Combining recordings from multiple DAWS

[ABECK]

I've done a handful of compilation cover tracks recently where each musician records their part using the original track as a guide. They send me their tracks and I mix in my DAW. I'm running into issues with drift, especially on the drums. They are locked at the beginning, but over time, they drift enough from the original that I need to do much editing to get them back in time.

 

What are some variables I should be looking at? Frame rate? Sample Rate? Clock Source?

Thank you!

[SamuelBLupowitz]

Sample rate is the big culprit here, in my experience. It's amazing the difference that will make in the length of the "same" recording over a period of time. I had to learn that the hard way myself.

 

Get everyone to agree on the lowest common denominator and you should be fine. I've yet to meet anyone who actually thinks they can "hear" sample rate, though I've heard legends... if that means you all have to record at 44.1, hey, it's still CD quality, and I would put money on none of your listeners being able to tell. And even on the off chance that somebody can, the difference it will make is so far beneath mic choice, mic placement, preamps, compressors, EQ... and it's all better than having to chop up a performance to make it fit.

[ABECK]

Good call. I'm thinking it's likely some were at 48, some at 44.

[harmonizer]

And ask for WAV files, not mp3.

[johnchop]

At different sample rates pitch will be off as well

[KuruPrionz]

And ask for WAV files, not mp3.

 

Yes, .WAV files of individual tracks - 24 bit @ 44.1. That way you won't lose any GarageBand users.

 

NO ProTools compiled multi-track WAV files, those are a PITA unless you own ProTools (I don't).

[mcgoo]

Most DAWs will convert sample rate to the host project's rate upon import, so I doubt that's the issue. What it sounds like to me is that not everyone is exporting their files to you at the EXACT same start point. Give everyone the same reference / guide track to work with and tell everyone to export the files they deliver to you from the exact same start point as the guide track.... and make sure that same guide track is in your project when you start assembling tracks.

[KuruPrionz]

Most DAWs will convert sample rate to the host project's rate upon import, so I doubt that's the issue. What it sounds like to me is that not everyone is exporting their files to you at the EXACT same start point. Give everyone the same reference / guide track to work with and tell everyone to export the files they deliver to you from the exact same start point as the guide track.... and make sure that same guide track is in your project when you start assembling tracks.

 

The OP is describing a situation where the tracks start out fine and then slowly drift apart. That doesn't indicate different start times. It does indicate different total times for individual tracks.

 

Different sample rates is a possibility that should be explored and eliminated as a cause.

 

The DAW converting different sample rates to the rate in the session could be the problem, maybe another DAW does it more accurately. I only have questions, not much in the way of answers.

It is an interesting puzzle and could happen to many of us so a good one to try and solve. Cheers, Kuru

[ABECK]

The next project I'm requesting consistent sample and frame rates, as well as no MP3. I'll report back!

[ksoper]

...They drift enough from the original that I need to do much editing to get them back in time.

 

 

Are you doing the fixing by hand or using something like Sonar's Audio Snap--a quantization tool?

[mcgoo]

Most DAWs will convert sample rate to the host project's rate upon import, so I doubt that's the issue. What it sounds like to me is that not everyone is exporting their files to you at the EXACT same start point. Give everyone the same reference / guide track to work with and tell everyone to export the files they deliver to you from the exact same start point as the guide track.... and make sure that same guide track is in your project when you start assembling tracks.

 

Different sample rates is a possibility that should be explored and eliminated as a cause.

 

If the sample rates are mismatched it will be glaringly, immediately obvious. Wrong speed, wrong pitch... and I can't imagine sample conversion by any DAW being flawed enough to change the duration of a file. If multiple people exported audio at arbitrary start points, but close to the guide track's start, it may take a bit before the lack of tightness is noticed. Another possibility is if someone's DAW had inadvertently enabled some type of "audio follows project tempo" command. Even if another user's project tempo is correct, if it's altering the guide track's tempo, all bets are off.

[ElmerJFudd]

A few thoughts.

What daw are you combining tracks into and what sample/bit rate and tempo do you intend to run that session at?

 

Ask all contributors to do the same and go with WAVs.

 

Your daw I am certain is capable of SRC and can fly in anything but you can avoid having to figure out what everyone did and converting by telling them up front.

 

Also, have them send you their stems all bouncing tracks from bar 1 beat 1 so dropping them in your project is as easy as placing the audio file at the beginning. Unless you intend to use a standard format like OMF or BWF to place the parts where they belong on import.

[WestHarp]

All A/D-converters in audio interfaces use crystal clocks to determine the sample rates. But: due to tolerances, these are not always in sync. 44,1k is not always EXACTLY 44,1k. Therefore, it's possible that one of your friend's DAW recorded e.g. 44007 samples per second and someone else's DAW 44201. Your DAW uses the same sample rate for playback of all files - therefore all tracks are slightly different in length. That's why there's drift.

 

And that's the reason why "master clock generators" exist to sync different digital devices to the exact same sample rate...

[Konnector]

Does your DAW allow you to stretch/shrink the length of a wav easily just by grabbing one end and adjusting the length? If it does, you could do that manually and line them up to sync to your click reference. You can always split the clips into shorter clips if long clips are more of a challenge to sync up.

[ElmerJFudd]

Does your DAW allow you to stretch/shrink the length of a wav easily just by grabbing one end and adjusting the length? If it does, you could do that manually and line them up to sync to your click reference. You can always split the clips into shorter clips if long clips are more of a challenge to sync up.

 

If imported and converted correctly - meaning the sample rate conversion was accurate so pitch was maintained - the length is going to be very close. Most DAWs today actually have very good sample rate conversion, it"s not a perfect process and there still exist some high end software that does it best. But today"s DAWs do a pretty darn good job within audible spectrum.

 

Most DAWs have time stretching algorithms so, yes, you can adjust with it, just keeping in mind you have to be happy with the results - the process always creates artifacts.

[harmonizer]

..... Frame rate? ......

 

Just noticed the words "frame rate".

Are you syncing the audio mix with video, and was that video captured from an iPhone?

Some iPhones drop frames as needed so they can keep up with the data capture

but the the video software you are using may not realize it, which can lead to drift, if you are syncing audio with a video.

[dazzjazz]

I"ve just been in the same situation, both with professionals and my uni students. It can be a nightmare.

 

Beyond the sample rate issue, some DAWs, like the latest version of Cubase, will process the imported track relative to the grid tempo in an attempt to quantise it. This issue cost me about 6 hours recently. So check your processing on import settings and turn it off if it"s on.

[ElmerJFudd]

I"ve just been in the same situation, both with professionals and my uni students. It can be a nightmare.

 

Beyond the sample rate issue, some DAWs, like the latest version of Cubase, will process the imported track relative to the grid tempo in an attempt to quantise it. This issue cost me about 6 hours recently. So check your processing on import settings and turn it off if it"s on.

 

Ha! Current version of Logic also looks in the audio file for tempo information. You have to tell it to ignore it. Live and learn!

[dazzjazz]

Another thought: even if your not on the grid, get everyone involved to set the grid to the same tempo. This might help, though I haven"t tried it.

[Reezekeys]

I"ve just been in the same situation, both with professionals and my uni students. It can be a nightmare.

 

Beyond the sample rate issue, some DAWs, like the latest version of Cubase, will process the imported track relative to the grid tempo in an attempt to quantise it. This issue cost me about 6 hours recently. So check your processing on import settings and turn it off if it"s on.

 

This bit me big time a few years ago when I sent a track to a guy using a Windows DAW (I'm on a Mac). It was a .WAV but what I didn't know was that .WAVs can contain "tempo" information â which apparently was in my file (probably my DAWs default 120bpm). I get a call from the guy saying my track sounds fine played on its own in Window Media Player, but sounded "warbly" when he imported it into Sonar.

[Zalman Stern]

Drummer's audio interface may be off enough in clocking to matter. Typically would be considered bad hardware, but it does happen.

 

Drummer may be playing back to the guide track running at a slightly different speed as well, which could be due to something in the mechanism used to do playback. (E.g. inadvertent use of a slowdowner, the unexpected time quantization in some DAWs, etc.)

 

You could check the clocking by recording a known source on the various folks' system and see how far apart they are. Normally you'd put 'em all in the same room to do this, but you could either play a song into the interface from known good sources or possibly use decent metronomes. This is only going to show up large errors, not tiny differences.

 

Sample rate difference are more likely to affect things due to clock error in the audio interface than due to resampling being off by that much. There is no class of product in the world where there isn't some cost reduced badly implemented version that is buggy in some unimaginable way, but most quality DAWs should be able to do sample rate conversion well enough for this task.

 

-Z-

[elif]

I've done a handful of compilation cover tracks recently where each musician records their part using the original track as a guide. They send me their tracks and I mix in my DAW. I'm running into issues with drift, especially on the drums. They are locked at the beginning, but over time, they drift enough from the original that I need to do much editing to get them back in time.

 

What are some variables I should be looking at? Frame rate? Sample Rate? Clock Source?

Thank you!

This could be an entire topic/thread, and probably is somewhere. This is how I intend to handle this for a project. Our drummer uses Pro Tools. I use Reaper. All critiques are welcome.

 

Create the source to a click track and send the click track on a separate track along with the source. Then the respondent would re-record the click track to its own track along with any tracks that were added to the original source. This would work with respondents that have either a DAW to record multiple channels (drums, stereo source) or only a 2-track recorder (horns, vox, etc.).

 

Once the returned track or tracks are combined in the DAW (including the re-recorded click), regardless of the sample rate or even if it was an mp3 source, you should be able to stretch them from first click to last to synchronize them with the existing tracks.

 

It is possible to add a click track to an existing recording. However, if the recording was not made to a click track, it will become quite evident how much the tempo varies over its duration. Adding the click to such a recording can be very tedious because the source must be adjusted to the click.

[ABECK]

I"ve just been in the same situation, both with professionals and my uni students. It can be a nightmare.

 

Beyond the sample rate issue, some DAWs, like the latest version of Cubase, will process the imported track relative to the grid tempo in an attempt to quantise it. This issue cost me about 6 hours recently. So check your processing on import settings and turn it off if it"s on.

 

Good call - I'll check that.

Stretching won't really work, because at the beginning everything is locked. So uniformly stretching will cause issues too. My current remedy is cut and shift. Depending on the track this can be a few times or many.

[Theo Verelst]

On Linux, the long existing Rosegarden sequencer/audio recording software appears to record real time clock time stamps while recording multi-track wav files. Audio clocks are usually based on a crystal, like a digital watch, that's reasonably stable and at well fixed frequency, but just like a watch will drift a little over a day time, it might be different audio cards at different times (and certain even the same audio card at a different temperature) will have a small change in frequency, which might make for a few samples more or less on a song. The real time clock in Linux usually is kept synchronized with some internet time reference, and will have less excursions from the "real time" and so can better serve as an actual time reference.

 

A phasing in high frequencies because of a few samples being added or lost is certainly audible if the tracks contain similar signal.

 

Qbase as an example at some point guaranteed "sample accurate processing" which would in practice should mean that if you import *the same* fines into a different cubase and different machine, the playback should consist of exactly the same mixed sample, so the output file should be exactly, bit for bit the same. Some DAW software allows you to adjust time, even real time to different audio cards on the same machine, which certainly can give you high hat phasing errors if you'd record the same sequence twice and add the results. If you have decent audio card clock regime (like UAD2 (IIRC) on some usb mixers) payback of tracks with reclocking "off" should lead to sample for sample the exact same audio every time you play the same sequence.

 

Re-sampling of any kind by multi track software isn't a great idea and always creates signal degradation.

 

T.V.