Problem with different Sample Rates getting out of sync.

[Fidget]

I'm so frustrated I made this mistake, but during my band's recording session I set one device to record at 48000Hz and another at 44100Hz. I'd like to mix everything in 44100, and I'm mixing in Audition so I set up the project with that rate. When I brought in the 48000 file, it prompted me to convert the sample rate and I did.

It was a very long recording session, and I'm trying to chop up the 2-hour-long files into tracks. I have a loud clap at the beginning to sync them to each other, but when I do, by the end of the 2 hours they're completely out of sync. The sample rate conversion doesn't seem to be doing the trick. It's as if time is still moving faster in one of the recordings than the other. If I try my best to match up the recordings before each song (without the clap, which is quite difficult) it sounds better, but I'm still worried about getting weird slight echoes since, again, time is moving faster in one of the recordings.

Is there any way to remedy this problem? Thanks so much for reading.

 

[bouldersoundguy]

What hardware are you talking about?

One or both of the ADC clocks were inaccurate. Setting the both devices to the same sample rate without syncing their clocks may not have prevented the problem. Syncing their clocks, slaving one to the other or using an external clock for both, would have prevented the problem.

A very slight adjustment with a speed/pitch stretching tool might correct this. You will want to let the pitch be changed along with the speed. On the plugin I have I would uncheck the "preserve pitch" option.

 

[Fidget]

What hardware are you talking about?

Recorded on two separate devices, a Zoom R24 8-track and a Zoom H2 recording from binaural ears.

One or both of the ADC clocks were inaccurate. Setting the both devices to the same sample rate without syncing their clocks may not have prevented the problem. Syncing their clocks, slaving one to the other or using an external clock for both, would have prevented the problem.

A very slight adjustment with a speed/pitch stretching tool might correct this. You will want to let the pitch be changed along with the speed. On the plugin I have I would uncheck the "preserve pitch" option.

That must be it. Before you said this I had never heard of ADC clocks and I assumed it was the sample rate.

Just to be clear, you don't recommend preserving the pitch along with the sound change? I assume this is because the change in pitch would be negligible? What's the downside to preserving the pitch?

I'm hoping that by using the clap at the beginning of the 2-hour recording and a similar peak near the end I can change the speed by the correct ammount... Thanks for the advice so far, much appreciated.

 

[bouldersoundguy]

You don't want to preserve the pitch when correcting the speed because you want to exactly reverse the process that caused the play speed error, and that play speed error caused a slight pitch error. It's exactly analogous to adjusting the play speed of an analog tape deck.

Preserving pitch does something entirely different, chopping the audio into short chunks and recombining them. That's not what you want.

 

[bouldersoundguy]

Recorded on two separate devices, a Zoom R24 8-track and a Zoom H2 recording from binaural ears.

As far as I can tell you can only sync the clock of a Zoom R24 to another R24 or an R16, using the USB connection. The H2 has no sync capability.

 

[Fidget]

As far as I can tell you can only sync the clock of a Zoom R24 to another R24 or an R16, using the USB connection. The H2 has no sync capability.

This makes so much sense. Thank you so much for the help. I just hope I can correct the speed precisely enough that nothing is noticable.

So, for future reference, sample rate doesn't affect the recording speed? I'm hoping I won't run into any problems caused by the accidentally mismatched sample rates of 48000 and 44100 (just convert the one down to 44100).

 

[bouldersoundguy]

This makes so much sense. Thank you so much for the help. I just hope I can correct the speed precisely enough that nothing is noticable.

So, for future reference, sample rate doesn't affect the recording speed? I'm hoping I won't run into any problems caused by the accidentally mismatched sample rates of 48000 and 44100 (just convert the one down to 44100).

One second equals one second regardless of how many times the signal is sampled in that second. The problem wasn't the samples, it was each device's different measurement of one second.

It's entirely possible that a device's clock is right at one sample rate and wrong at another. The Alesis HD24 is notorious for being inaccurate at 44.1k and dead on at 48k. As long as it's played back through the same converters it doesn't matter because the converter's record error is exactly reversed on playback. But when you record through one set of converters and play back through another the error becomes a problem, as you've discovered. So it may be that setting both devices to 44.1k will prevent this in the future, but it's just as possible that the clocking mismatch will be the same either way.

You could try playing the recording from one device into the other via the analog connections. That "should" reverse the clocking problem and resample at the same time. Of course that will take two hours to do in real time.

 

[Fidget]

Yeah that actually sounds like a brilliant solution, but first I'm trying the time stretch feature in Adobe Audition. It's pretty confusing to figure out what the effect's settings mean (as in, whether "pitch shift" refers to the natural change in pitch due to the change in time, or to the unwanted pitch correction) but I believe I've got it figured out, and disabled the unwanted settings.

In the future I'll probably avoid this problem by sacrificing a stereo slot on the R24 and plug the binaural microphone directly into it... or just buy another R24 and sync them. I can't believe I never heard of ADC clocks until today. I just assumed time was time. I blame Einstein.

I really appreciate the advice, and just realized that you're in Boulder, which is a wild coincidence as that's where I am too & where this recording was made.

 

[bouldersoundguy]

Yeah that actually sounds like a brilliant solution, but first I'm trying the time stretch feature in Adobe Audition. It's pretty confusing to figure out what the effect's settings mean (as in, whether "pitch shift" refers to the natural change in pitch due to the change in time, or to the unwanted pitch correction) but I believe I've got it figured out, and disabled the unwanted settings.

Whatever changes pitch and time together is what you want.

In the future I'll probably avoid this problem by sacrificing a stereo slot on the R24 and plug the binaural microphone directly into it... or just buy another R24 and sync them. I can't believe I never heard of ADC clocks until today. I just assumed time was time. I blame Einstein.

Imagine what it was like with analog machines. There was pretty much no hope of syncing anything without using SMPTE.

I really appreciate the advice, and just realized that you're in Boulder, which is a wild coincidence as that's where I am too & where this recording was made.

Well that's funny. Moderator jimmys69 is down in Denver (a little closer, actually). We met up once in Boulder. We may have to plan a Denver/Boulder get together. Are you by any chance a fan of sushi?

 

[Fidget]

Well that's funny. Moderator jimmys69 is down in Denver (a little closer, actually). We met up once in Boulder. We may have to plan a Denver/Boulder get together. Are you by any chance a fan of sushi?

Absolutely. And I'd love the chance to talk sound with you guys. Side note, I stretched the time the wrong way and made the problem worse so now I have to wait another hour and a half while the process runs again. Can't believe I made such a dumb mistake.