Inconsistent Recording Latency (Win 8)

I’m using version 2.0.6 and I am having trouble with the latency correction. It is currently set at -130 which seems to be very accurate about half the time but the other half it is up to 13ms inaccurate. Is this a normal variation in latency? If not how can I correct this issue. This issue is present through any mic/input I use. I have tried uninstalling all plugins so it is just the default Audacity plugins, stopping all unnecessary programs that may be slowing down processing but the issue remains.

So there is a clear demarcation? Some percentage of the time the performance goes down perfectly correctly in time and the mating percentage it goes bat-squeak nuts? No, I don’t think that’s normal. Normal would be tiny “noise” variation or slight drifting.

Does it always go off the same direction?

You can do statistical analysis by taping an earbud to the microphone, load up a click track as the rhythm reference and start overdub multiple times.

This is what it looks like after successful latency adjustment. You will get a stack of these things and they may or may not all line up.

You do start to get very serious timing problems and drifts when your hard drive fills up. What’s your drive Size and Space Available in actual numbers?

You can get insane behavior if your drive fragmentation is too high.


You are multiple Audacity versions out of date. I don’t expect that to cause this problem. It could.


So here is what happens when I stack recorded click tracks and here is how much space I have. Could this be a ram issue? I only have 4GB

stack recorded click tracks

I’m looking for stacking six or eight of these samples. I want to get a feel for the deviation and I can’t do that with only two. Still, that’s only a .03 second error. That’s really tiny. See what multiple passes does. It’s possible you can get a cloud of errors and adjust for the middle of the cloud.

4GB RAM is plenty, particularly because you told us you essentially shut down everything else on the machine.

Do you have a Skype account and do you leave it sleeping rather than shut down? Skype considers your sound services as belonging to Skype and there are no options.


It’s still audible though and it has made recording very difficult. I’ll post about 6 or so hold on.

Here are the next 6, same project, default click track.
click 2.jpg

I should have zoomed in a little more on the 2nd pic but it’s the same project same spot at the 1 second mark.

Closer. If you still have that project: View > Fit Vertically. That should put all six of them in one picture. I need to see all six plus the original clean track and how they relate to each other. It seems there aren’t any that line up ever. Didn’t you say up the thread that you performed the alignment and only some of them wander off?

Did you do the manual defrag?


I didn’t do the defrag, I’ll do that now, and I’ll try to fit more in a picture for you if that doesn’t help.

Just to be a ray of sunshine: It’s possible your machine will never be stable enough for precision overdubbing. It’s not a Digital Audio Workstation.

There was a recent post where someone was performing a live Skype entertainment show but recording each individual city on their own machine. Each city ships their very high quality wav sound file to the central location for editing into the “real” show. The ratty Skype audio is not used. It’s only there for coordination.

Nobody is shocked that some of the locations had minor timing issues. That’s a piffle; that’s a simple fix in Audacity. It was a shock when one of the cities provided a sound file that wandered. Constantly changing. Never sat still. And that’s the end of that portion of the show. There is no cure for that in post production and the only final solution is change the recorder and shoot the show again.

Sound familiar?

In your case, it’s possible to use the desperation method. Do one pass of overdub and full stop. Fix it. Use the Time Shift Tool to push the beginning of the clip into alignment and then Effect > Change Speed to make the end line up. You haven’t complained about the end of the clips yet. It’s possible they’re moving independent of the beginning.

Set up for the second overdub and repeat.


I am surprised that you didn’t complaint about machine latency. It’s very unusual (never) to be able to hear yourself live in real time during an overdub session. I reviewed three ways of “perfect overdubbing” for the forum. Listen to yourself while you’re performing. In all cases I used a special piece of hardware outside the computer.

I’m not urging you to mix music on earbuds. They happened to be photographically convenient.


Here it is after the defrag. Seemed to help a bit.
click defrag.jpg

You may not get it a lot better than that. The worst one, the late one on the right, is 1.008. That’s 0.008 seconds off. Nobody gets a band or orchestra to start with that kind of accuracy.

I’m not sure where to go from here. I think your system is normal. One of the other elves will chime in.


How did you get around Machine Latency? Listen to yourself sing into a cupped hand?


Or about 350 samples off.

I think it is pretty normal but a little below par.

Have you turned Skype off, if you have it?

How fragmented was the drive you are recording to? If it was not very fragmented, then fragmentation is not the cause.

There are a few things you could try. Open Task Manger, click “More Details” if necessary, then the “Details” tab. Right-click over Audacity, “Set priority” and raise it to “Above Normal” or “High”. Accept the warning about system instability. That setting does not persist after Audacity is closed, but you can see if it helps.

Update to Audacity 2.1.2: and try Windows Direct Sound host and WASAPI host in Device Toolbar, instead of MME.

Upgrade to Windows 10. Microsoft claim a reduction in audio engine processing delay in Windows 10. If it’s relevant, I think you will have to choose Windows WASAPI host to see any benefit.


Thanks for all the help I’ll let you guys know if I find out what’s going on, although it was pretty accurate after the defrag. I can’t remember what percent it was at before. Also, I don’t have Skype so that’s not getting in the way.

How are you listening to the guide tracks and are you listening to yourself during a live pass?