Audio out of sync in an OBS pre-recorded YouTube stream usually comes from either a fixed delay or drift that grows over time. First check the saved recording, then compare the beginning and end before changing OBS settings or editing the file.
A local recording uploaded to YouTube and a pre-recorded file played through OBS into YouTube Live are different workflows. The right diagnosis depends on which one you are using.
What “pre-recorded stream” means
People use “pre-recorded stream” to describe two different arrangements.
In the first, OBS records a video to your computer. You later upload that file to YouTube as a normal video, or use it as the source for another workflow. The important file is the local recording. If the saved file already has a delay, changing YouTube settings will not correct the original capture.
In the second, you play an existing video through OBS while OBS sends the output to a YouTube Live broadcast. The file may be perfectly synchronised when played locally, but the complete OBS-to-YouTube path still needs testing. This is closer to a live encoder workflow, even though the pictures and sound were prepared earlier.
Do not treat these as interchangeable. YouTube’s live encoder guidance covers encoder configuration and testing, but it does not establish that YouTube automatically repairs, causes or explains every sync problem in an uploaded pre-recorded file.
If you are deciding between an upload, Premiere and a live broadcast for a channel, the practical differences are covered in live vs Premiere vs upload. That choice is separate from the audio diagnosis, but it determines which test you should run.
Check the source recording before blaming YouTube
Start with the file produced by OBS. Play it in a desktop media player rather than judging only from the YouTube copy. Use headphones or speakers that do not introduce a noticeable Bluetooth delay, and avoid changing players halfway through the comparison.
Look for a moment where the picture and sound have an obvious relationship. A hand clap, a drum hit, a door closing, or the first consonant in a spoken word is easier to judge than continuous background music. Watch that event near the beginning, then find another clear event near the end.
Ask three questions:
- Is the sound already late or early in the local file?
- Is the gap about the same at both points?
- Does the gap become more obvious as the recording continues?
If the local file is wrong in the same way as the YouTube version, concentrate on the OBS scene, audio sources, recording settings and computer performance. If the local file is aligned but the YouTube version appears different, save a copy of the local file and note which delivery path produced the other version.
This comparison is particularly important for devotional loops, study channels and local news packages. A short sample may sound acceptable while a much longer recording gradually separates. Conversely, a single source can have a small constant delay that becomes distracting only when someone is watching a close-up speaker.
Do not use remuxing as your first sync repair. OBS explains that recording to MKV makes an interrupted recording less likely to lose the entire file, and that the recording can later be remuxed to MP4 when required. Remuxing changes the container workflow; it is not evidence of a universal cure for an audio offset or progressive drift. The OBS standard recording output guide explains that workflow.
Separate a fixed offset from growing drift
The shape of the error tells you which kind of problem you are investigating.
A fixed offset means the sound is early or late by roughly the same amount at the start and end. For example, a spoken word may begin slightly after the speaker’s mouth moves, but the relationship does not keep getting worse. One adjustment can address this kind of mismatch.
Drift is different. The recording may begin in sync, then the sound gradually moves ahead or behind the picture. By the end, the gap is noticeably larger than it was at the start. A one-time offset can move the whole soundtrack earlier or later, but it cannot correct a timing difference that accumulates.
There is also a third possibility: the apparent problem is in the playback chain. A particular media player, Bluetooth device, capture device or live delivery path may make the result appear different from the saved file. That does not prove where the fault lies, so compare the same file on a second sensible playback path before making a lasting change.
| What you observe | First check | More suitable response |
|---|---|---|
| The same gap is present near the start and end | Confirm whether sound leads or trails | Adjust the relevant OBS source for future recordings, or shift the track in an editor for an existing file |
| The gap grows during the recording | Compare sample rates, device clocks and OBS statistics | Investigate timing and performance rather than applying one fixed offset |
| The local file is aligned but the YouTube result differs | Identify upload or live encoder delivery | Preserve the local file and run a representative test in the intended workflow |
| The recording was interrupted or is difficult for your editor to open | Check the container and target software | Use OBS’s remux workflow where appropriate, without treating it as a sync repair |
For a long channel, this distinction saves repeated overnight tests. If the issue is fixed, measure and correct it once. If it is drift, find what is running at different rates or losing timing.
Measure the mismatch before changing settings
Do not begin by moving sliders until you know the direction of the error. A source that is 200 milliseconds late needs a different change from one that is 200 milliseconds early, and your installed OBS version may present the control differently from a tutorial you found.
Choose a clear event and note the relationship between the visible action and the corresponding sound. You do not need a special test duration or a claimed universal threshold. The useful question is whether the difference remains approximately stable between two points.
A simple written record is enough:
- At the first event, does the sound lead or trail the picture?
- At the later event, does it lead or trail by about the same amount?
- Does the local recording show the same result as the YouTube playback?
- Which source produces the sound: microphone, media file, browser audio, capture card or another device?
If the video contains several audio sources, test them separately where possible. A microphone may be aligned while music from a media source is delayed. If you change a global setting when only one source is wrong, you can create a new mismatch elsewhere in the scene.
For speech, use visible mouth movement and a strong consonant rather than relying on the beginning of a sentence. For music, use a percussive attack or a drum hit. For a bhajan or ambience channel with no obvious movement, add a temporary test scene containing a short visual cue and a clear sound, record it, and remove the scene after diagnosis.
Keep the original file before editing. If a change makes the timing worse, you need a clean comparison rather than a second-generation file whose timing is uncertain.
Adjust the relevant OBS audio source for future output
Use a fixed OBS audio sync adjustment only after you have established that the offset is stable and identified the source carrying it. The relevant source might be a microphone, media source, capture device or another audio input in the scene.
The control changes when that source is heard relative to the video. Make a small, documented adjustment in the direction suggested by your test, record a short sample, and compare the same visual event again. Do not copy a setting from a different computer, capture card or tutorial and assume it applies to your setup.
If the sound trails the picture, the source needs to be brought earlier relative to the video. If the sound leads, it needs to be delayed. The exact value should come from your measurement, not from a general recommendation. Verify the result in a saved recording, because monitoring and recording paths can differ.
If you use several scenes, check whether the source is shared between them or duplicated. A correction made to one source may not affect another instance. Likewise, a scene that contains a browser source, media file and microphone may need separate checks for each audio path.
If the error grows over time, do not try to hide it with a single offset. Check that the sample rate selected in OBS matches the rate used by the audio device. When two independent USB audio devices are involved, their clocks can run at slightly different rates. That is a likely cause to investigate, not a guaranteed explanation for every drift problem.
Where separate devices appear to drift against each other, routing the inputs through a suitable common audio interface can be worth considering. It is conditional, not necessary for a fixed delay, and it is not a guarantee of synchronisation. Test the actual arrangement before buying hardware.
Also inspect OBS statistics for missed or skipped frames and consider whether the computer is struggling to render or encode. The OBS encoding performance guide recommends reducing scene or workload demands when performance is a problem. Do not assume that every audio fault is caused by performance, and do not confuse network dropped frames during a live broadcast with every local recording sync issue.
For a 24/7 channel, a cloud workflow can remove the need to leave a computer recording and streaming overnight, but it does not make a damaged source file correct. StreamNeo is useful when you want to upload a prepared file once, connect it to your YouTube channel, and have the broadcast monitored and restarted without keeping your own computer switched on.
Shift the audio track in an existing file
If the recording already exists and the mismatch is fixed, correct the file in an editor rather than relying on a new OBS setting. Import the video, separate or expose the audio track, move the audio earlier or later by the measured amount, and export a new copy.
Keep the untouched original. Name the corrected file clearly, especially if you are preparing several episodes for a 24/7 loop. A name such as morning-loop-original and morning-loop-sync-corrected makes it easier to return to the source if you later discover that the timing was judged on the wrong playback device.
Preview the first clear event and one near the end before exporting. If the same offset remains at both points, the edit should improve both. If the gap is small at the start but large at the end, shifting the whole track will only choose where the error is least noticeable. You need to investigate drift instead.
When the sound needs a different correction in different parts of the recording, the file may contain a changing timing problem rather than one fixed offset. An editor can sometimes cut and realign sections, but that treats the result rather than the capture cause. For a programme you plan to repeat every day, it is usually better to find the OBS or device timing issue and make a clean recording.
If your editor rejects the recording container, follow the format requirements of that editor. OBS documents MKV, MP4 and other recording formats in its audio and video formats guide. A remux can make a file easier to use in a compatible workflow, but it should not be presented as a repair for the timing itself.
After export, inspect the new file independently. Do not assume that an editor’s preview, the exported file and YouTube playback will behave identically. Keep the corrected file available until the final delivery test is complete.
Test the workflow you will actually use
Once the local file is aligned, test the path that your viewers will receive.
For a normal upload, upload a short representative version or use an unlisted test where appropriate. Compare the uploaded playback with the local file using the same clear events. If the local recording is aligned and the uploaded result appears different, keep both copies and record the player, browser, device and delivery details before changing the source.
For a file played through OBS into YouTube Live, build a private or otherwise controlled test with the same media source, audio devices, scenes and encoder arrangement you intend to use. Include representative movement and sound rather than a still image with background music. YouTube’s encoder guidance recommends testing with the material and conditions that represent the planned stream, but that guidance is not a promise that every pre-recorded sync fault has one YouTube-side solution.
Watch the test at the beginning and near the end. If the local file and OBS preview are aligned but the live test differs, isolate the delivery path. Check the OBS output, the YouTube preview and the resulting playback separately. Avoid making several changes at once, because you will lose track of which change affected the result.
If your stream health display also reports a warning, treat that as a separate clue rather than proof of an audio fault. The guide on fixing yellow or red stream health can help you investigate the broadcast path without assuming that network status explains local recording drift.
Creators who run a recorded service, news bulletin or seasonal programme often need a repeatable pre-flight check. The recorded Sunday service rebroadcast guide covers the wider workflow. For this specific problem, the check remains simple: source file first, start and end comparison second, intended delivery path third.
Reduce the chance of a repeat fault
Before producing a long recording, make a short test with the actual microphone, media files, capture devices and scene collection. Listen to the saved file, not only to OBS monitoring. A source can sound correct in monitoring while the recorded or encoded path has a different relationship.
Use one clear audio plan. If music, speech and ambience are coming from separate devices, label each source and test each one. Avoid adding a second USB audio device merely because a tutorial uses one. Additional devices can introduce another clock and another point to investigate.
Keep the computer’s workload reasonable. Complex browser sources, animated scenes and heavy encoding settings can make troubleshooting harder. If OBS reports performance trouble, simplify the scene and retest instead of changing audio timing at the same time.
For recordings that may be interrupted, use a container suited to recovery and then remux when your editing or upload workflow requires it. The container decision protects the file from one kind of failure; it does not align audio and video that were captured at different rates.
Finally, document the working arrangement: OBS audio sample rate, devices in use, source names, measured offset and the file format used for the final output. That note is more useful than a screenshot of an unexplained slider when you return to the channel after several weeks.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Should I change the OBS sync offset if the audio drifts over time?
No. A fixed offset moves the whole audio track by one amount, while drift changes throughout the recording. Check sample-rate alignment, separate device clocks, OBS statistics and system performance before applying a one-time adjustment.
Is YouTube causing the audio delay?
Do not assume that from playback alone. Compare the local OBS recording with the YouTube version, then identify whether you used a normal upload or an OBS-to-YouTube Live encoder workflow. The comparison tells you which part of the path needs further testing.
Will remuxing an OBS recording fix sync?
Remuxing can change the container so another editor or upload workflow can use the file. It is not established as a universal repair for a fixed offset or growing drift, so check the timing before and after rather than treating remuxing as the fix.
What should I do with an existing recording that is always slightly late?
Measure the offset at a clear event near the beginning and another near the end. If it is stable, shift the audio track in an editor, export a new copy, and verify the result in the workflow your viewers will use.