad before it is detectible via listening. In your case, your best hope is that the same 16 bit samples are sent unmodified to your Marantz recorder. Any perceived mid-range noise improvement is imaginary. That is a very illuminating lesson for subjectivists, since they should realize the great difficulty in detecting subtle differences without controls such as blinding. Two identical presentations can sound different if they really believe that the two are different. If there are real differences in the samples, then you have added distortion and noise, and if you think it sounds better that way, well, it's your life... (snip) On the other hand, if one is tapping the SPDIF output from the Proceed DAC, as I am doing into the Marantz, one can only get a 16 bit signal. Whether it is truncated or dithered down from eighteen bits, I do not know. But the shaping of the original resampled noise should still give a two-bit gain in digital noise in the midrange versus a straight copy of the 16 bit signal, should it not. No, it would not. Because the output still is 16 bits. Because the noise is already set by the original CD samples. You can only hope to reproduce those samples as accurately as possible. You cannot hope to get 18 bit or 20 bit SNR's out of those CD samples. Even in the midrange? Then what does the noise shaping do? Why can't I have 16 bits with the noise and distortion moved up in frequency? Noise shaping after oversampling and followed by digital filtering can move the quantization noise of the *DAC* out. It does nothing to the *signal*. The 16-bit samples from the original CD are the original *signal*. (snip) The best way to not degrade the S/N of the original CD in the duplication process is simply to get bit-accurate copies. So let's see if I take in what you are saying. You are saying that you can reshape noise when digitizing to get more effective bits in the midrange while still maintaining 16 bits overall, but once you have recorded those sixteen bits, you can't "rearrange" the noise and distortion, is that it? I still am nonplussed, then, as to why the DTI goes to the trouble of noise shaping. Because it's high-end, and high-end needs tweaking, buzzwords and gimmicks? |
#140
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Harry Lavo wrote:
That doesn't mean I can't be wrong about something as arcane as digital theory, when even our group's leading EE acknowledges that he didn't know about or understand what Super Bit Mapping was until today. What does knowing about Super Bit Mapping have to do with being an EE? Well, my assumption was that also you were an audio hobbiest, and that the combination of two would have zeroed you in on Super Bit Mapping like a hawk. Turns out only the North-South radio beacon was in operation. :-) Another bad assumption. I just did a search on "Super Bit Mapping" at the IEEE Xplore site, where over a million of journal articles/papers are archived. No matching results found. Real EE's are not interested in Super Bit Mapping, or other marketing buzzwords. OTOH, a search for noise shaping brings up 461 results. In case you are not aware, IEEE (Institute of Electronic and Electrical Engineers) is the world's leading professional society for EE's. It has 365,000 members from over 150 countries. |
#141
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"chung" wrote in message
... Harry Lavo wrote: snip Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Yes, but can it give you a lower level of digital noise and artifacts down at the 16 bit level in the crucial midrange. Seems to me noise shaping should be able to. No, because the 16 bit samples sets the noise floor at the very output, aassuming the DAC is working well. You cannot get any lower noise than what has been recorded on those 16-bit samples. Look at it another way, let's say you somehow have obtained different 16-bit samples that are different than the original. In that case, the original samples have been changed. You no longer have the original performance. What oversampling and using high resolution DAC's can help is to make the overall DAC more accurate, by minimizing implementation errors. But the best you can do is to reproduce those 16-bits samples exactly and ideally. Although down below you finally admit that it can move quantizitation noise up in frequency and out of the audible range. That in part was what I was positing originally. It seems to me that this may help explain why the midrange sounds possibly "cleaner" through the Proceed. You may or may not know, but it seems to me relevant, that the DTI Pro does actually put out an eighteen bit, noise shaped signal to the Proceed DAC. A DAC nowadays is often the oversampling type for a lot of good reasons. But oversampling does not mean that you can get more "equivalent" bits. You still only get 16-bits of information. The DTI is not the DAC. The DTI reduces jitter, adds bits, and noise shapes...in this case to 18 bits which is what the Proceed DAC was designed with. So the proceed is receiving an upsampled and noise shapped signal to translate or downsample. It certianly looks like the DTI is not doing anything useful. First of all, the concept of a de-jittering device is suspect. The DAC, in this case the Proceed, should have clean clocks and circuitry that will reduce incoming jitter. The DAC's clock controls the final jitter in the output samples. If the DAC is well designed, then certainly it does not need any de-jittering circuits in front. If the DAC is poorly designed, then it will have poor jitter regardless of how good the incoming jitter is. This was Levinson's first DAC, and apparently they didn't implement the jitter reduction circuitry well at all. The DTI anti-jitter device makes a huge difference versus using straight toslink or coax input. Second, the DAC should have noise shaping if it is the oversampling high-resolution type. It is designed to work on 16-bit samples. I don't see how feeding it 18-bit samples can help. Does the DTI change the sampling rate, too? I don't think it changes the sampling rate, but it apparently does change the bit depth. The DTI Pro is selectable in output to feed either 18 or 20 bit DACs, as well as 16 bit DACs. The later DTI Pro 24 feeds either 20 bit or 24 bit DACs as well as 16. Third, the effect of jitter is so subtle, that you are wide open to expectation bias when you compare the results with and without de-jitters. If you really believe they work, then by golly, they work and they will sound better. There is no measurement test that I know of that shows having de-jitters improve the jitters of any competently designed DAC. Also be aware of lsiteners' low sensitivity to jitter. It has to be pretty bad before it is detectible via listening. This is one case where the words "dramatic" and "huge" are justified. In your case, your best hope is that the same 16 bit samples are sent unmodified to your Marantz recorder. Any perceived mid-range noise improvement is imaginary. That is a very illuminating lesson for subjectivists, since they should realize the great difficulty in detecting subtle differences without controls such as blinding. Two identical presentations can sound different if they really believe that the two are different. If there are real differences in the samples, then you have added distortion and noise, and if you think it sounds better that way, well, it's your life... (snip) On the other hand, if one is tapping the SPDIF output from the Proceed DAC, as I am doing into the Marantz, one can only get a 16 bit signal. Whether it is truncated or dithered down from eighteen bits, I do not know. But the shaping of the original resampled noise should still give a two-bit gain in digital noise in the midrange versus a straight copy of the 16 bit signal, should it not. No, it would not. Because the output still is 16 bits. Because the noise is already set by the original CD samples. You can only hope to reproduce those samples as accurately as possible. You cannot hope to get 18 bit or 20 bit SNR's out of those CD samples. Even in the midrange? Then what does the noise shaping do? Why can't I have 16 bits with the noise and distortion moved up in frequency? Noise shaping after oversampling and followed by digital filtering can move the quantization noise of the *DAC* out. It does nothing to the *signal*. The 16-bit samples from the original CD are the original *signal*. Doesn't moving the quantization noise out of the midrange potentially affect the quality of the sound, even if you feel it is at a marginal level to begin with. Isn't this one of the benefits of SACD and DVD-A? (snip) The best way to not degrade the S/N of the original CD in the duplication process is simply to get bit-accurate copies. So let's see if I take in what you are saying. You are saying that you can reshape noise when digitizing to get more effective bits in the midrange while still maintaining 16 bits overall, but once you have recorded those sixteen bits, you can't "rearrange" the noise and distortion, is that it? I still am nonplussed, then, as to why the DTI goes to the trouble of noise shaping. Because it's high-end, and high-end needs tweaking, buzzwords and gimmicks? I doubt it. When DTI introduced the units they were widely heralded and helped promote the benefits of dejittering. And in my case, it took the DTI Pro / Proceed combination to make CD's finally sound close enough to analog that I could relax and start enjoying them. My Phillips 880 used by iteself sounded "musical" (it was one of the few CD players at the time that did) but it had a "fade-to-black" that just fell off the edge of the earth. The sound just "stopped". This was disconcerting enough that I could not fully enjoy the unit. Even today, the combo reveals more depth and transparency than either the Marantz 63SE I once drove it with or the Sony C222ES CD/SACD player in CD mode. It is, however, not as transparent as my newer C2000ES Sony....but it is a lot smoother and more "high-end" in the mid-range when playing CD's. |
#142
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Harry Lavo wrote:
"chung" wrote in message ... Harry Lavo wrote: snip Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Yes, but can it give you a lower level of digital noise and artifacts down at the 16 bit level in the crucial midrange. Seems to me noise shaping should be able to. No, because the 16 bit samples sets the noise floor at the very output, aassuming the DAC is working well. You cannot get any lower noise than what has been recorded on those 16-bit samples. Look at it another way, let's say you somehow have obtained different 16-bit samples that are different than the original. In that case, the original samples have been changed. You no longer have the original performance. What oversampling and using high resolution DAC's can help is to make the overall DAC more accurate, by minimizing implementation errors. But the best you can do is to reproduce those 16-bits samples exactly and ideally. Although down below you finally admit that it can move quantizitation noise up in frequency and out of the audible range. That in part was what I was positing originally. It seems to me that this may help explain why the midrange sounds possibly "cleaner" through the Proceed. Harry, you just don't quite get it. Please try to understand what I said. Overampling can potentially improve the *DAC* accuracy, not the original samples. You cannot improve upon the original 16 bit samples. You were wrongly theorizing (to put it diplomatically) about the advantages of feeding those samples through the DTI/Proceed and somehow because of noise shaping, you get better than 16-bit samples into the recorder. The rest is snipped, since you failed to grasp this important concept. Maybe someone else can help you. |
#143
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Harry Lavo wrote:
"chung" wrote in message ... Harry Lavo wrote: That doesn't mean I can't be wrong about something as arcane as digital theory, when even our group's leading EE acknowledges that he didn't know about or understand what Super Bit Mapping was until today. What does knowing about Super Bit Mapping have to do with being an EE? Well, my assumption was that also you were an audio hobbiest, and that the combination of two would have zeroed you in on Super Bit Mapping like a hawk. Turns out only the North-South radio beacon was in operation. :-) Another bad assumption. I just did a search on "Super Bit Mapping" at the IEEE Xplore site, where over a million of journal articles/papers are archived. No matching results found. Real EE's are not interested in Super Bit Mapping, or other marketing buzzwords. OTOH, a search for noise shaping brings up 461 results. In case you are not aware, IEEE (Institute of Electronic and Electrical Engineers) is the world's leading professional society for EE's. It has 365,000 members from over 150 countries. Fair enough, but I think my other assumption is the real culprit...as you admit you are not really passionate about audio. If you were, you probably would have been aware of it. I was aware of it, but not interested enough to be an expert on it. I read about it at one time, and did not retain any specific information. However you were previously stating that since I had not heard of Super Bit Mapping, therefore I did not know anything about noise shaping. That's the belittlement I was talking about. |
#144
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
chung wrote:
Harry Lavo wrote: Why would not the digital noise and artifacts at the 16 bit level not be reduced when upsampled, noise shaped, and then reduced back. Because you start out with 16 bits and you end up with 16 bits. Can you do better than preserving those 16 bits you attempt to copy? Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Yes, but can it give you a lower level of digital noise and artifacts down at the 16 bit level in the crucial midrange. Seems to me noise shaping should be able to. No, because the 16 bit samples sets the noise floor at the very output, aassuming the DAC is working well. You cannot get any lower noise than what has been recorded on those 16-bit samples. Look at it another way, let's say you somehow have obtained different 16-bit samples that are different than the original. In that case, the original samples have been changed. You no longer have the original performance. What oversampling and using high resolution DAC's can help is to make the overall DAC more accurate, by minimizing implementation errors. But the best you can do is to reproduce those 16-bits samples exactly and ideally. Can't oversampling be used to lower the effective noise floor, by spreading the noise over a wider band -- after which a filter can cut more of it out? e.g as described here by Max Hauser as Reason #2 for using oversampling: http://groups.google.com/group/rec.a...e=source&hl=en Reason #2 for oversampling: quantization-noise reduction. What was in fact an original motivation for using oversampling in CD reproduction was not the relaxation of analog filtering requirements, although that was a definite benefit. A completely separate reason is to obtain high resolution from a lower-resolution basic D/A element. This is a quantization-error issue, concerning SNR and numbers of bits, rather than filter design. An N-bit basic D/A element (memoryless block with bits on the input and an analog signal on the output) can yield a higher-than-N-bit resolution if you distribute its quantization error over a wider bandwidth than occupied by the signal you are interested in, and then filter out the unneeded bandwidth subsequently. This is the essence of oversampling for quantization-error reduction. You take an N-bit signal representing an X-kHz analog bandwidth; instead of converting it to analog at near the Nyquist rate (~ 2X ks/s) you first change it to a higher sampling rate (let's say 2MX ks/s); run it through a fast N-bit D/A element; and lowpass-filter it (in analog form now) to recover the desired X-kHz baseband. Lo and behold, your analog signal has lower quantization error than you'd expect for N bits. But what is even more important, and again widely misapprehended, there is no fixed relation between the oversampling factor (M) and the resulting resolution enhancement. It depends on exactly how you distribute the quantization error in the frequency domain. An octave (M=2) of oversampling can yield half a bit of extra resolution if you just oversample with a straight D/A element (white quantization error); or 1.5 or 2.5 or even 4 bits of extra resolution if you embed the D/A element in a more sophisticated system that shapes the spectrum of the quantization error to be mostly in high frequencies ("noise shaping"); or no added bits at all if you try noise shaping and don't do it right. Noise shaping was the approach taken at NV Philips in 1983 when their first-generation CD D/A chip set (the SAA 7030 and TDA 1540) was designed (by Dijkmans, van de Plassche, and associates). Either this or the competing first-generation Sony chip set is in most CD players manufactured up to about mid-1985, regardless of who manufactured the analog filter electronics or the rest of the player. Philips employed 14-bit bipolar-technology D/A converters with 4:1 oversampling to achieve 16-bit resolution, whereas Sony used 16-bit converters (more expensive, and not as fast) with no oversampling. Philips chose 14-bit converters, at least according to one of the designers I know, because it was cheaper to make fast 14-bit units with the technology Philips used than to make slower 16-bit units. Also, there were fringe benefits associated with oversampling: Reason #1 (above) and, as it later fortuitously turned out, Reason #3 (below). [Addendum: By 1985 there were plenty of audiophiles on net.audio gloating about their Sony CD players with "full 16-bit resolution," blissfully unaware that the competing players with 14-bit internal D/A blocks also had full 16-bit resolution, thanks to the oversampling process. I have, of course, saved those old articles and may someday post them.] In 1985 Philips Elcoma Division began producing a more-compact CD D/A block, a stereo 16-bit converter in bipolar technology, capable of operating at 200 ks/s; it therefore accommodated 4:1 oversampling but did not rely on the oversampling process to achieve full resolution, i.e. SNR; this fact simplified the analog reconstruction filter a bit further. Details on this chip can be found in Schouwenhaars et al., "A Monolithic Dual 16-bit D/A Converter," IEEE Journal of Solid-State Circuits vol. SC-21 no. 3 pp. 424-429, June 1986. In 1986 a joint Philips/Mullard/Valvo team developed a CMOS single-chip two-DAC CD/DAT D/A subsystem, a very nice design, employing 256:1 oversampling internally to yield full 16-bit resolution in each channel from internal 1-bit DACs (Naus et al., "A CMOS Stereo 16-bit D/A Converter for Digital Audio," IEEE Journal of Solid-State Circuits vol. SC-22 no. 3 pp. 390-395, June 1987). This is more or less the current state-of-the-art and is used in many CD players under the Philips labels (Magnavox, NAP, etc.) and by many other firms as well, since all but the largest consumer audio OEMs lack the expensive capacity to develop and manufacture their own high-performance data-converter chips. In this current Philips/Mullard part, because the actual D/A element inside resolves only one bit, most of the final 16-bit resolution derives from the oversampling and noise-shaping process. The requirements on the analog reconstruction filters are somewhat different now because their purpose is largely to filter out quantization error instead of (completely distinct) high-frequency images. By the same token, oversampling by 256 is in no sense inherently "better" than oversampling by 4; the difference in oversampling factor was motivated by what can be produced economically in IC technology, not by further relaxing reconstruction filter specifications, and the oversampling factor should not be construed as a figure of merit or a status symbol. Summary remark 2a: the resolution, SNR and other performance measures of a complete D/A system are not revealed by the number of bits going into the raw D/A element within. Depending on how the full conversion-reconstruction system is configured, perfect 16-bit performance may be achieved with a 16-bit, a 14-bit or even a 1-bit D/A block. Summary remark 2b: do not assume that more oversampling means fewer analog problems. Many CD players employ oversampling not to aid the analog filtering problem but to permit a lower-resolution D/A element. Indeed, this can even aggravate the analog filtering problem at high O/S factors. -- -S "If men were angels, no government would be necessary." - James Madison (1788) |
#145
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"Steven Sullivan" wrote in message
... chung wrote: Harry Lavo wrote: Why would not the digital noise and artifacts at the 16 bit level not be reduced when upsampled, noise shaped, and then reduced back. Because you start out with 16 bits and you end up with 16 bits. Can you do better than preserving those 16 bits you attempt to copy? Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Yes, but can it give you a lower level of digital noise and artifacts down at the 16 bit level in the crucial midrange. Seems to me noise shaping should be able to. No, because the 16 bit samples sets the noise floor at the very output, aassuming the DAC is working well. You cannot get any lower noise than what has been recorded on those 16-bit samples. Look at it another way, let's say you somehow have obtained different 16-bit samples that are different than the original. In that case, the original samples have been changed. You no longer have the original performance. What oversampling and using high resolution DAC's can help is to make the overall DAC more accurate, by minimizing implementation errors. But the best you can do is to reproduce those 16-bits samples exactly and ideally. Can't oversampling be used to lower the effective noise floor, by spreading the noise over a wider band -- after which a filter can cut more of it out? e.g as described here by Max Hauser as Reason #2 for using oversampling: http://groups.google.com/group/rec.a...e=source&hl=en Reason #2 for oversampling: quantization-noise reduction. What was in fact an original motivation for using oversampling in CD reproduction was not the relaxation of analog filtering requirements, although that was a definite benefit. A completely separate reason is to obtain high resolution from a lower-resolution basic D/A element. This is a quantization-error issue, concerning SNR and numbers of bits, rather than filter design. An N-bit basic D/A element (memoryless block with bits on the input and an analog signal on the output) can yield a higher-than-N-bit resolution if you distribute its quantization error over a wider bandwidth than occupied by the signal you are interested in, and then filter out the unneeded bandwidth subsequently. This is the essence of oversampling for quantization-error reduction. You take an N-bit signal representing an X-kHz analog bandwidth; instead of converting it to analog at near the Nyquist rate (~ 2X ks/s) you first change it to a higher sampling rate (let's say 2MX ks/s); run it through a fast N-bit D/A element; and lowpass-filter it (in analog form now) to recover the desired X-kHz baseband. Lo and behold, your analog signal has lower quantization error than you'd expect for N bits. But what is even more important, and again widely misapprehended, there is no fixed relation between the oversampling factor (M) and the resulting resolution enhancement. It depends on exactly how you distribute the quantization error in the frequency domain. An octave (M=2) of oversampling can yield half a bit of extra resolution if you just oversample with a straight D/A element (white quantization error); or 1.5 or 2.5 or even 4 bits of extra resolution if you embed the D/A element in a more sophisticated system that shapes the spectrum of the quantization error to be mostly in high frequencies ("noise shaping"); or no added bits at all if you try noise shaping and don't do it right. Noise shaping was the approach taken at NV Philips in 1983 when their first-generation CD D/A chip set (the SAA 7030 and TDA 1540) was designed (by Dijkmans, van de Plassche, and associates). Either this or the competing first-generation Sony chip set is in most CD players manufactured up to about mid-1985, regardless of who manufactured the analog filter electronics or the rest of the player. Philips employed 14-bit bipolar-technology D/A converters with 4:1 oversampling to achieve 16-bit resolution, whereas Sony used 16-bit converters (more expensive, and not as fast) with no oversampling. Philips chose 14-bit converters, at least according to one of the designers I know, because it was cheaper to make fast 14-bit units with the technology Philips used than to make slower 16-bit units. Also, there were fringe benefits associated with oversampling: Reason #1 (above) and, as it later fortuitously turned out, Reason #3 (below). [Addendum: By 1985 there were plenty of audiophiles on net.audio gloating about their Sony CD players with "full 16-bit resolution," blissfully unaware that the competing players with 14-bit internal D/A blocks also had full 16-bit resolution, thanks to the oversampling process. I have, of course, saved those old articles and may someday post them.] In 1985 Philips Elcoma Division began producing a more-compact CD D/A block, a stereo 16-bit converter in bipolar technology, capable of operating at 200 ks/s; it therefore accommodated 4:1 oversampling but did not rely on the oversampling process to achieve full resolution, i.e. SNR; this fact simplified the analog reconstruction filter a bit further. Details on this chip can be found in Schouwenhaars et al., "A Monolithic Dual 16-bit D/A Converter," IEEE Journal of Solid-State Circuits vol. SC-21 no. 3 pp. 424-429, June 1986. In 1986 a joint Philips/Mullard/Valvo team developed a CMOS single-chip two-DAC CD/DAT D/A subsystem, a very nice design, employing 256:1 oversampling internally to yield full 16-bit resolution in each channel from internal 1-bit DACs (Naus et al., "A CMOS Stereo 16-bit D/A Converter for Digital Audio," IEEE Journal of Solid-State Circuits vol. SC-22 no. 3 pp. 390-395, June 1987). This is more or less the current state-of-the-art and is used in many CD players under the Philips labels (Magnavox, NAP, etc.) and by many other firms as well, since all but the largest consumer audio OEMs lack the expensive capacity to develop and manufacture their own high-performance data-converter chips. In this current Philips/Mullard part, because the actual D/A element inside resolves only one bit, most of the final 16-bit resolution derives from the oversampling and noise-shaping process. The requirements on the analog reconstruction filters are somewhat different now because their purpose is largely to filter out quantization error instead of (completely distinct) high-frequency images. By the same token, oversampling by 256 is in no sense inherently "better" than oversampling by 4; the difference in oversampling factor was motivated by what can be produced economically in IC technology, not by further relaxing reconstruction filter specifications, and the oversampling factor should not be construed as a figure of merit or a status symbol. Summary remark 2a: the resolution, SNR and other performance measures of a complete D/A system are not revealed by the number of bits going into the raw D/A element within. Depending on how the full conversion-reconstruction system is configured, perfect 16-bit performance may be achieved with a 16-bit, a 14-bit or even a 1-bit D/A block. Summary remark 2b: do not assume that more oversampling means fewer analog problems. Many CD players employ oversampling not to aid the analog filtering problem but to permit a lower-resolution D/A element. Indeed, this can even aggravate the analog filtering problem at high O/S factors. Thanks, Steven. If I am reading this right, you are supporting my initial belief that the noise-shaping, bit-enhancing DTI *may* (notice I do not say does for sure) be improving S/N in the midrange through reduced digital artifacts? Is that correct? |
#146
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Steven Sullivan wrote:
chung wrote: Harry Lavo wrote: Why would not the digital noise and artifacts at the 16 bit level not be reduced when upsampled, noise shaped, and then reduced back. Because you start out with 16 bits and you end up with 16 bits. Can you do better than preserving those 16 bits you attempt to copy? Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Yes, but can it give you a lower level of digital noise and artifacts down at the 16 bit level in the crucial midrange. Seems to me noise shaping should be able to. No, because the 16 bit samples sets the noise floor at the very output, aassuming the DAC is working well. You cannot get any lower noise than what has been recorded on those 16-bit samples. Look at it another way, let's say you somehow have obtained different 16-bit samples that are different than the original. In that case, the original samples have been changed. You no longer have the original performance. What oversampling and using high resolution DAC's can help is to make the overall DAC more accurate, by minimizing implementation errors. But the best you can do is to reproduce those 16-bits samples exactly and ideally. Can't oversampling be used to lower the effective noise floor, by spreading the noise over a wider band -- after which a filter can cut more of it out? Yes, with oversampling and subsequent filtering, it is possible to move the quantization noise of the DAC (not the CD samples) to a different band where there are less perceivable effects. However, simply oversampling the 16-bit CD samples and then sending the 16/44.1 bit stream to be recorded as red-book CD, cannot possibly improve the original CD samples. |
#147
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Steven Sullivan wrote:
........ Can't oversampling be used to lower the effective noise floor, by spreading the noise over a wider band -- after which a filter can cut more of it out? ...... The whole argument between Chung and Harry Lavo could be simplified this way: Digital audio is all about going from D to A (or A to D) with the least bit of distortion as the music is converted from Digital data to Analog audio or vice versa. To do that you always always always want to minimize how many times you go from D to A or vice versa. Ideally you want to change from A to D or D to A only once. In Harry Lavo's example, he was going from D to A and then A to D. Because something is most definitely lost when you go from D to A and then a 2nd time from A back to D, there is no way that the 2nd D is as accurate as the 1st. If the 2nd D sounds better, then whatever distortion was added in the process is really what is being appreciated. Harry Lavo likes his DAD. Super Bit mappping takes a very high quality A, like a master tape, then uses a very high quality D to finally produce a decent "regular" D, the CD audio disc. In other words A to D to D, the 1st D being more highly resolute than the 2nd and final D. Sony is marketing this ADD process. Incidentally, XRCD does a similar process to Super bit mapping with much much more attention paid to getting the very high quality A into the veyr high quality D, then finally to the regular Audio CD the final D. XRCD is ADD HDCD was also an ADD process. With Chung's example he went from D to D, no middleman, no A whatsoever, and hence no more added distortion. Chung's example is DD. Nuff said. You can see how all these processes, with exception to Harry Lavo's example, go from A to D only once. Remember when CDs 1st came out and they would specify how they were mastered? Some would say DDD, ADD, DAD, etc. I always scratched my head when I saw the DAD. Couldn't quite get that but basically that the jist of what this discussion's about. On a slightly different note, is anyone familiar with Quincy Jones "Back on the Block" CD? It was mastered DDD, but I feel whatever digital equipment they used was awful. Any horns (trumpets, sax etc) were murderously harsh to listen to. I was wondering what thoughts anyone else had on that CD. Just my 3 cents CD |
#148
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
On 24 Feb 2006 20:40:18 GMT, Steven Sullivan wrote:
chung wrote: Harry Lavo wrote: Why would not the digital noise and artifacts at the 16 bit level not be reduced when upsampled, noise shaped, and then reduced back. Because you start out with 16 bits and you end up with 16 bits. Can you do better than preserving those 16 bits you attempt to copy? Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Yes, but can it give you a lower level of digital noise and artifacts down at the 16 bit level in the crucial midrange. Seems to me noise shaping should be able to. No, because the 16 bit samples sets the noise floor at the very output, aassuming the DAC is working well. You cannot get any lower noise than what has been recorded on those 16-bit samples. Look at it another way, let's say you somehow have obtained different 16-bit samples that are different than the original. In that case, the original samples have been changed. You no longer have the original performance. What oversampling and using high resolution DAC's can help is to make the overall DAC more accurate, by minimizing implementation errors. But the best you can do is to reproduce those 16-bits samples exactly and ideally. Can't oversampling be used to lower the effective noise floor, by spreading the noise over a wider band -- after which a filter can cut more of it out? e.g as described here by Max Hauser as Reason #2 for using oversampling: http://groups.google.com/group/rec.a...e=source&hl=en Reason #2 for oversampling: quantization-noise reduction. What was in fact an original motivation for using oversampling in CD reproduction was not the relaxation of analog filtering requirements, although that was a definite benefit. A completely separate reason is to obtain high resolution from a lower-resolution basic D/A element. This is a quantization-error issue, concerning SNR and numbers of bits, rather than filter design. That's correct, but as has already been pointed out, that is a method for reducing noise and increasing resolution in the *DAC*, it cannot change the noise floor of the CD itself. The Proceed is ancient technology, any modern 'universal' player will have 24/192 DACs, which have much lower noise and vastly higher resolution than that old 18-bit device, *but* when playing CD, they will produce an output limited to 22kHz with a dynamic range of 93dB, although the master tape itself will limit the dynamic range to less than 85dB. Exactly the same applies to the more sophisticated 'upsampling' players which digitally reprocess the incoming 16/44 signal to anything up to 24/192. What comes out may be a nicely clean signal with gentle filtering, but it remains limited to less than 93dB dynamic range and 22kHz bandwidth, because the reconstruction filter simply restores the original analogue signal, *exactly* as it would in a well-made 'plain Jane' 16-bit player. The effect Harry is claiming, simply cannot exist - except as a deliberate *degradation* of the original signal, maybe even with a small EQ lift in the midrange. This kind of trickery is not unknown in so-called 'high-end' gear, which is often deliberately broken so that it will indeed sound *different* - but less accurate. snip historical information which simply proves the point about DAC resolution -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#149
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
On 25 Feb 2006 17:32:33 GMT, "Harry Lavo" wrote:
If I am reading this right, you are supporting my initial belief that the noise-shaping, bit-enhancing DTI *may* (notice I do not say does for sure) be improving S/N in the midrange through reduced digital artifacts? Is that correct? NO, HE ISN'T!!! The S/N ratio will remain that of the *original* signal, just as it does if you're not using that ancient 18-bit Proceed technology, but a modern 'upsampling' 24/192 DAC-equipped player. Note that such players are *vastly* superior to the Proceed/DTI approach, but make no such nonsensical claims. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#150
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Stewart Pinkerton wrote:
On 24 Feb 2006 20:40:18 GMT, Steven Sullivan wrote: chung wrote: Harry Lavo wrote: Why would not the digital noise and artifacts at the 16 bit level not be reduced when upsampled, noise shaped, and then reduced back. Because you start out with 16 bits and you end up with 16 bits. Can you do better than preserving those 16 bits you attempt to copy? Oversampling a 16 bit signal at higher bit rates and resolution does not give you any additional information. Can you see that? Yes, but can it give you a lower level of digital noise and artifacts down at the 16 bit level in the crucial midrange. Seems to me noise shaping should be able to. No, because the 16 bit samples sets the noise floor at the very output, aassuming the DAC is working well. You cannot get any lower noise than what has been recorded on those 16-bit samples. Look at it another way, let's say you somehow have obtained different 16-bit samples that are different than the original. In that case, the original samples have been changed. You no longer have the original performance. What oversampling and using high resolution DAC's can help is to make the overall DAC more accurate, by minimizing implementation errors. But the best you can do is to reproduce those 16-bits samples exactly and ideally. Can't oversampling be used to lower the effective noise floor, by spreading the noise over a wider band -- after which a filter can cut more of it out? e.g as described here by Max Hauser as Reason #2 for using oversampling: http://groups.google.com/group/rec.a...e=source&hl=en Reason #2 for oversampling: quantization-noise reduction. What was in fact an original motivation for using oversampling in CD reproduction was not the relaxation of analog filtering requirements, although that was a definite benefit. A completely separate reason is to obtain high resolution from a lower-resolution basic D/A element. This is a quantization-error issue, concerning SNR and numbers of bits, rather than filter design. That's correct, but as has already been pointed out, that is a method for reducing noise and increasing resolution in the *DAC*, it cannot change the noise floor of the CD itself. The Proceed is ancient technology, any modern 'universal' player will have 24/192 DACs, which have much lower noise and vastly higher resolution than that old 18-bit device, *but* when playing CD, they will produce an output limited to 22kHz with a dynamic range of 93dB, although the master tape itself will limit the dynamic range to less than 85dB. It's what I thought, but thanks go to you and chung and Codifus for clarifying this further -- also clarified in Nika Aldrich's book, which I've consulted in the interim. -- -S "If men were angels, no government would be necessary." - James Madison (1788) |
#151
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
Codifus wrote:
Steven Sullivan wrote: ....... Can't oversampling be used to lower the effective noise floor, by spreading the noise over a wider band -- after which a filter can cut more of it out? ..... The whole argument between Chung and Harry Lavo could be simplified this way: Digital audio is all about going from D to A (or A to D) with the least bit of distortion as the music is converted from Digital data to Analog audio or vice versa. To do that you always always always want to minimize how many times you go from D to A or vice versa. Ideally you want to change from A to D or D to A only once. In Harry Lavo's example, he was going from D to A and then A to D. Because something is most definitely lost when you go from D to A and then a 2nd time from A back to D, there is no way that the 2nd D is as accurate as the 1st. If the 2nd D sounds better, then whatever distortion was added in the process is really what is being appreciated. Harry Lavo likes his DAD. Well, Harry changed his setup during the discussions. First he said, on 2/13: "I made a mistake here and gave you my CD recording setup, which is taken from the digital out of the Proceed after passing from CD player to DTI-Pro to DAC." So it appears that there is no analog involved. Then, he said, on 2/17: "What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? " So perhaps the Proceed is not even involved. But wait, on 2/18, he said: "The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine. But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent." So now it seems like the Proceed DAC is doing the downsampling, and is actually involved in this process. This is his contention on 2/20, when I questioned him about about his theory that sampling a CD stream at 20 bits and then downsampling back to 16 bits improves the perceived audio mid-range: "No, I am saying that noise-shaping a 16 bit signal to a 20 bit-equivalent signal of 18 bits, and then downsampling the noise-shaped signal back to 16 bits may improve the "Perceived Audio Mid-Range" by the functional equivalent of two bits." What's also amazing is that he said, on 2/23, that the DTI does not change the sample rate, but simply outputs 18 bit samples when receiving 16-bit samples (!). This is certainly not oversampling the way we understand it. Nevertheless, Harry contends the DTI makes a "dramatic" or "huge" difference. Which leads me to wonder if his Proceed DAC is broken. The more likely explanation is that Harry swallowed the high-end marketing propaganda line, hook and sinker. Super Bit mappping takes a very high quality A, like a master tape, then uses a very high quality D to finally produce a decent "regular" D, the CD audio disc. In other words A to D to D, the 1st D being more highly resolute than the 2nd and final D. Sony is marketing this ADD process. It is also possible to start with high-rez digital, and carefully generate an excellent 16/44.1 version. I think Sony also calls that SBM. The difference, if any, between what Sony does and what other people do is in the type of dither they use. Or that's how much I can tell from their descriptions Incidentally, XRCD does a similar process to Super bit mapping with much much more attention paid to getting the very high quality A into the veyr high quality D, then finally to the regular Audio CD the final D. XRCD is ADD HDCD was also an ADD process. With Chung's example he went from D to D, no middleman, no A whatsoever, and hence no more added distortion. Chung's example is DD. Nuff said. In fact, using software such as EAC, it is possible and likely to get bit accurate copies because of better error detection/correction and jitter removal, when direct digital connection to the Marantz may fail to do so. You can see how all these processes, with exception to Harry Lavo's example, go from A to D only once. Remember when CDs 1st came out and they would specify how they were mastered? Some would say DDD, ADD, DAD, etc. I always scratched my head when I saw the DAD. Couldn't quite get that but basically that the jist of what this discussion's about. Perhaps they started with a digital recording and somehow generated an analog master for vinyl. Or perhaps they did the mastering in analog. Then that analog master got converted to digital for CD mastering. Just my guess. On a slightly different note, is anyone familiar with Quincy Jones "Back on the Block" CD? It was mastered DDD, but I feel whatever digital equipment they used was awful. Any horns (trumpets, sax etc) were murderously harsh to listen to. I was wondering what thoughts anyone else had on that CD. Just my 3 cents CD |
#152
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"chung" wrote in message
... Codifus wrote: Steven Sullivan wrote: ....... Can't oversampling be used to lower the effective noise floor, by spreading the noise over a wider band -- after which a filter can cut more of it out? ..... The whole argument between Chung and Harry Lavo could be simplified this way: Digital audio is all about going from D to A (or A to D) with the least bit of distortion as the music is converted from Digital data to Analog audio or vice versa. To do that you always always always want to minimize how many times you go from D to A or vice versa. Ideally you want to change from A to D or D to A only once. In Harry Lavo's example, he was going from D to A and then A to D. Because something is most definitely lost when you go from D to A and then a 2nd time from A back to D, there is no way that the 2nd D is as accurate as the 1st. If the 2nd D sounds better, then whatever distortion was added in the process is really what is being appreciated. Harry Lavo likes his DAD. I'd like to make two points about this. 1) The discussion evolved out of a discussion of analog recording...from records and from SACD. Then evolved digital to digital. Chung asked why I didn't just use the computer. I answered I found it simpler just to hit auto-track-record on my Marantz and to record in real time when listening to a CD. This used the transport - DTI Pro - Proceed DAC - Marantz digital recorder pathway. I said I felt I got as good or better recording this way for permament archieveing on low speed (4x) disks. The discussion evolved from there... 2) I also said up front, twice, that I could be wrong in my supposition a) that the disks sounded better, and b) my theory about why they might be better. That bit of honesty was never transmitted in this summary, for whatever reason. Well, Harry changed his setup during the discussions. First he said, on 2/13: "I made a mistake here and gave you my CD recording setup, which is taken from the digital out of the Proceed after passing from CD player to DTI-Pro to DAC." So it appears that there is no analog involved. Then, he said, on 2/17: "What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? " So perhaps the Proceed is not even involved. But wait, on 2/18, he said: "The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine. But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent." So now it seems like the Proceed DAC is doing the downsampling, and is actually involved in this process. This is his contention on 2/20, when I questioned him about about his theory that sampling a CD stream at 20 bits and then downsampling back to 16 bits improves the perceived audio mid-range: "No, I am saying that noise-shaping a 16 bit signal to a 20 bit-equivalent signal of 18 bits, and then downsampling the noise-shaped signal back to 16 bits may improve the "Perceived Audio Mid-Range" by the functional equivalent of two bits." What's also amazing is that he said, on 2/23, that the DTI does not change the sample rate, but simply outputs 18 bit samples when receiving 16-bit samples (!). This is certainly not oversampling the way we understand it. Nevertheless, Harry contends the DTI makes a "dramatic" or "huge" difference. Which leads me to wonder if his Proceed DAC is broken. The more likely explanation is that Harry swallowed the high-end marketing propaganda line, hook and sinker. Let's be honest, Chung. I also said it was probably because of the anti-jitter of the DTI, since the first Proceed DAC apparently didn't handle de-jittering very well at all (a fact confirmed by its input cable sensitivity according to Arny). I never claimed that it was due to bit-mapping, or any other feature of the DTI. I also went out of my way to point out that noise-shaping (as well as de-jittering) was what DTI promoted, and that I had no independent expertise to doubt it, but couldn't understand why they would make the claim if they didn't actuall do it. Expierentially, all I can tell you is that the DTI Pro - Proceed DAC combination FINALLY (in 1990) made a CD system that sounded equivalent to my analog system and I could start enjoying CD's as music, and start amassing a serious CD collection. And to this day, it is more transparent in direct A-B with a) a Marantz 63SE, b) a Marantz Pro recorder, and c) a Sony C222ES SACD machine, all when playing CD's. But it is not as transparent as my later Sony C2000ES in playing CD's, although it is smoother and more pleasing in the critical mid-range. Super Bit mappping takes a very high quality A, like a master tape, then uses a very high quality D to finally produce a decent "regular" D, the CD audio disc. In other words A to D to D, the 1st D being more highly resolute than the 2nd and final D. Sony is marketing this ADD process. It is also possible to start with high-rez digital, and carefully generate an excellent 16/44.1 version. I think Sony also calls that SBM. The difference, if any, between what Sony does and what other people do is in the type of dither they use. Or that's how much I can tell from their descriptions Incidentally, XRCD does a similar process to Super bit mapping with much much more attention paid to getting the very high quality A into the veyr high quality D, then finally to the regular Audio CD the final D. XRCD is ADD HDCD was also an ADD process. With Chung's example he went from D to D, no middleman, no A whatsoever, and hence no more added distortion. Chung's example is DD. Nuff said. In fact, using software such as EAC, it is possible and likely to get bit accurate copies because of better error detection/correction and jitter removal, when direct digital connection to the Marantz may fail to do so. Well, the DTI took care of the jitter of a direct digital transfer, and if the Marantz also does a decent job, then it is "de-jittered" twice, so I doubt this is a problem. And I thought that error correction was a "solved problem" in CD recording. You can see how all these processes, with exception to Harry Lavo's example, go from A to D only once. Remember when CDs 1st came out and they would specify how they were mastered? Some would say DDD, ADD, DAD, etc. I always scratched my head when I saw the DAD. Couldn't quite get that but basically that the jist of what this discussion's about. Perhaps they started with a digital recording and somehow generated an analog master for vinyl. Or perhaps they did the mastering in analog. Then that analog master got converted to digital for CD mastering. Just my guess. It's very simple. Most studio's had expensive analog consoles. When they added digitial mastering machines, they recorded on them, mixed through the console, and recorded back to digital (DAD). Unless they already had or were recording to analog master tapes, in which case they mixed the session tapes through the console and recorded them in digital (AAD). Digital consoles came along much further on, and took time to replace analog due to their expense and a feeling on the part o many recording engineers that they weren't up-to-snuff sound-wise. But when installed, they allowed (DDD). Obviously, nowadays, the workstation has replaced the console in many recordings. Whether that's a good thing.... On a slightly different note, is anyone familiar with Quincy Jones "Back on the Block" CD? It was mastered DDD, but I feel whatever digital equipment they used was awful. Any horns (trumpets, sax etc) were murderously harsh to listen to. I was wondering what thoughts anyone else had on that CD. Just my 3 cents CD |
#153
Posted to rec.audio.high-end
|
|||
|
|||
Percpetion
"chung" wrote in message
... Harry Lavo wrote: "chung" wrote in message ... Codifus wrote: Steven Sullivan wrote: ....... Can't oversampling be used to lower the effective noise floor, by spreading the noise over a wider band -- after which a filter can cut more of it out? ..... The whole argument between Chung and Harry Lavo could be simplified this way: Digital audio is all about going from D to A (or A to D) with the least bit of distortion as the music is converted from Digital data to Analog audio or vice versa. To do that you always always always want to minimize how many times you go from D to A or vice versa. Ideally you want to change from A to D or D to A only once. In Harry Lavo's example, he was going from D to A and then A to D. Because something is most definitely lost when you go from D to A and then a 2nd time from A back to D, there is no way that the 2nd D is as accurate as the 1st. If the 2nd D sounds better, then whatever distortion was added in the process is really what is being appreciated. Harry Lavo likes his DAD. I'd like to make two points about this. 1) The discussion evolved out of a discussion of analog recording...from records and from SACD. Then evolved digital to digital. Chung asked why I didn't just use the computer. I answered I found it simpler just to hit auto-track-record on my Marantz and to record in real time when listening to a CD. This used the transport - DTI Pro - Proceed DAC - Marantz digital recorder pathway. I said I felt I got as good or better recording this way for permament archieveing on low speed (4x) disks. The discussion evolved from there... 2) I also said up front, twice, that I could be wrong in my supposition a) that the disks sounded better, and b) my theory about why they might be better. That bit of honesty was never transmitted in this summary, for whatever reason. I thought Codifus captured your sentiment fairly. You do like the Marantz copies better than the PC copies right? If you later admitted you could be wrong, it was after we explained to you why you could be wrong. Not so. I said I might be wrong when I first commented on the subject. Well, Harry changed his setup during the discussions. First he said, on 2/13: "I made a mistake here and gave you my CD recording setup, which is taken from the digital out of the Proceed after passing from CD player to DTI-Pro to DAC." So it appears that there is no analog involved. Then, he said, on 2/17: "What pray tell is the extra step in going transport - DTI Pro (noise shaping, dejittering) - digital input of the Marantz CDR? " So perhaps the Proceed is not even involved. But wait, on 2/18, he said: "The 44.1/16 bit CD goes into the DTI Pro (which is where the noise shaping takes place) and comes out an 18 bit, noise-shaped 20 bit equivalent signal into the 18 bit DAC, which which then downrates it back to 44.1/16 and passes it to the Marantz CD-R machine. But the noise shaping is still in place, only now it is 18 bit equivalent instead of 20 bit equivalent." So now it seems like the Proceed DAC is doing the downsampling, and is actually involved in this process. This is his contention on 2/20, when I questioned him about about his theory that sampling a CD stream at 20 bits and then downsampling back to 16 bits improves the perceived audio mid-range: "No, I am saying that noise-shaping a 16 bit signal to a 20 bit-equivalent signal of 18 bits, and then downsampling the noise-shaped signal back to 16 bits may improve the "Perceived Audio Mid-Range" by the functional equivalent of two bits." What's also amazing is that he said, on 2/23, that the DTI does not change the sample rate, but simply outputs 18 bit samples when receiving 16-bit samples (!). This is certainly not oversampling the way we understand it. Nevertheless, Harry contends the DTI makes a "dramatic" or "huge" difference. Which leads me to wonder if his Proceed DAC is broken. The more likely explanation is that Harry swallowed the high-end marketing propaganda line, hook and sinker. Let's be honest, Chung. I thought I was being honest; I even quoted your posts. You did say the DTI makes a "dramatic" or "huge" improvement, right? Right, and it did. I also said it was probably because of the anti-jitter of the DTI, since the first Proceed DAC apparently didn't handle de-jittering very well at all (a fact confirmed by its input cable sensitivity according to Arny). I thought I would never see the day when you have to depend on Arny for information to back you up . I do learn things here, unlike some others (I guess). And hopeful occassionally I may contribute to others learning as well. I may be unrealistic, but that is what I always hope will come out of a newsgroup discussion. BTW, if a DAC cannot handle de-jitter well at all, what can it handle? I never claimed that it was due to bit-mapping, or any other feature of the DTI. You claimed that those upsampling and oversampling was what got your the percived mid-range noise floor improvement...I even quoted you saying that. I didn't claim anything, other than the perceived clarity of my setup...I asked if it wasn't possible that noise-shaping and oversampling contributed. There is a big difference. I also went out of my way to point out that noise-shaping (as well as de-jittering) was what DTI promoted, and that I had no independent expertise to doubt it, but couldn't understand why they would make the claim if they didn't actuall do it. Which we answered for you. Well, not technically, you didn't although you described why it wasn't possible. But the reason you ascribed to them was marketing puffery, and I'm still not sure I buy that. |