MP3s for email collabs, DON'T DO IT!

[ohtravioso]

Hi everyone,

I did a little investigation that I wanted to share with everyone. When you convert a WAVE file to an MP3 file to email to your buddy for him to work on and send you back a track for you to mix in, there will be an unknown amount of latency added to beginning and end of the MP3 file. At least 528 samples, but it depends on both the encoder and decoder you're using. So, if you're working to a grid in your DAW, and you bounce out a track, convert to an MP3, and send it to your buddy. And your buddy sends you back an MP3 to drop into your DAW to keep working with, that file will not align to the grid. It will be late. Your buddy isn't as off time as he sounds. He just worked to your MP3 latent file and then sent you back an MP3 to further compound the problem.

So, you can either nudge that track around until you get it just right. Or, you can only work with sample accurate audio and save everyone the headache. I'm doing the latter from now on.

Just thought you should know.

Hayden

[mojobone]

When I do this sort of (increasingly common) thing, I may send an MP3 for reference, but I expect a WAV file back; I'd never mix with degraded MP3 audio without a dang good reason. (such as having no choice, and you'd have to pay me extra to do it, and I'd probably demand that my name not appear in the credits) I'd be adjusting the time-alignment of the resulting track in any case; it's jes' good housekeeping. I usually send a stereo WAV two-mix for monitoring purposes, unless my collaborator lacks a broadband connection, so kinda moot, in my case.

[llama]

Good advice Hayden. I may send send an MP3 to a collaborator just to say "hey what d'ya think of this" But for tracks that will be used in the final mix I'll send WAV files via box.net or yousendit. Even my crappy ears can hear the degradation of MP3 from WAV files. I don't care about that when listening to stuff on my iPod but not when I'm trying for the best sound possible on my stuff.

Derek.

[eeoo]

What about if you're converting to mp3 directly from your daw session as opposed to bouncing to wav and then using third party conversion?

[jazzstan]

I did a collaboration with an MP3 without problem. I brought the MP3 into a track in Sonar - which it converted it somehow. I added a measure of quarter-note clicks upfront and then did my guitar tracks, I sent the WAV files with click (of just my guitar parts) to my collaborator who never reported any problem syncing 'em up with his original tracks.

[chris carter]

There should not be any problem. What encoder are you using?

It's possible that you are using a really bad mp3 encoder/decoder. It's more likely that one or both of you is using a DAW that doesn't not have any delay compensation and you aren't dealing with it apropriately. If this is the case, you will have this problem regardless of whether you are using an mp3 or a wav file.

mp3s are fine for reference when tracking. In fact, they are used this way in professional situations all the time. Like mixing down an instrumental, sending an mp3 to load into a DAW, then tracking vocals with the mp3 as reference, and later swapping the mp3 for the original instrumental tracks. Everything lines up perfectly doing this.

But if you are using the file for produciton purposes, it is best to send consolidated audio files. Or, like with the example above, ultimately swap out the mp3 for the consolidated audio files when it comes time to mix. You can read about how to create consolidated audio files here: www.vonpimpenstein.com/mixguidelines.htm It's a guide for delivering tracks to a mix engineer, but it's the same and says how to consolidate audio files.

[orest]

Have you tried to send the file via YouSendit instead?
Create an account and you are able to send files up to 100 meg for free.

[Mark Kaufman]

There should not be any problem. What encoder are you using?

It's possible that you are using a really bad mp3 encoder/decoder. It's more likely that one or both of you is using a DAW that doesn't not have any delay compensation and you aren't dealing with it apropriately. If this is the case, you will have this problem regardless of whether you are using an mp3 or a wav file.

mp3s are fine for reference when tracking. In fact, they are used this way in professional situations all the time. Like mixing down an instrumental, sending an mp3 to load into a DAW, then tracking vocals with the mp3 as reference, and later swapping the mp3 for the original instrumental tracks. Everything lines up perfectly doing this.

But if you are using the file for produciton purposes, it is best to send consolidated audio files. Or, like with the example above, ultimately swap out the mp3 for the consolidated audio files when it comes time to mix. You can read about how to create consolidated audio files here: http://www.vonpimpenstein.com/mixguidelines.htm It's a guide for delivering tracks to a mix engineer, but it's the same and says how to consolidate audio files.

Glad to see Chris's take on this, because I keep hearing about this same problem, and yet never experiencing it.

I definitely use WAVs for tracking, but MP3s for reference tracks have never given me this timing problem. So far, that is...

[cardell]

Thanks Hayden...but...

...this is done successfully at kompoz.com (an on-line collaboration website) all the time. They use a sync tone and never have any problems lining up tracks from all over the world.

They have two videos on this page that describe there use in detail:

http://www.kompoz.com/compose-collabora ... rial.music

Stuart

P.S. I also like Flac & WavePack (both lossless compressions) as alternatives to MP3.
http://flac.sourceforge.net/
http://www.wavpack.com/

[ohtravioso]

Hey guys,

The sync tone approach is interesting.

@Chris Carter - I understand people are using MP3s for reference. But I've tried several different encoders and decoders, and I've found an indeterminate number of samples added to the beginning of the audio file. Sometimes as few as 528. Sometimes as much as over 2000. So if encoders behave differently, the decoder must be able to compensate for every encoder type to correctly compensate for the latency.

I honestly believe that people are using MP3 for reference, and they are adding at least 528 samples of latency between when they return to using WAV files.

If you have some free time, try this experiment. Make a wave file mix with zero silence, something that has the waveform start at the beginning. Pick any MP3 conversion method you want. iTunes or a LAME encoder. Anything at all. Now open both wave file and MP3 in an audio editor. Zoom in to the start of the waveform on both down to the sample level. You should see added latency.