Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Is relative phase audible?
I know I'm hardly the first person to study whether relative phase is
audible :-) Probably not even the first person this week. Nonetheless, I never actually tried it myself till now. So: Two samples each containing a signal comprising 220Hz, 440Hz, 660Hz, and 880Hz sines at the same relative levels but with different phase relationships sound pretty different, to me. (That is, I can reliably tell them apart in blind randomized trials.) Thought a few others might be interested to take a listen. So, I put the ..wav files and a bit of discussion up on my web page, at http://www.cafewalter.com/cafewalter/signals/phase.htm. By the way, does anyone know offhand whether MP3 encoders preserve relative phase? -walter |
#2
|
|||
|
|||
On Thu, 9 Dec 2004 23:47:13 -0800, "Walter Harley"
wrote: I know I'm hardly the first person to study whether relative phase is audible :-) Probably not even the first person this week. Nonetheless, I never actually tried it myself till now. So: Two samples each containing a signal comprising 220Hz, 440Hz, 660Hz, and 880Hz sines at the same relative levels but with different phase relationships sound pretty different, to me. (That is, I can reliably tell them apart in blind randomized trials.) Thought a few others might be interested to take a listen. So, I put the .wav files and a bit of discussion up on my web page, at http://www.cafewalter.com/cafewalter/signals/phase.htm. By the way, does anyone know offhand whether MP3 encoders preserve relative phase? -walter They do indeed sound very different. The one you call inphase has an "aaaaahhh" sound, and the inverted an "ohhhhhh" sound. Roughly- Saving them as MP3 preserves the phase information very nicely. d Pearce Consulting http://www.pearce.uk.com |
#3
|
|||
|
|||
I feel like my ears are being sucked out when I'm between two sound sources out of phase with
each other. -- Stephen Sank, Owner & Ribbon Mic Restorer Talking Dog Transducer Company http://stephensank.com 5517 Carmelita Drive N.E. Albuquerque, New Mexico [87111] 505-332-0336 Auth. Nakamichi & McIntosh servicer Payments preferred through Paypal.com "Don Pearce" wrote in message ... On Thu, 9 Dec 2004 23:47:13 -0800, "Walter Harley" wrote: I know I'm hardly the first person to study whether relative phase is audible :-) Probably not even the first person this week. Nonetheless, I never actually tried it myself till now. So: Two samples each containing a signal comprising 220Hz, 440Hz, 660Hz, and 880Hz sines at the same relative levels but with different phase relationships sound pretty different, to me. (That is, I can reliably tell them apart in blind randomized trials.) Thought a few others might be interested to take a listen. So, I put the .wav files and a bit of discussion up on my web page, at http://www.cafewalter.com/cafewalter/signals/phase.htm. By the way, does anyone know offhand whether MP3 encoders preserve relative phase? -walter They do indeed sound very different. The one you call inphase has an "aaaaahhh" sound, and the inverted an "ohhhhhh" sound. Roughly- Saving them as MP3 preserves the phase information very nicely. d Pearce Consulting http://www.pearce.uk.com |
#4
|
|||
|
|||
On Fri, 10 Dec 2004 04:09:34 -0700, "Stephen Sank"
wrote: I feel like my ears are being sucked out when I'm between two sound sources out of phase with each other. They aren't out of phase with each other. It is the relative phase of the harmonics from the fundamental that has been changed. But both channels are identical. d Pearce Consulting http://www.pearce.uk.com |
#5
|
|||
|
|||
Walter,
I think your experiment is a good one, but any conclusions need to be carefully drawn for at least two reasons that I can think of: [a] It's crucial to define exactly what question is being answered. Psychoacoustics describes (and the anatomy of human hearing supports) an ability to hear relative phase in the sense that you're using the term, but only below 1500 Hz or so. Above that range the ability disappears, but all your test signals were well below that point. So let's beware the false dichotomy: "we can hear relative phase" / "we cannot hear relative phase" when it's not quite so simple. [b] When you vary the phase relationships among signal components, you alter the peak levels of the composite signal (sometimes greatly) even though the effective values remain the same. Some of your test equipment may well behave differently given these changes in peak level--a power amplifier or a recorder used in the experiment may produce audibly higher or lower distortion levels, for example. Listeners may well respond differently to these differences in peak level. Of course you can't be blamed the impossibility of keeping both the effective loudness and the peak levels constant while altering the phase relationships. Point [b] can indeed be considered a valid reason to maintain relative phase relationships carefully. But this uncontrolled variable limits the conclusions that can fairly be drawn from any such experiment. --best regards |
#6
|
|||
|
|||
|
#7
|
|||
|
|||
I know I'm hardly the first person to study whether relative phase
is audible :-) Certainly not, since, by definition, phase IS relative. There is no such thing as absolute phase. I think it depends on a definition. To me "absolute phase" is often the term used when someone really means "polarity". But really, polarity is also a relative term, requiring some reference, i.e. "point of origin" (such as "positive polarity is a positive voltage on pin 2 and a positive excursion of the woofer"). But it has been determined that "absolute phase" can be heard, if the original waveform is asymetrical, such as from a single reed instrument (sax, clarinet, etc.), human voice, drums and string instruments. In other words, if the kick drum produces a compression of the local volume of air on a hit, then the woofer reproducing it should also create an increase in pressure, i.e. it is in "absolute phase" with the point of origin. Which brought me to a thought I had yesterday... many engineers use two mics on the toms and snare of the drum kit. And to have the "absolute phase" match for these two mics, they flip the bottom mic out of polarity. However, I think it would be better to flip the *top* mic out of polarity, since when the skin is first hit, it goes *down* on the heads, thus pulling the diaphragm of the top mic out and pushing the diaphragm of the bottom mic in. Thus to get a positive excursion at the speaker, the bottom mic should be used as the reference, with the top mic "flipped" to match. Undoubtedly, there are engineers who are already doing this, and maybe I've exposed a "secret". Sorry about that... Karl Winkler Lectrosonics, Inc. http://www.lectrosonics.com |
#8
|
|||
|
|||
"Karl Winkler" wrote in message oups.com... Which brought me to a thought I had yesterday... many engineers use two mics on the toms and snare of the drum kit. And to have the "absolute phase" match for these two mics, they flip the bottom mic out of polarity. However, I think it would be better to flip the *top* mic out of polarity, since when the skin is first hit, it goes *down* on the heads, thus pulling the diaphragm of the top mic out and pushing the diaphragm of the bottom mic in. Thus to get a positive excursion at the speaker, the bottom mic should be used as the reference, with the top mic "flipped" to match. Karl Winkler Lectrosonics, Inc. http://www.lectrosonics.com But why? When you sit and listen to a drummer play, you are at the side of the snare. So, micing a snare the way you suggest (or even the other way) would produce something other than what is heard in the room. That might be why minimalist micing of drums is prefered by many...more realistic. Mike |
#9
|
|||
|
|||
If you combine the fundamental with its harmonics and vary the phase of
the harmonics, the peak amplitude of the waveform will change. If you are not carful with scaling etc, the peaks can be clipped. It is because of this distortion and other non-lineatieies that you might be able to tell. If the playback system (and your ears) were linear, you should not be able to perceive a change in the phase of the harmonics relative to the fundamental. Mark |
#10
|
|||
|
|||
Mark wrote:
If you combine the fundamental with its harmonics and vary the phase of the harmonics, the peak amplitude of the waveform will change. If you are not carful with scaling etc, the peaks can be clipped. It is because of this distortion and other non-lineatieies that you might be able to tell. If the playback system (and your ears) were linear, you should not be able to perceive a change in the phase of the harmonics relative to the fundamental. What the original poster is measuring is the audibility of group delay. When someone says "relative phase" I figure they are talking about phase differences between channels. There is some good research on the audibility of group delay out there. Including Koray Oczam's paper, AES preprint 5740. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#11
|
|||
|
|||
"Michael Putrino" wrote in message ... "Karl Winkler" wrote in message oups.com... Which brought me to a thought I had yesterday... many engineers use two mics on the toms and snare of the drum kit. And to have the "absolute phase" match for these two mics, they flip the bottom mic out of polarity. However, I think it would be better to flip the *top* mic out of polarity, since when the skin is first hit, it goes *down* on the heads, thus pulling the diaphragm of the top mic out and pushing the diaphragm of the bottom mic in. Thus to get a positive excursion at the speaker, the bottom mic should be used as the reference, with the top mic "flipped" to match. Karl Winkler Lectrosonics, Inc. http://www.lectrosonics.com But why? When you sit and listen to a drummer play, you are at the side of the snare. So, micing a snare the way you suggest (or even the other way) would produce something other than what is heard in the room. That might be why minimalist micing of drums is prefered by many...more realistic. Mike Unfortunately, 'realistic' is very often not do-able because of bad rooms. Close miking a drum kit allows the producer to at least have a chance at creating a space for the kit which actually fits a mix, rather than spend the time figuring out how to eliminate the horrible sounding room that came with the miniscule number of tracks available, without destroying the source. In my experience, minimal miking only works in a special set of circumstances, which span every variable from the drummer's performance to the room itself. With a little practice, more sources will allow a good engineer to re-create what may have been missing from the tracking quality. Sound design is a big part of putting together the mix... it doesn't have to be something bigger than life and can still end up sounding "natural". Two or three sources that are crammed with multiple and possibly unbalanced, awful tones with waaay too much 'room' are often nigh on impossible to work with if drums are meant to cut through the mix at all. In most cases that I have witnessed (in my 30 years of doing this) where minimal miking was used on the drums are either very genre' specific, as in jazz - but even more so lately, the main reasons people are purporting to be 'minimalist' in technique, is that they simply don't have the space, the mics, the available tracks, or the experience to take a bigger picture to work with. It may be preferred by many, but in the global scheme of things, in large studios, that's actually very few. -- David Morgan (MAMS) http://www.m-a-m-s DOT com Morgan Audio Media Service Dallas, Texas (214) 662-9901 _______________________________________ http://www.artisan-recordingstudio.com |
#12
|
|||
|
|||
Mike, this is indeed a good point you are making. In the "for my
method" camp I would say that by presenting a positive waveform to the listener, it would create an "impact" that is associated with the snare. But what you say is true, and really, what would be needed is something that recreates a realistic impression of *height* as well as L-R information. And at this point, no one seems to be working on this principal. Clearly, that's what Ambisonics was/is doing... but none of the current 5.1, 7.1 etc. systems seem to take this into account. I'm personally a huge fan of minimal drum miking when it can be done right. In fact, for several years when I was touring with the Air Force jazz big band, I used 3 mics for the drums: two overheads and a kick mic. I would often get comments about how "realistic" the overall sound was. My goal was to maintain the impact of the drums, and the natural relationships between the different sources in the kit, rather than trying to isolate and present each source. Karl Winkler Lectrosonics, Inc. http://www.lectrosonics.com |
#13
|
|||
|
|||
"David Morgan (MAMS)" wrote in message news:EKlud.3709$mn6.3377@trnddc07... "Michael Putrino" wrote in message ... "Karl Winkler" wrote in message oups.com... Which brought me to a thought I had yesterday... many engineers use two mics on the toms and snare of the drum kit. And to have the "absolute phase" match for these two mics, they flip the bottom mic out of polarity. However, I think it would be better to flip the *top* mic out of polarity, since when the skin is first hit, it goes *down* on the heads, thus pulling the diaphragm of the top mic out and pushing the diaphragm of the bottom mic in. Thus to get a positive excursion at the speaker, the bottom mic should be used as the reference, with the top mic "flipped" to match. Karl Winkler Lectrosonics, Inc. http://www.lectrosonics.com But why? When you sit and listen to a drummer play, you are at the side of the snare. So, micing a snare the way you suggest (or even the other way) would produce something other than what is heard in the room. That might be why minimalist micing of drums is prefered by many...more realistic. Mike Unfortunately, 'realistic' is very often not do-able because of bad rooms. Close miking a drum kit allows the producer to at least have a chance at creating a space for the kit which actually fits a mix, rather than spend the time figuring out how to eliminate the horrible sounding room that came with the miniscule number of tracks available, without destroying the source. In my experience, minimal miking only works in a special set of circumstances, which span every variable from the drummer's performance to the room itself. With a little practice, more sources will allow a good engineer to re-create what may have been missing from the tracking quality. Sound design is a big part of putting together the mix... it doesn't have to be something bigger than life and can still end up sounding "natural". Two or three sources that are crammed with multiple and possibly unbalanced, awful tones with waaay too much 'room' are often nigh on impossible to work with if drums are meant to cut through the mix at all. In most cases that I have witnessed (in my 30 years of doing this) where minimal miking was used on the drums are either very genre' specific, as in jazz - but even more so lately, the main reasons people are purporting to be 'minimalist' in technique, is that they simply don't have the space, the mics, the available tracks, or the experience to take a bigger picture to work with. It may be preferred by many, but in the global scheme of things, in large studios, that's actually very few. -- David Morgan (MAMS) http://www.m-a-m-s DOT com Morgan Audio Media Service Dallas, Texas (214) 662-9901 _______________________________________ http://www.artisan-recordingstudio.com Yes, I agree...room is everything. What I was trying to convey, but did so badly, was that I don't think that it would make much difference whether switching the phase on the top mic or the bottom mic. An outward or inward pressure from the micing is not what you would normally hear in a performance anyway...so pick one...hopefully the one that sounds better to you at the time...mixed in with the whole band. Record them on seperate tracks and you can flip to your hearts content. Mike |
#14
|
|||
|
|||
On Thu, 9 Dec 2004 23:47:13 -0800, "Walter Harley"
wrote: Two samples each containing a signal comprising 220Hz, 440Hz, 660Hz, and 880Hz sines at the same relative levels but with different phase relationships sound pretty different, to me. (That is, I can reliably tell them apart in blind randomized trials.) snip This has been known for years. The Hammond "organ", first sold in 1935, uses additive synthesis in a failed attempt to recreate the sound of various organ stops. Due to the construction of the tonewheel generator, all the tonewheel are in a different phase relationship every time the organ starts up, since all the wheels are clutch driven and slip slightly upon startup. The combination you cite above would be equivalent to A below Middle C with the 8', 4', 2 2/3' and 2' drawbars pulled out. However, there's a catch here, since Hammonds are roughly tuned (not exactly) to Equal Temperatment, so the 2 2/3' pitch would be slightly flat from Just Temperament, and thus, not exactly 660 Hz. In this case, the 2 2/3' pitch would be 659.255 Hz in ET, and Hammonds "stretch and shirnk" ET just a tad here and there because of the limitations of the mathematics of the tonewheel generator. In any event, there'll be some beating from this, regardless of phase. Every time the organ is started up, this combination will sound different to the ear. The same will go when you use tones derived from a top octave generator, or a bank of free running oscillators locked with a PLL; vary the phase, the tone will sound slightly different there and there. So, to answer your question, yes, phase angle can change the timbre of harmonically complex tones. How much is a matter of conjecture. Tests done by the Allen Organ Company showed that relative phase is far less a determining factor in timbre "footprint" than is relative amplitude of harmonics, but, contrary to what many had said in the past, it IS discrenable. However, as Messrs. Fletcher and Munson learned at Bell Telephone Labs in the '20s, pitch recognition up above the midrange area, say above 1800 Hz, becomes less accurate as frequency increases. Thus, it can be argued quite well that changes in phase angle that mostly affect the top end, such as we see in digital PCM, do not materially affect the tonality of the sound, but rather become mostly indecipherable to the human ear. Again, amplitude is the prime consideration, with phase angle ranking way down the list. At frequencies above about 8 KHz, pitch recognition in most people goes away, anyway, so phase is totally irrelevant UNLESS the change in phase angle changes the difference products of IM distortion. THEN you open a whole new kettle of fish! dBdB |
#15
|
|||
|
|||
"Michael Putrino" wrote in message ... Yes, I agree...room is everything. What I was trying to convey, but did so badly, was that I don't think that it would make much difference whether switching the phase on the top mic or the bottom mic. An outward or inward pressure from the micing is not what you would normally hear in a performance anyway...so pick one...hopefully the one that sounds better to you at the time...mixed in with the whole band. Record them on seperate tracks and you can flip to your hearts content. Ahhh... gotcha'. In my main recording room, the owner has a dozen or so polarity reversed cables which he uses religiously on toms and other things wherein he believes that the downward motion of the instrument, due to the initial attack being a movement away from the microphone, will make a better and more accurate recording. I've tried it and could tell diddly-squat difference in my results, but he lives by it and won't consider anything else. So... point now well taken. ;-) -- David Morgan (MAMS) http://www.m-a-m-s DOT com Morgan Audio Media Service Dallas, Texas (214) 662-9901 _______________________________________ http://www.artisan-recordingstudio.com |
#16
|
|||
|
|||
"David Morgan (MAMS)" wrote:
Unfortunately, 'realistic' is very often not do-able because of bad rooms. OT to the subject, but here is my thought on *realistic* sounds and drums. I don't think Realistic sounding drums is possible since *realistic* is too relative a term for anything as eclectic as drums. First where do we hear drums? Clubs? Bars? Concerts? Living Rooms? What constitutes a realistic drum sound then? It can't be in a nice recording studio room with great acoustics, since very few places that we listen to drums are like that, and rarely do people listen to drums in those kinds of rooms. But we don't record drums to sound like the aforementioned locations either (well mostly we don't). So it appears recording *natural* sounding drums is not about letting them sound natural like they really do in the majority of rooms we listen to them in, rather it is about eliminating the bad acoustical drum noise to make room in a recording for some good acoustical drum noise. Which BTW I think you do point out later in your post. Am I off base in this thinking though? -- Nathan "Imagine if there were no Hypothetical Situations" |
#17
|
|||
|
|||
|
#18
|
|||
|
|||
"Karl Winkler" wrote in message
oups.com... However, I think it would be better to flip the *top* mic out of polarity, since when the skin is first hit, it goes *down* on the heads, thus pulling the diaphragm of the top mic out and pushing the diaphragm of the bottom mic in. Thus to get a positive excursion at the speaker, the bottom mic should be used as the reference, with the top mic "flipped" to match. Undoubtedly, there are engineers who are already doing this, and maybe I've exposed a "secret". Sorry about that... No, I don't think so, anyway... the response of the head is fast enough that unless you're both top & bottom mic'ing a given drum, the difference is negligible, relative to the whole kit... unless by doing so you generate other phase related issues such as how that mic's signal now relates to, let's say, that of the overheads. Think about it... you've got all these variables to consider: 1.) Is the mic ONLY picking up the sound generated by the center of the head, where the stick (assumedly) strikes? No, it's also picking up the sound from the shell, and from the outer edges of the head (which generate a wave faster than the center does, since the edges have less distance to travel on "recoil" than the center of the head does. So it's a pretty complex sound that a drum mic is picking up, even apart from reflections from room surfaces. 3.) Is the mic pointed with the capsule directly downward right at the strike point? Never - except in the case of a mic stuck right in front of a kick drum beater, perhaps... therefore you've got more "relative" phase happening than "absolute" phase in every circumstance on each drum. 3.) The mic is also picking up reflections from the floor, and any surrounding walls - how do these reflections relate to, for example, those picked up from the overheads if you were to flip the phase on a top snare mic? 4.) Is the few microseconds of difference in when the sound arrives at the mic on a single close-mic'ed snare (again assuming top-micing only) if you flip the polarity, going to make a detectable difference in phase relative to - let's say - the kick mic, which is normally/often mic'ed so that the head is in excursion relative to the mic diaphragm when it's "kicked"? Probably not - what's more likely to happen is that the combination of the waves generated by the kick shell & the floor & the surrounding wall surfaces are going to have more of an impact as to whether it sounds more in-phase or out of phase. Same goes for what the in-phase kick mic picks up from the snare when the snare is struck (which is normally/often at a much lower level if the mic is located inside the kick). There's more, I'm sure, but off the top of my head, that's the stuff that immediately comes to mind. I've tried messing around with what you mentioned before, just out of curiosity, and to me it just makes more sense to keep everything in the same relative polarity, but just maintain awareness of the normal stuff that can cause phase issues (distances between mics, esp. how the overheads are set up, dealing with reflections, etc). Anyway, having said all that; Karl, have you tried that which you mentioned, and do you prefer it that way? Neil Henderson |
#19
|
|||
|
|||
"David Satz" wrote in message oups.com... Walter, I think your experiment is a good one, but any conclusions need to be carefully drawn for at least two reasons that I can think of: [a] It's crucial to define exactly what question is being answered. Psychoacoustics describes (and the anatomy of human hearing supports) an ability to hear relative phase in the sense that you're using the term, but only below 1500 Hz or so. Above that range the ability disappears, Hey David... since the ear has a natural presence peak at around 3k, wouldn't one be able to detect things in that range even more readily? Neil Henderson |
#20
|
|||
|
|||
DeserTBoB wrote: On Thu, 9 Dec 2004 23:47:13 -0800, "Walter Harley" wrote: Two samples each containing a signal comprising 220Hz, 440Hz, 660Hz, and 880Hz sines at the same relative levels but with different phase relationships sound pretty different, to me. (That is, I can reliably tell them apart in blind randomized trials.) snip This has been known for years. The Hammond "organ", first sold in 1935, uses additive synthesis in a failed attempt to recreate the sound of various organ stops. Due to the construction of the tonewheel generator, all the tonewheel are in a different phase relationship every time the organ starts up, since all the wheels are clutch driven and slip slightly upon startup. The combination you cite above would be equivalent to A below Middle C with the 8', 4', 2 2/3' and 2' drawbars pulled out. However, there's a catch here, since Hammonds are roughly tuned (not exactly) to Equal Temperatment, so the 2 2/3' pitch would be slightly flat from Just Temperament, and thus, not exactly 660 Hz. In this case, the 2 2/3' pitch would be 659.255 Hz in ET, and Hammonds "stretch and shirnk" ET just a tad here and there because of the limitations of the mathematics of the tonewheel generator. In any event, there'll be some beating from this, regardless of phase. Every time the organ is started up, this combination will sound different to the ear. The same will go when you use tones derived from a top octave generator, or a bank of free running oscillators locked with a PLL; vary the phase, the tone will sound slightly different there and there. This is not the same thing as the OP talked about. The OP talked about one note and harmonics of that note. What you described are differenct notes of an organ and the fact that when they are not exactly equally tempered it gives the organ body which is a good thing. These are two different things. Your ear cannot perceive a change in phase between the fundamental and the harmonics unless there is a non-linearity that distorts the waveform which then changes the amplitude of additional harmoincs. Mark |
#21
|
|||
|
|||
"David Satz" wrote in message
oups.com... Walter, I think your experiment is a good one, but any conclusions need to be carefully drawn for at least two reasons that I can think of: [...] Both very good points. I do discuss those in my writeup. -w |
#22
|
|||
|
|||
"Mark" wrote in message
oups.com... Your ear cannot perceive a change in phase between the fundamental and the harmonics unless there is a non-linearity that distorts the waveform which then changes the amplitude of additional harmoincs. Your ear might not be able to, but I just demonstrated that my ear can. I do not believe there is any substantive nonlinearity in the system on which I explored this (described in my writeup). It is precisely this (mis-)conception which I hoped to address. Out of interest: Mark, can you hear the difference between the two .wav files? -walter |
#23
|
|||
|
|||
"Scott Dorsey" wrote in message
... What the original poster is measuring is the audibility of group delay. When someone says "relative phase" I figure they are talking about phase differences between channels. Thanks for the correction in terminology. I'll update my web page. There is some good research on the audibility of group delay out there. Including Koray Oczam's paper, AES preprint 5740. ....and thanks for the reference. It was an article in the latest JAES that motivated me to go explore this (I've always wondered about the schism between people saying it's inaudible and people complaining about graphic EQ's screwing up the sound, but never done anything about it). I've just downloaded Oczam's paper and will check it out. -walter |
#24
|
|||
|
|||
Walter Harley wrote:
...and thanks for the reference. It was an article in the latest JAES that motivated me to go explore this (I've always wondered about the schism between people saying it's inaudible and people complaining about graphic EQ's screwing up the sound, but never done anything about it). I've just downloaded Oczam's paper and will check it out. Well, graphic EQs screw up the sound in enough different ways that the group delay issue may not even be the most serious one. Just looking at the actual frequency response of a graphic configured for a gradual rise will make you feel queasy. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#25
|
|||
|
|||
Walter Harley wrote: "Mark" wrote in message oups.com... Your ear cannot perceive a change in phase between the fundamental and the harmonics unless there is a non-linearity that distorts the waveform which then changes the amplitude of additional harmoincs. Your ear might not be able to, but I just demonstrated that my ear can. I do not believe there is any substantive nonlinearity in the system on which I explored this (described in my writeup). It is precisely this (mis-)conception which I hoped to address. Out of interest: Mark, can you hear the difference between the two ..wav files? -walter Why no Walter, I could not hear a difference but that means nothing. Have you demonstated that you hear a difference using a double blind test as you describe in the article? If you wish to address the (mis)conception as you say, then you need to also verify the validity of the experiment by checking the 2 waveforems with an oscilloscope and a spectrum analyzer. Please use the scope to verify that neither waveform is being distorted and use the SA to verify that all the harmonics are at the same relative amplitude in both cases. This will verify that the ONLY thing different is the phase and that there are no additional harmoincs being added or that the amplitude of the ones you put in are changed. If you can verify the experimient is valid this way, and can still statistically hear a differnce in a double blind test, then you will begin to get my attention. I applaud you for questioning the party line but you must apply rigorous checks or else its cold fussion. Mark |
#26
|
|||
|
|||
Walter, I just tried your experiment in my lab in a different way. I used 2 audio oscillators, set 1 to 440 and the other to 880. I summed them and looked at the combination waveform on the scope and listend on my monitors. Since the 880 is not exactly 2x the 440, the relative phase drifts through slowly. To my surprise, I could hear a difference as the waveform changed. But then I thought about it realized that my 440 generator (and yours) is not perfect and generates some 880. This 880 combines with the 880 that I added from the other generator and as the phase relationship changes, the AMPLITUDE of the 880 changes. This in fact is what it sounded like. So again, I belive you need to verify your experimental setup to ensure that ALL the harmonics are at the same AMPLITUDE in both waveforms. A change in the amplitude of the harmoics will obviously change the "timbre" of the sound. Thanks Mark |
#27
|
|||
|
|||
Hi Walter,
This is an interesting experiment. However, I feel, as some other posters do, that there are some sources of error in it. 1) The maximum peak to rms ratio becomes larger as the number of tones added together is increased. Hence, to keep the maximum peak to rms ratio to a minimum, I suggest that you use only two tones. 2) Sound generation equipment generates harmonic distortion. By chosing tones that are harmonically related, you will augment or diminish the harmonics generated by the equipment thereby accentuating the effect. So, I suggest that you not only use two tones but that they be not related harmonically. Perhap 220 Hz and pi*220 Hz may work. 3) Putting two or more tones through an amplifier results in intermods. I suggest that you put one tone in the left channel and the other in the right channel just to get this out of the picture. Btw, I did listen to your experiment on junky cheap powered computer speakers I bought many years ago and could not hear any difference. (This being rec.audio pro should I run for cover? But I do believe that on better equipment the difference would be audible. And I did see that the in-phase set had higher peaks than the out-of-phase set. Joe |
#28
|
|||
|
|||
the ear has a natural presence peak at around 3k
That isn't quite what equal loudness curves (e.g. Fletcher/Munson) represent, but OK--your meaning is clear enough. wouldn't one be able to detect things in that range even more readily? That just doesn't turn out to be true in practice. So you've just offered a highly intelligent explanation for a fact that doesn't exist. --best regards |
#29
|
|||
|
|||
"David Satz" wrote in message oups.com... the ear has a natural presence peak at around 3k That isn't quite what equal loudness curves (e.g. Fletcher/Munson) represent, but OK--your meaning is clear enough. wouldn't one be able to detect things in that range even more readily? That just doesn't turn out to be true in practice. So you've just offered a highly intelligent explanation for a fact that doesn't exist. --best regards Hmmm... I don't get what you're saying - can you run that by me again? That was a serious question, BTW - not a flippant remark, if that's what you thought. Neil Henderson |
#30
|
|||
|
|||
"Neil Henderson" wrote in message
. com "David Satz" wrote in message oups.com... Walter, I think your experiment is a good one, but any conclusions need to be carefully drawn for at least two reasons that I can think of: [a] It's crucial to define exactly what question is being answered. Psychoacoustics describes (and the anatomy of human hearing supports) an ability to hear relative phase in the sense that you're using the term, but only below 1500 Hz or so. Above that range the ability disappears, Hey David... since the ear has a natural presence peak at around 3k, wouldn't one be able to detect things in that range even more readily? For what this "Me too" post is worth, David's got it exactly right. The normal given frequency for the point where the ear is most sensitive based on intensity, is more like 4 KHz than 3. The reason why is usually given is the ear-canal resonance that you seem to be referring to. While the ear is most sensitive to sound based on intensity at about 4 KHz , the ear is most sensitive to other aspects of sound at other frequencies. For example, the ear is most sensitive to FM distortion when the FM modulation occurs at very low frequencies, a few Hz. The ear is often most sensitive to nonlinear distortion when the test signal is at some frequency other than 4 KHz, but there is a spurious tone generated by the distortion that appears around 4 KHz, and so on. |
#31
|
|||
|
|||
On Sun, 12 Dec 2004 01:20:50 -0500, David Satz wrote
(in article .com): the ear has a natural presence peak at around 3k That isn't quite what equal loudness curves (e.g. Fletcher/Munson) represent, but OK--your meaning is clear enough. wouldn't one be able to detect things in that range even more readily? That just doesn't turn out to be true in practice. So you've just offered a highly intelligent explanation for a fact that doesn't exist. --best regards The FM curve illustrates that at low levels the 3kHz area is dominant. As the SPL rises, however, the peak is not as dominant. Regards, Ty Ford -- Ty Ford's equipment reviews, audio samples, rates and other audiocentric stuff are at www.tyford.com |
#32
|
|||
|
|||
"Arny Krueger" wrote in message ... While the ear is most sensitive to sound based on intensity at about 4 KHz , the ear is most sensitive to other aspects of sound at other frequencies. OK, I get what you're saying now. Thanks! Neil Henderson |
#33
|
|||
|
|||
On 10 Dec 2004 20:35:10 -0800, "Mark" wrote:
This is not the same thing as the OP talked about. snip Yes, it is. Read this post. The OP talked about one note and harmonics of that note. snip ....which is what I was talking about. What you described are differenct notes of an organ and the fact that when they are not exactly equally tempered it gives the organ body which is a good thing. These are two different things. snip Wrong on two counts. On a "real" organ, mutation stops, which coincide with the third, fifth, and sixth (and on up) harmonics of the fundamental, are tuned to Just Temperament. Unit organs, such as the Wurlitzer used in theaters in the '20s, and other highly unified organs, have derived mutations from their fundamental ranks, which is tuned to Equal Temperament, and thus all dissonent mutation stops will be technically "out of tune." The Hammond (except for the failed G-100) use this same sort of unification, and thus, only the third, fifth, sixth and the third harmonic of the suboctave fundamental are mutations, or dissonant false "harmonics", which are actually tuned approximately to Equal Temperament. All other tones (the consonent harmonics) will be locked in phase once the tonewheel generator gets up to speed, unless there is a defective tonewheel clutch somewhere. You can easily prove this with a scope on the output. Your ear cannot perceive a change in phase between the fundamental and the harmonics unless there is a non-linearity that distorts the waveform which then changes the amplitude of additional harmoincs. snip Sensitivity to phase angle of harmonics of a fundamental was thought for years to be imperceptible to the human ear, but research in the '80s proved otherwise, but nothing major enough that really changes any predating basic theories about timbral synthesis. As stated earlier, Allen Organ, and developers of the Musicom system in the UK, discovered that changes in relative phase DO influence the listener's perception of "timbre," which, as Helmholz defined, is the sum total effect of a fundamental and its various harmonics at varying amplitudes which makes up the signature sound of a tuned instrument. However, this effect only happens in the most pitch sensitive area of human hearing, determined in the 1920s to be that area between 200 and 1800 Hz. Thus a 220 Hz (A below middle C) note on, say, a slender scaled organ stop, will have a harmonic train of as many as 50 harmonics, all of which will have their own pecular phase relationship to the fundamental. Now, the first three harmonics are both the most discernable as to pitch, whereas pitch accuity above that point becomes progressively poorer. Not a worry, since these harmonics are locked in tune with the fundamental...or are they? In this example, they surely are NOT. Due to the resistance to the propagating wave front within the slender scaled pipe body of such a stop, these first three harmonics actually undlate in phase enough to cause a very tiny, almost inperceptable "beating" amongst themselves. Tests determined that the listener CAN determine when this phase shift occurs, even though the relative amplitudes of all concerned harmonics varies less than .5 dB, an almost imperceptible change in amplitude to the human ear even under the best of conditions. As the pitch of each related harmonic rises, this "off phase" behavior increases with frequency until, in the upper regions of audibility, they even waver off pitch. The ear, however, cannot discern this pitch change per se, but CAN detect the sum and difference byproducts (think IM distortion) and the changes in amplitude caused by beating and phase cancellation. It was found by Fast Fourier Analysis that the phase relationships of this harmonic train don't stay in relative phase with each other, either, which is to say that a 30° negative shift in phase of one particular harmonic doesn't necessarily guarantee a 60° negative shift of the next consonant harmonic in the train. In another example I gave, using a top octave generator/divider scheme, the harmonics WILL indeed stay in abosulute locked phase, regardless of how high in frequency the harmonics extend, the limit of which is the frequency outputs of the TOG itself. You can, however, by using LC components, shift the phase of various harmonics after they're "divided out" and bought out seperately, and thus you can test the question that way, as has been also done in the past. Locked phase relationship is why "divider organs" (Thomas, Lowery, others) always sounded even more sterile than the Hammond, which was bad enough. At least the imperfections of the Hammond due to tonewheel magnetization and other noises would provide a LITTLE interest to the composite tone; the divider organs had none. Example of a divider organs that failed to live up to the Hammond name: The X-66. All of this is why it took makers of electronic posuers so long to figure out why their imitation strings would sometimes sound like reeds (or, worse, "frying bacon.") It's still very difficult indeed for a digital system, locked in phase such they are unless using VERY long samples/models and VERY fast clocks, to faithfully produce the sound of an organ string stop, especially one of very fine scale and bright timbre. For that matter, a real violin or viola is a challenge not met by synthesizers either, for much the same reasons, except that now we're dealing with a mechanically excited string instead of a narrow air column, and you're also dealing with the technique of the player. Back in the box. Go into your room, fire up that old tonewheeled "coffee grinder" Hammond, connect a scope to one of the G-G terminals (yes, old Hammonds have a tip-ring output, but with no center tap), draw out the first, third, fourth, sixth and ninth drawbars to 8. Observe. You will see LOTS of non-pitch related stuff going on there (due to magnetized tonewheels, preamp distortion and noise, the works) but you won't see much, if any, variation in relative phase between the various consonent harmonics playing. Fine. If you have a memo scope, take a snapshot now. Shut down the Hammond and let it coast down. Fire it back up again, then look at the same note played the same way. The waveform will be different, because all (or most) of the tonewheels involved have slipped their clutches upon start up. Does it sound the exact same? Close...but listen REAL close. It's a LITTLE different now. You have to be able to filter out all the "crap" that in even good Hammonds (most players in rock prefer really screwed up Hammonds), as magetized tonewheels will provide inharmonic "thumping" that's very obvious in the output waveform. Concentrate on the sinusoids...they will either be phase locked, or pretty damned close to it. Now, add a touch of the fifth drawbar, which is a "borrowed" mutation at 2 2/3' pitch. The wave form will start undulating wildly, as the Equal Temperament of this pitch beats with the fundamental and all its harmonics. Thus, you can't include any dissonant harmonics in such a test of phase accuity when using a Hammond organ. By the way, it's NOT that feature which provides the "warmth" of a Hammond organ...it's that 22-H or 122 Leslie over there. Played through the board (G-G goes to tip and ring, and in ya go...source impedance: 220 ohms at around a +12 wide open...pad it down!) or even through one of Hammond notoriously lousy tone cabinets, the Hammond sounds horrid...unmusical as all hell. I don't even like them through a Leslie, but I'm a classically trained organist, so Hammonds to me are an abortion for real organ music. For R&B, pop and rock, they have a signature sound WITH the Leslie that's and unavoidable part of the fabric of American popular music. Only until very recently have digital poseurs have even come close to emulating the sound of a tonewheel organ because of 1.) failure to realize that real Hammonds are NOT exactly tuned to Equal Temperament due to the mathematics of running 91 tonewheels from only one 1200 RPM motor shaft, and 2.) there's a lot more than just a bunch of added sinusoids going on in that tonewheel generator. Best one I've heard to date: The Voce. The Hammond products...wellll, they tried. Roland's VK-7 and VK-77 missed the mark by a bit, too, but are pretty good. The Hammond-Suzuki "new B-3" (not the XB-3...eh) is REAL good (uses bifurcated palladium keying switches, just like the original), but at $50K a copy, I wouldn't look for them in your local garage band anytime soon. The XB-3? Well, let's just say it was a nice try. dB |
#34
|
|||
|
|||
DeserTBoB wrote:
By the way, it's NOT that feature which provides the "warmth" of a Hammond organ...it's that 22-H or 122 Leslie over there. Hey, those'll even warm up a Telecaster on bridge pickup. g -- ha |
#35
|
|||
|
|||
snipped lots of interesting information about tone wheels etc. That was interesting and I thank you for it. However, if I understood you correctly the harmonics in a tone wheel organ are created by various wheels which may slip at startup and therefore have arbitrary phase relationships to the fundamental. Fine. But you also said yourself that these wheel generated harmoincs combine with the "actual" harmonics created by the fundamental wheel and as the phase changes the AMPLITUDE of the combined harmonics change. There is no argument that these amplitude changes are audible. You also said yourself that the harmonic amplitudes changed about 0.5 dB. This is a lot of change for the kind of subtle effects we are talking about here. You also seem to be saying that if the harmonic is off pitch, that is audible. OK fine thats a frequency change and I certainly agree that a frequency change can be audible as a pitch shift. So I stand my my original contention, the phase relationship of the harmonics to the fundamental are not audbile. Any AMPLITUDE changes to the harmonic are audible as a change in timbre. Any FREQUENCY changes to the harmonic can be audible as a pitch shift. You have not cited a case where the organ gnerates a harmonic phase change that was audible without also a change to the amplitude or frequency of the harmonic. thanks Mark |
#36
|
|||
|
|||
I don't even like them through
a Leslie, but I'm a classically trained organist, so Hammonds to me are an abortion for real organ music. Nah, it's just a whole different art form, as different from classical organ as ballet is from sculpture. For R&B, pop and rock, they have a signature sound WITH the Leslie that's and unavoidable part of the fabric of American popular music. BRBR And jazz, too. Scott Fraser |
#37
|
|||
|
|||
it's that 22-H or 122 Leslie over there.
Hey, those'll even warm up a Telecaster on bridge pickup. g BRBR They've done some great stuff to some violin & viola tracks of my acquaintance. too. Scott Fraser |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Equalizers | Audio Opinions | |||
Doppler Distortion - Fact or Fiction | Pro Audio | |||
Transient response of actively filtered speakers | Tech | |||
science vs. pseudo-science | High End Audio | |||
Blindtest question | High End Audio |