Fixing Audio Drift in Long Recordings and Streams

Audio Drift in Long Recordings: Fix Progressive A/V Desync Over Time

If sync gets worse as the recording gets longer, you are probably dealing with drift rather than a fixed offset. Audio drift is a different problem from a static offset. A static offset is a fixed gap present from the first second of a recording; you measure it, enter a correction, and it stays fixed. Drift is a progressive gap that grows while the recording keeps running. A 100 ppm mismatch can produce about 180 ms of drift after 30 minutes and about 360 ms after an hour, which is enough to make a long recording feel out of sync even when the start looked clean.¹

PPM means parts per million, so a clock error becomes elapsed-time error as the session grows longer.² Every USB audio device runs on its own internal clock that nominally targets the same sample rate, such as 48,000 Hz, but no two devices tick at exactly the same rate. Measuring the offset at the start and end of a long recording confirms whether drift is your issue.

How to confirm you have drift rather than a static offset

Run the 5-clap test at the very beginning of your recording session and note the offset. Then, after 30 or more minutes of the session, pause and run the 5-clap test again without stopping the recording. If the second measurement shows a noticeably different offset, more than about 20 ms different from the first, you have a practical sign of drift.³

Alternatively, clap once at the beginning and once at the end of a test recording. Play the recording back and measure the gap between audio and video at each clap point. A difference in gap size between the two points confirms progressive drift rather than a static offset. This two-point method works with any recording software, so you can use it inside OBS, Streamlabs, or even a simple camera recording to verify whether the drift is happening within the capture pipeline or being introduced later during encoding or playback.

Two-point drift check

The two-point test for confirming drift is straightforward. Run the five-clap test at the start of your recording session and write down the offset. Without stopping the recording, open a second browser tab after 30 minutes and run the five-clap test again. Compare the two values. An offset that has grown by more than 20 ms over the 30-minute window confirms drift is present. An offset that stays within 10 ms of the starting value points to a static offset, so matching sample rates is less urgent for your setup.

Take the measurement at the same camera and microphone settings you used at the start. Changing the USB port or audio device between the two points changes the offset for a different reason and hides whether the clock relationship is the real cause. If the two readings move apart by more than 20 ms while the hardware stayed identical, the growing gap is drift rather than a setup change.

What causes sample rate drift

Every USB audio device, such as a webcam microphone, USB headset, audio interface, or desktop audio capture device, operates on its own internal oscillator. Inside OBS or any recording software, multiple audio sources mix at a shared sample rate. If one device oscillates at 48,010 Hz and another oscillates at 47,990 Hz, both are nominally 48 kHz but differ by 20 Hz. Over time, the audio samples from the two devices arrive at different rates, and the recording software resamples them to maintain a consistent mix.¹

Yet the resampling corrects audio-to-audio drift between sources, not necessarily video clock mismatch. The video clock runs on its own USB host controller timing chain. Consequently, audio that resamples correctly against the audio clock can still drift against the video clock if the video USB controller runs at a different rate.

Aligning sample rates across all audio devices in OBS

Open Windows Sound settings, select each audio input device in the list, click Properties, go to the Advanced tab, and set the Default Format to "16 bit, 48000 Hz (Studio Quality)" or "24 bit, 48000 Hz (Studio Quality)". Windows documents this Default Format path in the multimedia control panel, and Microsoft also shows 48-kHz, 16-bit stereo as a Windows endpoint format example.⁴⁵ Repeat for every microphone, headset, and audio interface connected to your streaming machine. Open OBS Settings > Audio and set the Sample Rate to 48 kHz. This matching step reduces the systematic sample rate conversion that OBS performs when a device format does not match the sample rate selected in OBS.⁶

Fixing audio drift in OBS and Streamlabs

The definitive fix for audio drift is matching all audio devices to the same sample rate: 48,000 Hz across every device in your OBS scene. Open OBS Settings > Audio and set the Sample Rate to 48 kHz. In Windows Sound settings, open each audio device's properties and set the sample rate to 48 kHz / 24-bit. Setting every device to the identical sample rate eliminates inter-device resampling as a drift source.⁴⁶

Building on this, for extremely long recordings over 4 hours, even matched sample rates can accumulate small drifts because crystal oscillators tolerate +/-100 PPM variance in the consumer-device scenario described above. Professional audio systems address that remaining clock relationship with a master clock or reference clock, especially where audio must stay locked to video timing.⁷ For most streamers and podcasters, matching all devices to 48 kHz reduces drift below the perceptible threshold for sessions under 3 hours.

For recording sessions longer than four hours, matched sample rates reduce drift but do not eliminate it entirely. Crystal oscillators at the same nominal 48 kHz still differ in practice, and that difference compounds across elapsed time. Professional broadcasters use a hardware word clock generator to lock all devices to a shared external reference, eliminating inter-oscillator drift completely. For streamers and podcasters producing sessions under four hours, matched 48 kHz settings are sufficient and cost nothing beyond the time to configure.

Drift check after a long session

For long recordings, catch a growing PPM mismatch before it hits 20 ms by running a second measurement after the session reaches the point where viewers usually notice drift. Keep the same webcam, microphone, USB port, and sample-rate settings active. If the second offset differs by more than 20 ms from the first, treat the setup as drifting and fix sample-rate matching before relying on a single static Sync Offset value.

When to use this

Run this test when recordings start in sync but sound increasingly desynchronized after 20 to 30 minutes, when you produce long-form content like podcasts or gaming streams that run over an hour, or when you notice audio at the end of a recording is further out of sync than at the beginning.

Examples