Two separate live recordings of same show, how to sync?

[muze]

So I used my Zoom H4 to record a recent gig and a friend also used a portable digital recorder to capture the performance. He FTPed me the files and I imported both recordings into Sonar 7. Both are stereo, his was 16 bit, mine 24.

They play fine but they don't stay in sync with each other. I tried stretching the audio by CTRL dragging the edge of one until everything was in sync, but then the stretched clip sounds.... stretched (is this jitter?) I downsampled mine to 16 to see if that made a difference and it doesn't. It's just separate equipment, separate clocks I guess.

I have word clock inputs available, but no idea if they would help, or how. I'm hoping maybe there's a solution here that doesn't require stretching. Maybe there's software that specializes in this skill?

Any suggestions?

 

[michaelst]

how about using your ears?
Convert to the same sample rate and depth, and make them mono.
Import both tracks into your software.
Pan one hard left, the other hard right.
Find the first drummer count off. Hopefully he clicks his sticks.
Move one until the clicks come from the middle of the stereo field.

voila.

 

[muze]

Hmmm I don't think you understand. I can line them up no problem but then they drift out of sync from each other later. It happens over the course of one song. The two audio tracks are slightly different speeds from each other....

 

[michaelst]

Ahhh...

OK.. well.. you're SOL then.

Sorry

 

[NeveSSL]

You could possibly spend some time chopping them up and adjusting the time of different segments that don't match, but you're going to have to spend A LOT of time and use your choice of time stretching plug-in.

IE, its probably not worth it.

Brandon

 

[dgatwood]

Hmmm I don't think you understand. I can line them up no problem but then they drift out of sync from each other later. It happens over the course of one song. The two audio tracks are slightly different speeds from each other....

The problem is not really the time. The problem is that the sample rates are slightly different. I'm assuming that we're not talking about radically different pitch like you'd get if one is 44.1kHz and the other is 48kHz. If it is, just do a sample rate conversion to make them the same rate and you're done.

Otherwise, you'll have to do some math to fix the problem. First, select all the samples from an easily identifiable drum hit near the beginning to an easily identifiable drum hit near the end. Copy this into your clipboard, then paste it at the start of a track. This will get you a really exact period of time for calculations. Repeat this for the other copy of the track. For each of these shortened clips, count the number of seconds and the number of additional samples beyond the last whole second.

Take the longer of the two shortened clips and call that one "correct". Now take a calculator (or use Google since this will overflow most calculators). Multiply the larger number of samples by the sample rate in Hz (e.g. 44100). Divide that value by the smaller number of samples.

Download Audacity. Import the complete shorter track. Set the sample rate to the value you calculated. Next, tell it to do a sample rate conversion to 44.100 kHz (or whatever the rate of the other track is). Use the resulting file instead of the original for that track and you should be in the neighborhood.

 

[muze]

Otherwise, you'll have to do some math to fix the problem. First, select all the samples from an easily identifiable drum hit near the beginning to an easily identifiable drum hit near the end. Copy this into your clipboard, then paste it at the start of a track. This will get you a really exact period of time for calculations. Repeat this for the other copy of the track. For each of these shortened clips, count the number of seconds and the number of additional samples beyond the last whole second.

Take the longer of the two shortened clips and call that one "correct". Now take a calculator (or use Google since this will overflow most calculators). Multiply the larger number of samples by the sample rate in Hz (e.g. 44100). Divide that value by the smaller number of samples.

Download Audacity. Import the complete shorter track. Set the sample rate to the value you calculated. Next, tell it to do a sample rate conversion to 44.100 kHz (or whatever the rate of the other track is). Use the resulting file instead of the original for that track and you should be in the neighborhood.

So are you saying that (for example) one is 44.104 and the other is 44.099 or something? Shouldn't an audio editor be able to figure that out and change to 44.100 ?

 

[apl]

So are you saying that (for example) one is 44.104 and the other is 44.099 or something?

Exactly. That's why when people use multiple interfaces they have to pick a master clock so that all the samples happen at the same time.

How much out of whack are they? A couple of seconds for a four minute song?

Shouldn't an audio editor be able to figure that out and change to 44.100 ?

The wave file is just a list of numbers and it's nominal sample rate is assumed to be correct. Both interfaces thought they were at exactly 44.10000000 but obviously that's not possible. There's nothing in the data that will give the actual sample rate.

Has a similar problem when I had my drum machine on 120BPM and the click track in the editor on 120BPM and they didn't come out the same, either.

 

[muze]

So I finally got this to work with very acceptable results. Thanks to dgatwood and apl for your suggestions.

Initially I tried the formula given but found that my shorter file was actually doing the opposite of what I needed. The number I was getting was 44,119.78 which was rounded up to 44,120 by every program I tried ( Sound Forge, Wavelab, and Audacity). Anyways I found that I had to enter 44,080 (20 less than 44,100 instead of 20 over) and then resample to 44,100 to get the correct results.

And it wasn't perfect towards the end of each track, probably because I couldn't include the decimal. For the hell of it I chopped and more accurately lined up the second half of each clip.