USB Mic Latency

I have a big problem. I use a Rode USB mic to record my voice and a separate camera to record me talking. When I put it all together in Adobe Premiere I noticed… a delay… ~0-2 seconds after 10 minutes (constant, starting at around 2 minutes).

It is a huge issue for voice over. I googled around and it is a well-known issue for usb mics…

Is there ANYTHING I could do? How can I fix it in Audacity if it is a constant delay, probably something like 0,00x seconds / 1 minute :frowning:

What I’ve tried:

  • tested on 2 PCs
  • I even tested a webcam and it produced a slight delay as well comparing to the Lumia video file

Effect > Change Speed. Once you determine how far apart they are, it will always be the same correction.

Assuming you used WAV or one of the other uncompressed formats, it should be possible to import the work, corrected it and export it again. Just like that. Dump it in your video editor and go. If you insisted on MP3, it’s going to be messy. Audacity doesn’t edit MP3. It makes a personal copy of the work and then makes a new MP3 when you’re done, picking up compression sound damage along the way. You can’t stop it.

Anyway, that’s how to do it. And yes it’s perfectly normal to have drift like that. You have two carefully timed digital generators. Unfortunately, they’re not timed to each other.

If you start using a different camera or audio system, then you’ll need to find the correction again.

There’s a thing about 44100 versus 48000 sample rates. That’s Audio CD versus Video. Most video editors I ever met will deal with either one with no fuss, but 48000 is the video sample rate if you have to pick one.

Describe how you’re shooting the work. Remember, this is a forum, not a help desk.


Thanks, I use a Panasonic standalone HD camera (saving it as 1080p mp4) and a USB Rode Mic for better audio with Audacity on my PC (wav). Of course I save and edit it all as WAV.
I was really shocked to discover it can gradually desync like this… I know I can edit it in Audacity but I have like 100 clips, 10 minutes each to record… it will be a nightmare as I am not sure if it desyncs in the same way every time…

Is there like a ‘buffer’ option in Audacity to improve it or it is not related to Audacity?

Thanks for you help!

I am not sure if it desyncs in the same way every time…

It should. Each timebase has to be stable within itself. If it’s not, it will produce garbage instead of a digital sound signal. They can’t drift, warble, or wander, but they can be any sample rate within their manufacturing tolerances.

The worst thing that can happen is you catch an equipment combination that happens to match. You go all day thinking you found the holy grail of youth and that lasts until the first time you have to change equipment. “What on earth happened to my sound tracks?” (not in those words).

I get it. You’re recording in the sound track in Audacity. Unsurprisingly, you want us to correct it in real time. Audacity does very little in real time and effects is not in that list.

Pick three monologs at random and correct them. Keep track of what you did. I bet they all come in within a teensy fraction of each other. Nobody can see one frame off lip sync and many people can’t see two or three. So right there you can miss it by 1/24 of a second and nobody will know (assuming 24-frames).

You don’t even have to struggle with existing shoots. Start a shoot with a slate. “Camera Mark!” [bang]. then just let it go for an hour and shoot an end slate (hold it upside down) “End Mark!” [bang]. That will give you a frame accurate, literally, shoot to derive your correction numbers.

I bet you thought that slate thing was just for Hollywood.

Back to the microphone. Which Rhode and how are you speaking into it? Driving a microphone is my thing.

Screen Shot 2015-09-27 at 6.34.59 PM.png

Thx. It is Rode-NT USB mic

I guess I missed one. The Time Bases are established inside the camera and USB microphone. You’re stuck.

That’s why the grownups spend the bux to get good timebases. You can start a movie camera and a sound recorder and come back in an hour and they’re still in nearly perfect lock step. You can lock them, too, but that’s harder when you’re in the back of a truck shooting a western at full gallop.

“Where’s the sound guy?”

“Ummmm. He was in a Jeep behind us…”


From your experience: is it due to the USB? I am thinking of getting an XLR mic and recorder for the next project. It is expensive :frowning: Will it work better than USB mics?

I am trying to complete a pro project using amateur equipment :frowning:


Thx. It is Rode-NT USB mic

Thx, but I mean how are you using it? Are you on camera and the microphone is suspended over your head (18" up and 18" forward).


No, the camera is 1 meter from me, I am at the desk with the mic 20cm from my face

I am trying to complete a pro project using amateur equipment

Certainly nobody else is doing that. You only get your picture in the magazines when you get it to work, like that movie shoot they did all on iPhones. It helps that a world-famous director did it.

Now you want me to pull the magic out of my hat. The big kids still use slates for exactly this problem. Given they may not always use end slates.

I’m not sure there is any way to know who’s off, the camera or the microphone. They could both be off.

Personally, if you got the sound perfect in every way (I can’t stress enough how important that is), I would continue to shoot this way and build the correction into your production pathway. I think you can create a Chain to do this. That’s Audacity for “Batch.”

“Correct all those sound files. Call me when you’re finished. I’m going to Starbucks”


Shoot the hour-long test show to see how far off they are—accurately. Then batch correct your brains out. I bet it works. You don’t need the clapboard. You can clap your hands on camera.

Thx, will do

Make safety backups of all your shoots.


To measure it, you need an extremely stable reference clock.

In this case, the clocks are probably both off frequency, but both in a different direction. An sich, that’s relatively easy to correct, but usually the drift also changes in time with the temperature.

You might start a recording with one clock that is -0.5 % and the other at +0.2%. That’s easy enough to correct. While recording, the first clock glides to -0.3% and the other to +0.3%. And that’s no longer easy to correct.

A third clock affecting this, is the internal clock of your computer. That one is not affected by USB transport fi, and in general is a lot more stable. Some equipment has hardware that will auto-align with the internal clock, but that’s not always possible.

A simple way to minimise the error, is recording very short cuts. The error is at minimum when you start recording and could slide to worse or better. When the clip is short, it’s not noticeable.

The problem with game recording from a console is that you probably want to record longer games. Some consoles have weird graphics output that might just worsen the clock drift on the recording device by adding some frequencies that are not visible on the analog composite or VGA outputs if you attach a TV or an oldskool monitor. They tend to confuse video digitizers some times.

Which digitizer and software are you using?

Which digitizer and software are you using?

I’m not sure the problem is that deep. It’s drift between a USB microphone and a digital camera. Ordinary Separate Sound. They’re both at room temperature through the performances.

There is one very serious possibility past simple manufacturing tolerances. The computer is dropping bits from the USB microphone. People post with those problems pretty regularly. It usually shows up by popping or clicking but it doesn’t have to. That’s erratic and pretty deadly. That may not respond to simple speed corrections.

If three or more takes correct to normal with simple speed correction, and I think they will, then this show should correct with a simple batch/chain.

And again, the only reason I’m suggesting this is the poster didn’t complain about the voice quality.



As usual, you’re right. Wrong thread. Sorry…