Digital volume and bit depth.

[Taupe1]

Hey there so I have a question regarding sampling off streaming services for making music.

I’m using qobuz 16/44 only tracks

Now if I want to record these at -6dbfs peaks I have to lower either the qobuz app or windows Master volume to avoid clipping and especially to get -6dbfs peaks. Mostly to 50-60% volume. From what I grasp digital volume reduction reduces bit depth and I’m wondering if it really matters because most of the time the samples would be mixed lower at -10dbfs or so but sometimes boosted a bit and I don’t want the noise floor or other artifacts to become and issue later.

For context it’s drum breaks for old boom bap style hip hop so fidelity isn’t super important but I’d still like to avoid messing the audio up more than I need to.

Here’s some options I thought of and would like to know which would be the preferred option ( in theory) all recorded with loopback in audacity

Record the 16/44 source with the volume reduced to reach -6dbfs peaks and deal with the “bit loss” and noise etc ( if it even becomes a problem with the above in mind)

Record the 16/44 source with volume lowered just enough to not clip minimizing the changes to the source and worry about gain reduction /boosting later in the DAW when mixing (since there won’t be as much of a change as the first option this may be better)

Record the 16/44 source at 24/44 with volume lowered enough to peak at -6dbfs (would those extra bits help?)

Record at 16/44 with volume lowered just enough to avoid clipping and use the “amplify” option in audacity with a negative gain value to set the peaks at -6dbfs

Sorry if this is long winded just want to get it right before I do a ton of sampling and have to do it over lol.

 

[DVDdoug]

Now if I want to record these at -6dbfs peaks I have to lower either the qobuz app or windows Master volume to avoid clipping

If you don't boost the volume you won't get clipping (although there can be existing clipping).
With integer formats, 0dB is defined as the maximum you can "count to" (positive & negative) with a given number of bits. Nothing bad happens when you get close to 0dB but you'll clip if you try to go over.

and especially to get -6dbfs peaks. Mostly to 50-60% volume. From what I grasp digital volume reduction reduces bit depth and I’m wondering if it really matters because

No it's not a big deal. When the volume in the recording drops or fades to -6dB you're only using 15-bits. When it fades to -48dB you're down to 8-bits. But you don't notice any loss of resolution because it's quieter and the quantization noise is still down around -96dB.

If you make a full-volume 8-bit file, you'll hear quantization noise, especially on the quiet parts.

If you record at 24-bits and then you can lower the volume without losing resolution. By default, Audacity converts everything to 32-bit floating-point.

For context it’s drum breaks for old boom bap style hip hop so fidelity isn’t super important but I’d still like to avoid messing the audio up more than I need to.

If you are mixing, you'll have to lower the volume because mixing is done by summation. And because of the summing, digital mixing can increase the bit depth. Analog mixers are built-around summing amplifiers but of course there are level controls for each input, plus a master volume control.

 

[Taupe1]

That makes sense now, I’ve noticed on some recorded tracks (especially older songs) there is some clipping present in the track but lowering down the volume just a smidge in qobuz seems to at least not make it clip in my recording of it which is all I can ask for I think. I’ve started recoding everything as close to 0dbfs as I can which is typically full volume, checking with audacity using Check for clipping, show clipping and then amplify to see if there is a tiny bit of headroom by seeing how far it would have to be amplified to hit 0dbfs (some of these tracks only leave like 0.3 or less guess you learn something Every day !) so i suppose that’s my due diligence to get a good recording. I appreciate you educating me though.


So the “loss of bits” by just lowering the volume on qobuz a tiny bit (maybe 5-10% at most) would probably be so negligible if you’re saying -6 is only 1 bit in a 16bit track that I won’t even bother giving it any mind ?

In my case if the source format is 16/44 is there really a point in recording at 24bits if the volume reduction I’m doing is so little and you mentioned the loss in resolution wouldn’t even be noticeable ? The 24 bits recording and loss in resolution only applies to volume decreases or increases when I’m recording it right? I can lower it Via faders in the DAW without a resolution loss ?

 

[Taupe1]

If you don't boost the volume you won't get clipping (although there can be existing clipping).
With integer formats, 0dB is defined as the maximum you can "count to" (positive & negative) with a given number of bits. Nothing bad happens when you get close to 0dB but you'll clip if you try to go over.


No it's not a big deal. When the volume in the recording drops or fades to -6dB you're only using 15-bits. When it fades to -48dB you're down to 8-bits. But you don't notice any loss of resolution because it's quieter and the quantization noise is still down around -96dB.

If you make a full-volume 8-bit file, you'll hear quantization noise, especially on the quiet parts.

If you record at 24-bits and then you can lower the volume without losing resolution. By default, Audacity converts everything to 32-bit floating-point.


If you are mixing, you'll have to lower the volume because mixing is done by summation. And because of the summing, digital mixing can increase the bit depth. Analog mixers are built-around summing amplifiers but of course there are level controls for each input, plus a master volume control.

So basically you’re saying that I shouldn’t really need to lower the volume at all because if I’m understanding correctly any clipping in the song (especially since they are all older before 1990 or so ) is likely to be analog and not sound bad in my audacity wasapi loopback recording

AND

That it shouldn’t digitally clip my recording because it’s been recorded at a fixed volume previously ?

If I check audacity “amplify” it’s always below 0dbfs so it should be fine to record at 100% then right ?

Thanks again

 

[bachatero]

So basically you’re saying that I shouldn’t really need to lower the volume at all because if I’m understanding correctly any clipping in the song (especially since they are all older before 1990 or so ) is likely to be analog and not sound bad in my audacity wasapi loopback recording

you don't need to lower the volume, but not because of the analog part. rather, any clipping in the original won't just go away when you lower the volume, because all volume does is scale the waveform's amplitude.

also, make sure you keep your recording volume as high as possible for maximum dynamic range and thus the best quality preservation.

check out these two screenshots I got of the same part of a song. the difference is that I decreased the volume of one of them. however, the parts with clipping still remain.

 

[Taupe1]

you don't need to lower the volume, but not because of the analog part. rather, any clipping in the original won't just go away when you lower the volume, because all volume does is scale the waveform's amplitude.

also, make sure you keep your recording volume as high as possible for maximum dynamic range and thus the best quality preservation.

check out these two screenshots I got of the same part of a song. the difference is that I decreased the volume of one of them. however, the parts with clipping still remain.

Got it so basically leave the volume at 100% in both app and windows mixer and record away.

More or less if the clipping is present in the track I can't make it go away and it won't cause any issue or clipping for my recording because it's already in the original and if it sounds good there or not noticeable it won't be an issue for me?

Ie people who recorded those songs I'm sampling if the clipping in those songs was cool with them I shouldn't worry on my end about it?

Whenever I check the files in audacity though interestingly enough even if there is clipping or not the "amplify" option shows the highest peak and it's always below 0db even if it's by .15 or whatever. I figure that's all that matters is that it's not clipping my recording ?

Recording the tracks with clipping at max volume which is very close to 0db if they sound good on the original won't cause me any issues later with clipping right?

Thanks for the help!

 

[bachatero]

think of it like this:

we can define clipping as any corner point on the waveform where the slope of the left is nonzero and zero to the right, or zero to the left and nonzero to the right.

or

basically, what that means is clipping is anywhere you've got somewhere curvy and somewhere flat at the same point.

and, what that means for us doing audio recording is that clipping doesn't necessarily depend on gain, but rather "just" what the waveform looks like.

let's put it all together with the original recording and your new sample.

when you record that sample at 100%, you won't be adding any more clipping of your own because that can only happen if the waveform goes above 100% by definition. and if you go lower, you aren't getting rid of any clipping from the sample because all normal audio recorders don't know about "corner points" and instead just scale the whole waveform as much as you want it to.

that's why it's 100% safe to stay at 100%!

Thanks for the help!

glad to help!

 

[Taupe1]

think of it like this:

we can define clipping as any corner point on the waveform where the slope of the left is nonzero and zero to the right, or zero to the left and nonzero to the right.

or

basically, what that means is clipping is anywhere you've got somewhere curvy and somewhere flat at the same point.

and, what that means for us doing audio recording is that clipping doesn't necessarily depend on gain, but rather "just" what the waveform looks like.

let's put it all together with the original recording and your new sample.

when you record that sample at 100%, you won't be adding any more clipping of your own because that can only happen if the waveform goes above 100% by definition. and if you go lower, you aren't getting rid of any clipping from the sample because all normal audio recorders don't know about "corner points" and instead just scale the whole waveform as much as you want it to.

that's why it's 100% safe to stay at 100%!


glad to help!

Awesome I really appreciate the info, side question, then I could theoretically never clip MY new recording of these originally recorded songs in audacity because they are set levels in the original that don't peak past 100% that I've noticed ,or even if they did go past 0dbfs originally they clipped on the original recording I'm streaming (and recording from) and could never clip on the new recording now ? Just so I have a greater understanding. The only way I could actually clip a recording while making it would be recording my own actual input while making a recording would be recording new input with a mic or instrument or amplifying the streaming source in audacity before recording?

Is there any books you could suggest to a beginner on audio science to help understand some do the math / applications.

 

[pablolie]

Hey there so I have a question regarding sampling off streaming services for making music.

I’m using qobuz 16/44 only tracks

Now if I want to record these at -6dbfs peaks I have to lower either the qobuz app or windows Master volume to avoid clipping and especially to get -6dbfs peaks. Mostly to 50-60% volume. From what I grasp digital volume reduction reduces bit depth and I’m wondering if it really matters because most of the time the samples would be mixed lower at -10dbfs or so but sometimes boosted a bit and I don’t want the noise floor or other artifacts to become and issue later.

For context it’s drum breaks for old boom bap style hip hop so fidelity isn’t super important but I’d still like to avoid messing the audio up more than I need to.

Here’s some options I thought of and would like to know which would be the preferred option ( in theory) all recorded with loopback in audacity

Record the 16/44 source with the volume reduced to reach -6dbfs peaks and deal with the “bit loss” and noise etc ( if it even becomes a problem with the above in mind)

Record the 16/44 source with volume lowered just enough to not clip minimizing the changes to the source and worry about gain reduction /boosting later in the DAW when mixing (since there won’t be as much of a change as the first option this may be better)

Record the 16/44 source at 24/44 with volume lowered enough to peak at -6dbfs (would those extra bits help?)

Record at 16/44 with volume lowered just enough to avoid clipping and use the “amplify” option in audacity with a negative gain value to set the peaks at -6dbfs

Sorry if this is long winded just want to get it right before I do a ton of sampling and have to do it over lol.

in most playback systems these days there will be more bits added to take care of the volume function. And you dont have to do anything to benefit from it.