Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Recording level and S/N ratio
Arny,
In another thread, you suggested recording my albums at 6-10dB down, and then normalizing up closer to 0.0 dB once all of my noise reduction and de-clicking was done. I understand the reasoning behind this - so I have adequate headroom during the entire recording. My question is, doesn't recording at 6-10 dB down lower my signal-to-noise ratio by the same amount? I'm only concerned about it because I think I have a high noise floor to begin with. (I'm not sure if I'm using that term correctly). When I have my turntable and preamp turned on, but I'm not playing an album, the level meter in Sound Forge is at about -45 dB. The noise is coming from my $45 preamp. I didn't want to disrupt my main stereo system for the couple of months it's going to take me to record all these albums, so I'm using an old turntable and a cheap phono preamp. I know the noise is coming from the preamp, because if I disconnect the turntable from the preamp, the meter stays at -45 dB, but if I disconnect the preamp from the soundcard, it drops to -69 dB, which is almost the bottom of SoundForge's scale. The recordings I've made so far sound fine - I guess the 44 dB difference between my "noise floor" (-45 dB) and the peaks in my music (~-1.0 dB) is more than adequate, but I'm concerned about the effects on my S/N ratio if I lower the recording level by 6-10 dB. Scott Gardner |
#2
|
|||
|
|||
Recording level and S/N ratio
"Scott Gardner" wrote in message ... When I have my turntable and preamp turned on, but I'm not playing an album, the level meter in Sound Forge is at about -45 dB. The noise is coming from my $45 preamp. I didn't want to disrupt my main stereo system for the couple of months it's going to take me to record all these albums, so I'm using an old turntable and a cheap phono preamp. I know the noise is coming from the preamp, because if I disconnect the turntable from the preamp, the meter stays at -45 dB, but if I disconnect the preamp from the soundcard, it drops to -69 dB, which is almost the bottom of SoundForge's scale. Sound Forge has switchable level meters which can extend down to over -120 dB if you want to. The recordings I've made so far sound fine - I guess the 44 dB difference between my "noise floor" (-45 dB) and the peaks in my music (~-1.0 dB) is more than adequate, but I'm concerned about the effects on my S/N ratio if I lower the recording level by 6-10 dB. Your noise floor is, by most standards, pretty goddamn high. You're essentially recording at 8 bits, which is not very good. If you know that your problems are coming from your preamp, I'd suggest that you invest in a bit more expensive and better preamp. |
#3
|
|||
|
|||
Recording level and S/N ratio
"Scott Gardner" wrote in message
Arny, In another thread, you suggested recording my albums at 6-10dB down, and then normalizing up closer to 0.0 dB once all of my noise reduction and de-clicking was done. I understand the reasoning behind this - so I have adequate headroom during the entire recording. My question is, doesn't recording at 6-10 dB down lower my signal-to-noise ratio by the same amount? I'm only concerned about it because I think I have a high noise floor to begin with. (I'm not sure if I'm using that term correctly). Recording 6-10 dB below peak levels will reduce your effective dynamic only if you are recording a source that has equal or greater dynamic range than your recording device. When I have my turntable and preamp turned on, but I'm not playing an album, the level meter in Sound Forge is at about -45 dB. OK. The noise is coming from my $45 preamp. I didn't want to disrupt my main stereo system for the couple of months it's going to take me to record all these albums, so I'm using an old turntable and a cheap phono preamp. OK I know the noise is coming from the preamp, because if I disconnect the turntable from the preamp, the meter stays at -45 dB, but if I disconnect the preamp from the soundcard, it drops to -69 dB, which is almost the bottom of SoundForge's scale. OK The recordings I've made so far sound fine - I guess the 44 dB difference between my "noise floor" (-45 dB) and the peaks in my music (~-1.0 dB) is more than adequate, but I'm concerned about the effects on my S/N ratio if I lower the recording level by 6-10 dB. Currently, your noise floor is -45 dB and it's set by your phono preamp. As you drop the gain coming into your sound card, every dB you drop the peaks will also drop the noise by one dB. This happens until you the noise floor drops down around -60 dB or. Some place around -60 dB, the noise from the sound card will start becoming significant. Obviously, when your noise floor due to the preamp goes down to -69 dB, the noise from the sound card will keep the total system noise floor from dropping any further. You should be able to verify these estimates without too much work. Try it and tell us what happens! |
#4
|
|||
|
|||
Recording level and S/N ratio
"Tommi" wrote in message
"Scott Gardner" wrote in message ... The recordings I've made so far sound fine - I guess the 44 dB difference between my "noise floor" (-45 dB) and the peaks in my music (~-1.0 dB) is more than adequate, but I'm concerned about the effects on my S/N ratio if I lower the recording level by 6-10 dB. Your noise floor is, by most standards, pretty goddamn high. You're essentially recording at 8 bits, which is not very good. If you know that your problems are coming from your preamp, I'd suggest that you invest in a bit more expensive and better preamp. I'll second that suggestion. When I digitize vinyl, the needle-up noise floor is well over 65 dB below peak recorded levels. I use a Conrad Johnson preamp or a Apt/Homan preamp, and a DAL Card Deluxe sound card. If I turn off the preamp, the noise floor is more than 90 dB down. The actual needle-down noise floor varies tremendously, from about 50 dB down to about 60 dB down. |
#5
|
|||
|
|||
Recording level and S/N ratio
On Tue, 9 Dec 2003 13:51:21 -0500, "Arny Krueger"
wrote: Currently, your noise floor is -45 dB and it's set by your phono preamp. As you drop the gain coming into your sound card, every dB you drop the peaks will also drop the noise by one dB. This happens until you the noise floor drops down around -60 dB or. Some place around -60 dB, the noise from the sound card will start becoming significant. Obviously, when your noise floor due to the preamp goes down to -69 dB, the noise from the sound card will keep the total system noise floor from dropping any further. You should be able to verify these estimates without too much work. Try it and tell us what happens! Okay, back from the doctor's. I haven't re-recorded any more music, but I have played around with the level meters and the gain on the sound card. I also figured out a more accurate way to read the meters, so the noise floor is actually -40 dB, not -45 dB. Also, the minimum noise with the input muted is -76 dB , not -69 dB. With the preamp plugged into the sound card and the input gain on the sound card set all the way up, the noise floor is -40 dB. If I turn the input gain on the sound card all the way down, the noise floor drops to -76 dB, so I guess that's the noise from the sound card that I won't be able to get rid of without switching cards. With the input gain set at halfway, the noise is about -48 dB. So, where would be the best place to put the input gain on the sound card? Aim for the same 6-10 dB down that you were talking about earlier? Thanks, Scott Gardner |
#6
|
|||
|
|||
Recording level and S/N ratio
Here's an update.
I grabbed the first album I found that I had both on LP and CD. This turned out to be the "Cocktail" soundtrack. I recorded two songs off of both formats - "Kokomo" and "Rave On". Here are the peak and RMS levels for both songs on both formats: Kokomo CD: Peak -1.3 RMS -18.3 Kokomo LP: Peak -2.2 RMS -21.9 Rave On CD: Peak -0.7 RMS -18.1 Rave On LP: Peak -0.3 RMS -21.9 The two copies of "Kokomo" only differed by 0.8 dB on the peak, but there was a difference of 3.6 dB in the RMS level. Likewise, the difference between the two copies of "Rave On" were 0.4 dB peak, but 3.8 dB RMS. So, in both cases, it looks like the recording engineer that made the CD decided to raise the RMS level, thus compressing the dynamic range. I normalized the LP versions of both songs to an RMS of -18 dB, and they now have the same apparent volume as the CD copies. Thanks so much for all the suggestions. I would never have guessed that the CD versions would actually have a *smaller* dynamic range than the LP versions. I mistakenly assumed that since CD had a higher potential dynamic range, that the range for the CDs would in fact be higher. So now that I know I can normalize the RMS levels to keep the volume level more consistent across my songs, this should make things easier. So, back to the original question that started these threads. I'm assuming it's better to do the noise reduction and de-clicking before I adjust the volume levels. Does this sound correct? Thanks again for all the suggestions. Scott Gardner |
#7
|
|||
|
|||
Recording level and S/N ratio
"Scott Gardner" wrote in message
So, back to the original question that started these threads. I'm assuming it's better to do the noise reduction and de-clicking before I adjust the volume levels. Does this sound correct? As long as the volume levels are reasonable, the de-clicking algorithms generally automatically adjust for them. |
#8
|
|||
|
|||
Recording level and S/N ratio
"Sang Li" wrote in message
"Arny Krueger" wrote: Arny, In another thread, you suggested recording my albums at 6-10dB down, and then normalizing up closer to 0.0 dB once all of my noise reduction and de-clicking was done. I understand the reasoning behind this - so I have adequate headroom during the entire recording. My question is, doesn't recording at 6-10 dB down lower my signal-to-noise ratio by the same amount? I'm only concerned about it because I think I have a high noise floor to begin with. (I'm not sure if I'm using that term correctly). Recording 6-10 dB below peak levels will reduce your effective dynamic only if you are recording a source that has equal or greater dynamic range than your recording device. ********! Easy to say, hard to prove. How typical of you to disregard noise just because it's there. This sentence doesn't make any sense. I'm sure the OP would like to retain the quality of the captured recording, That would be a reasonable goal. in which case he may choose to convert the file to 24bit mode before making gain adjustments. He may choose to do it, but it would be an exercise in futility. The noise floor he's working with is sooo high that the difference between 16 and 24 bits is so thoroughly swamped that is would be an exercise in futility. His system noise floor is something like 45 dB down, while the noise floor of 16 bits is more than 90 dB down. Remember, dBs are logarithmic, so this is far more than a simple arithmetic difference of 2:1. The difference involves twice as many powers of 10. |
#9
|
|||
|
|||
Recording level and S/N ratio
Sang Li wrote in
: "Arny Krueger" wrote: Arny, In another thread, you suggested recording my albums at 6-10dB down, and then normalizing up closer to 0.0 dB once all of my noise reduction and de-clicking was done. I understand the reasoning behind this - so I have adequate headroom during the entire recording. My question is, doesn't recording at 6-10 dB down lower my signal-to-noise ratio by the same amount? I'm only concerned about it because I think I have a high noise floor to begin with. (I'm not sure if I'm using that term correctly). Recording 6-10 dB below peak levels will reduce your effective dynamic only if you are recording a source that has equal or greater dynamic range than your recording device. ********! Since even a pristine LP is about 20dB noisier (not even counting infrasonic noise) than a good soundcard running 16 bits then recording at 6-10 dB below will have only a miniscule effect on signal to noise. That said my target is 3-6 dB below. Tim -- "The strongest human instinct is to impart information, and the second strongest is to resist it." Kenneth Graham |