Imported track is drifting

Hey all - got an interesting question…

Im recording multiple tracks from my keyboard and my guitar. I have David Bowie’s “Rebel Rebel” playing in my earbuds and i have my headphones over that. The headphones allow me to hear what i am recording.

For keyboard, the flow is: keyboard out → zoom h4n recorder. My headphones come out of the zoom.
For guitr, the flow is: guitar amp out → zoom h4n recorder. My headphones come out of the zoom.

For every track i record, i am playing aling WITH the origianl track so there is no possibility of drift, hiwever i have noticed that over time my guitar track starts to lag. It doesnt make any sense to me. Its recorded perfectly in time.

Any suggestions? Im a sound man in TV so i also have more high end recorders (Sound Devices 633) but i cant imagine that the zoom is not recording in time. It seems like an audacity issue to me.

It seems like ONLY the guitar track is drifting. The bass and drum tracks were both recorded from the keyboard to the zoom and they both stay perfectly in sync. The guitar track, which was also recorded with the zoom, is the only track the drifts. The track seems to be about .1 second off - very odd!

very odd!

Not that odd.

Any time you have separate sound, the two sounds are only as accurate as the original sampling rate generators. For example, if one recorder was running at 44100.000 and the other was at 44100.001, the two tracks are guaranteed to drift with respect to each other.

I can make this worse. One is at 44099.900 and the other is 44100.001. They’re both wrong. The movie people go to extremes to avoid this problem with expensive crystal-controlled time base generators, etc.

It’s a snap to correct. Effect > Change Speed. Once you figure out the error, it’s always the same. One recorder (for example) is running 0.003% fast. One correction no matter how long the recording is.

When you do a multi-point podcast, it is recommended that all parties record their own high quality voice and not use the Gargly Skype voice. Everybody ships their own voice sound file to the Producer who marries them into an excellent, perfect whole. Nobody is ever in sync.

And now you know what this does.
Screen Shot 2018-05-22 at 10.27.10.png
This is how they do picture and sound sync on a movie set.


Actually, you may be in this condition: Any time you play back a file under different conditions than it was recorded, you get a speed change. You get slow motion in the theater by speeding up the movie camera. So you may have three variables: The record speed of one recorder, the record speed of the other and the playback speed of the player. They could all be wrong. It’s up to the producer to wave their pencil and declare one of them to be the master.

Use Change Speed, not the other two. Change Speed assumes both the pitch and duration changed and it’s relatively harmless to change them back. The other two filters rip the sound apart and put it back together again. Some with better success than others.


If there are tiny pieces of audio being skipped, the recording will increasingly drift out-of-sync, (specifically lag-behind). …

Oh, right.

My comments are assuming there’s nothing broken. If your machine can’t keep up, the recording may skip tiny pieces of the show here and there. That will cause the presentation to play too fast.

Even worse, it may change with time, temperature, humidity, and phase of the moon making corrections a nightmare. That’s one you just have to fix.


I am a sound man in TV/Film so I know all about drift and Timecode sync boxes.

Here’s the thing: all tracks were recorded with the SAME recorder, so there shouldn’t be any discrepancies between them it’s not a huge deal as I just have a little snip here and a snip there to bring it back together but it’s odd. I suppose I could use my more expensive recorder. The zoom h4n is an inexpensive piece of trashola.

I have an H4. The fuzzy idea is the external preamps are trash but the built-in mics are OK. About a year ago I did a system update so it could use larger memory cards.

If the recording is too long, that’s not dropout or other shortcoming in recording. Those produce short recordings.

That could be bad Sampling Rate management? Is everybody running at 44100? Can you tell a percent error? Sometimes that can tell you things, such as falling into a common error. I still have nightmares of 30.00 and 29.976.

I’m out of ideas.


This doesn’t have anything to do with frame rate. I’m not syncing the audio to video. It’s odd.

There’s no 29.976 frame rate. There’s 29.97 and 23.976, but no 29.976.

Oh well!

Forgive me.


That was just the illustration of a “standard error.”

The difference between 44100 and 48000 would be another standard error.

It is odd that you can make a couple of minor corrections and have everything come out apparently even. That could be something is unstable.

Can you recreate the error?