Serious drift problem during podcast recording

I regularly record tabletop rpg sessions for my podcast. Recently I’ve started asking all the participants to record their own voice during the game. And I meanwhile also record the entire Google Hangouts call so I can use that later as a reference to sync up all the separate audio recordings. And while this has increased the audio quality of our podcast episodes considerably, there’s a serious problem I keep running into.

Audio drift.

While most of the audio tracks have a small drift problem that can be easily solved with adding a few seconds of silence in here and there to sync them up again, there is one person in the group who has an unreasonable amount of drift in his recording. It is not uncommon that I have to spend almost an entire day fixing the desyncing issues with his recording and I am tired of it. The amount of drift also doesn’t make any sense. Sometimes it’s just a few microseconds and a bit later it’s over a minute long and later still it’s back down to several second.
It’s driving me insane!

Just see for yourself. I took a look at his recording and compared it to my recording of the google hangouts session.
On the left you see what time it is during the recording and on the right you see how much time the main recording needs to catch up with his recording. (Yes, his audio is always early).


This data doesn’t make any sense to me…
Sometimes it stays constant and other times it’s shifting wildly.

First I thought there was a problem with his computer’s internal clock or something, that it was running fast, but now that I see these numbers I don’t think that’s the case.
We’ve also made certain we’re all recording at the same frequency, 44100 Hz, so that can’t be the problem either.

Can anyone please offer me any advice on what could possibly be causing this problem and how we can prevent it in the future?
There are usually around 6 people in the call and he’s the only one with this issue.

I’m at my wit’s end. Please help!

The audio interface he’s using, has a bad clock. The only sensible solution is to change the interface, imho.

In rare cases, with onboard audio on overclocked PC’s with real cheap onboard audio, the fix is not to overclock. This kind of hardware has no built-in clock generator and gets it’s clock from one of the main clock generators. But that solution is probably not acceptable to the user :laughing:

I’d get him a small, affordable external interface and see if that helps.

Oh no, that sounds like it will cost money…
Thanks for the advice, but is there a possibility that something else could be the problem?
And yes, I don’t know that much about computer hardware, just want to get that out there.

There’s always a possibility. I don’t see how, but maybe someone else can come up with one?

Split recordings are recommended because as you found, they can sound tons better than trying to clean up a Skype or Chat session. The down side is that drift thing. If you have a long show, chances of everybody’s computer running at exactly the same speed is zero. Usually, you can figure out a correction for each one.

Fort Lauderdale is consistently 0.03% slow, so I’ll correct for that with Effect > Change Speed.

Schenectady, on the other hand was 0.1% slow the first half-hour and then started drifting faster over the course of the next hour.

Schenectady has a broken microphone or Microphone Preamplifier. Or, more accurately, using devices not up to multi-player standards.

Usually you can do OK with home electronics. Everybody is the same percent off and it’s easily corrected, but nobody is writing big checks for accurate digital generation if they don’t need to. Every so often, somebody gets a really poor microphone or MicPre or USB interface and the digital bitstream just sucks. They don’t notice it until they start comparing it to something else. Like you.

How are they digitizing their voice? In detail. Model numbers.

You can’t fix what you have. That will take you a week to get one show out the door.

Side Note: This is exactly the problem people have when they demand to record two USB microphones at the same time. My machine asks me which one I want to use as a master, because it can’t use both. The other will slowly drift out of time.


Unfortunately changing the speed doesn’t help in this case (I’ve tried) because the drift is never consistent.
I’ll try talking to him about getting an external audio device, and if need be I’ll cover part of the costs, because I’d rather spend some money than to spend more hours on fixing his drift problem. After all time = money.

What are they doing now? It’s remotely possible they have a clear bottleneck that can be fixed.

There are technical minimums to doing some of these jobs, and no, you can’t fix everything in software. You got a significant boost when you found how well this technique can work with everybody else in the group. It’s not much of a mystery what needs to be fixed.

This musical piece was done with the “shipping files back and forth” technique. It looks like a Skype recording.


So if I am understanding this correctly he probably needs an external USB sound card.
Is that correct?
(Again, I don’t know much about computer hardware)

That is correct.

Alright, thank you for the advice!
I’ll make certain he gets one!

external USB sound card.

That may be a little too simplified. A USB soundcard doesn’t have a microphone. It would be really good to know the system as it is now used. If the performer is using a bad USB microphone, a new soundcard isn’t going to do any good.