Recording vocals using 24-bit card, how to export from Live?

[dafeda]

Hi,

I'm recording vocals using an Apollo Twin card which is a 24-bit/192kHz card.
Say I do some editing to the vocals in Live and want to export a single track to wav.
If I pick 32-bit/44.1kHz in the export options and import the wav back into Live
it sounds thinner and has a lower gain than the original track.
However, if I export using 24-bit/44.1kHz it sounds ok to my ears.

Could someone please explain the "proper" way to export audio in this specific case?
Also, do I need to use dithering when exporting audio to 24-bit that was recorded using
a 24-bit audio interface?

Thanks.

[Tarekith]

When you import the wav back into Live, make sure you turn off warping on that audio file, you don't need it anymore. Not sure why a 32bit file would be quieter than a 24bit one though, never heard of that myself.

You only need to dither when going from a higher to lower bit depth, IE from 32bit to 24, or 24 bit to 16bit. If you're exporting at the same bit depth as you recorded, then I would not add dither yet.

[dafeda]

Thank you for replying.
Can't remember whether I had Warp turned on, will check after work and post my results.
I read somewhere that Ableton internally processes all audio as 32-bit and that I should use
dithering even if I record at 24-bit. Does this make any sense?

Thanks.

[dafeda]

I might be an idiot but please hear me out.
I record a vocal track, and then pan it completely to the left in Ableton, i.e. at 50L.
The peak of a section of this track is -19.77 and the RMS is -26.77.
If I now export this as either 32bit or 24bit at 44100Hz and import it back in I get
PEAK=-22.78 and RMS=-29.78. I get the same drop whether I convert to mono or not during exporting.
However, if I do not pan my original track and export I retain the same PEAK and RMS.
This is probably the way it should be but I´m a bit slow, and would greatly appreciate it if someone could explain.

Thank!

[jlgrimes]

Hi,

I'm recording vocals using an Apollo Twin card which is a 24-bit/192kHz card.
Say I do some editing to the vocals in Live and want to export a single track to wav.
If I pick 32-bit/44.1kHz in the export options and import the wav back into Live
it sounds thinner and has a lower gain than the original track.
However, if I export using 24-bit/44.1kHz it sounds ok to my ears.

Could someone please explain the "proper" way to export audio in this specific case?
Also, do I need to use dithering when exporting audio to 24-bit that was recorded using
a 24-bit audio interface?

Thanks.

There are differing opinions on this.

24 bit 44.1 kHz with no dither is pretty standard and shouldn't be an issue. This should also give you exactly what you recorded uncolered in any way.
32 bit 44.1 kHz with no dither though would be better if you applied any processing in the DAW after you recorded.

That said there really shouldn't be any perceivable differences between the two, so it is more preference than anything.

Another advantage 32 bit has is that it doesn't clip (practically speaking), so it can be ideal if sending stems of bounced VSTis or processed files, but you still shouldn't clip a vocal recording while recording because it is still 24 bit at that point.

[dafeda]

Hi, thanks for the info.
I still don't understand why peak and rms are lower than the original though...

[Stromkraft]

Hi, thanks for the info.
I still don't understand why peak and rms are lower than the original though...

What about your export settings in detail? I assume no normalize was active?

[Tarekith]

Panning the signal to the left and exporting is of course going to be different from having the right channel in there as well, they are two different signals.

[Stromkraft]

Could someone please explain the "proper" way to export audio in this specific case?
Also, do I need to use dithering when exporting audio to 24-bit that was recorded using a 24-bit audio interface?

When you use a mono mike connected to and recorded from one channel you'd get a mono file in Live. But you have a stereo file, indicating that you recorded for some unknown reason a stereo input. What went into those 2 channels?

If you're going to make any further processing digitally there's hardly any need of dithering. I'm not as certain concerning analog processing though as I have never done that from a digital source or if I did I've forgotten.

[dafeda]

When I record with my mic I do get a mono file. When I then pan it hard left or right, export and import it back it, the new file has a lower gain than the original. I do not understand why exporting a panned mono file should affect its volume. Thanks for your time.

[H20nly]

maybe because you're effectively cutting the signal in half when you hard pan??

I can only offer that guess and my interest in the topic though... *bookmark*

[Stromkraft]

When I record with my mic I do get a mono file. When I then pan it hard left or right, export and import it back it, the new file has a lower gain than the original. I do not understand why exporting a panned mono file should affect its volume.

I don't understand why you are panning a mono file before export in the first place, but it's also unclear what you mean with "the original" as what you compare with. You mean you compared to the original panned exactly the same way? Or how?

[Stromkraft]

maybe because you're effectively cutting the signal in half when you hard pan??

I can only offer that guess and my interest in the topic though...

I did a quick test with one my own vocal tracks, any mono source will work though.

The meters are the uninteresting part here.
This will cancel out with Utility: Phase Invert:

1. Export not normalized, panned hard to any side with the fader at 0 (AFAIK the fader is made part of the export volume-wise)
2. Import to a new track fader at 0. Do not pan. If you listen you see there is only info in one channel.
3. Now add Utility: Phase Invert to one track and play back.

The tracks will cancel out to full silence. If you move the pan they will not as that affects volume.

What I find a little freaky though and is what is the interesting part here is that what you have created with a panned export is a panned mono file with only information in one side. Something I didn't hold as possible. I'd expect a mono file with an affected volume, not pan information. Obviously I missed something about digital mono files and they contain pan information. This is reflected also in other apps. I tried in OcenAudio.

That's cool info to me. There is no problem in Live though.

[Stromkraft]

When I record with my mic I do get a mono file. When I then pan it hard left or right, export and import it back it, the new file has a lower gain than the original. I do not understand why exporting a panned mono file should affect its volume.

You have to acknowledge the fact that what Live offers first of all is balance panning, that is you are raising/lowering the volume of the left and right channels. Already there you are affecting volume. As you see in my little experiment above, the files if compared right will sound identical, which is proven by them canceling out with reversed polarity on one.

In contrast, what Live doesn't offer natively, though there are free Max For Live tools that do, is true stereo panning. But that's possibly out of scope here.

[dafeda]

I don't understand why you are panning a mono file before export in the first place, but it's also unclear what you mean with "the original" as what you compare with. You mean you compared to the original panned exactly the same way? Or how?

I was recording a bunch of vocals and panned some tracks hard left and hard right to get a feel for how it would sound. I then wanted to make stems so that I could import them into another project. I did not think it was necessary to center all the tracks before exporting.
What I am refering to as The Original is the recording I did in the first place. I then pan this original recording and export it. When I import the exported file into the project I see a drop in volume compared to the original. Your next comment is probably the reason why this is so.

In contrast, what Live doesn't offer natively, though there are free Max For Live tools that do, is true stereo panning. But that's possibly out of scope here.

This is probably the reason for my confustion. So if I record a mono track and pan it hard it will reduce in volume? That would not happen with true stereo panning right? Could you recommend a decent max for live device that does true stereo panning?

1. Export not normalized, panned hard to any side with the fader at 0 (AFAIK the fader is made part of the export volume-wise)
2. Import to a new track fader at 0. Do not pan. If you listen you see there is only info in one channel.
3. Now add Utility: Phase Invert to one track and play back.

Thank you for taking the time to perform this test!
The way I read this is that if I, say, record a bunch of vocals and do a rough mix with panning etc. I should probably use true stereo panning instead of the default panning in Live if I want to export stems for use in another project. Does this make sense or am I still misunderstanding something?

Thanks!