Fixing Audio That Leads Video in Webcam Recordings

Audio Leads Video Fix: When Your Voice Arrives Before Your Lips Move

Start by checking the direction label before deciding which track needs correction. Audio leading video is less common than video delay, but it happens. When audio arrives at OBS before the matching video frame, viewers hear your words before seeing your lips move; it produces the same uncanny experience as watching a poorly dubbed film, but in reverse. The most common cause is a camera with unusually low processing latency, a particularly fast video pipeline, or audio being delayed relative to video by another factor in the signal chain.

A common source of audio-leads-video is a Bluetooth microphone combined with a wired webcam. Bluetooth audio codecs add 40–200 ms of codec processing delay after the hardware microphone samples the sound.¹ Yet a wired USB webcam delivers video on the standard 60–100 ms path.² If your Bluetooth codec delay is shorter than your webcam video delay (a narrow but possible window), audio arrives at the browser after video, producing a negative offset.

What causes audio to lead video

Audio leads video when the video processing path is longer than the audio path by more than the typical gap. Inside a CPU under heavy load, video frame decode can complete faster than audio buffer handoff when the audio driver uses a larger latency buffer for power savings. On certain macOS builds with Apple Silicon, the video pipeline is significantly faster than on Intel Macs, producing negative offsets on configurations that were positive on the previous machine.

Bluetooth headsets add a known fixed latency from their codec: SBC adds 150–200 ms, aptX adds 40–70 ms, and aptX HD adds around 200 ms.¹³ If your video pipeline delay is shorter than the Bluetooth codec delay, audio arrives at the browser after video, producing a negative offset in the test result.

When Apple Silicon or Bluetooth reverses the offset

Apple Silicon Macs can reduce video pipeline delay enough to flip the offset polarity compared with older Intel machines.⁴ A setup that measured positive on Intel might measure closer to zero or even negative on Apple Silicon, so measure on the exact machine and OS you use for streaming because the offset cannot be predicted from specs alone.

Treat the measured value as machine-specific rather than model-specific. Two streamers with the same camera and microphone but different host hardware can see opposite polarities, so copying a friend's number is risky. Run the 5-clap test on your own Apple Silicon or Intel machine whenever you change the OS version, because a system update can move a borderline reading across the zero line.

Measuring and interpreting a negative offset

Run the 5-clap test with your actual webcam and microphone. If the result displays an "Audio Leads" direction label and a negative offset value (for example, −45 ms), your audio is arriving at the browser after your video frame. The magnitude tells you how far after, and that number is the correction value you enter directly into the OBS Sync Offset field for the audio source that lags behind the video timeline.

In OBS, a negative sync offset uses buffered audio data so the selected audio source becomes the reference and the rest of the mix is delayed around it.⁵ This is counterintuitive, because you might expect to delay the video instead. Building on this, OBS's Sync Offset field only delays audio sources, not video sources. For audio that arrives late relative to video, entering a negative offset in the field lets OBS align the tracks without advancing time.

A negative offset number in the result tells you the audio path is slower than the video path by that many milliseconds on your current setup. A −10 ms reading is perceptually insignificant; it falls within the In Sync range the tool displays. A −45 ms reading falls at the human perception threshold for audio-leading-video; most viewers perceive it as a slight uncanny mismatch, and a −90 ms reading is clearly detectable and requires correction.⁶ Entering the negative value in the OBS Sync Offset field lets OBS align the tracks without advancing time. The negative offset is less common than a positive one, but it is not an error; it simply means your particular combination of camera, microphone, and system configuration produces a faster video path than audio path, which can happen with certain Bluetooth setups or on machines with unusually efficient video decode pipelines.

How much negative offset needs correction

The magnitude matters as much as the direction. A −10 ms reading is perceptually insignificant; it falls within the In Sync range the tool displays. A −45 ms reading falls at the human perception threshold for audio-leading-video; most viewers perceive it as a slight uncanny mismatch, and a −90 ms reading is clearly detectable and requires correction. The practical takeaway is that any negative offset above −45 ms should be corrected in OBS before you go live, while readings between −5 ms and −15 ms fall near the measurement noise floor and can generally be left alone.

Why a negative value is still an audio correction

A negative offset does not mean the test failed. It means the audio path is the slower path in your current setup. OBS can only delay audio sources directly, so the negative value tells OBS to delay the audio source by that amount. The video stays where it is, and the audio moves into alignment with the already-arrived video frame.

Fixing audio-leads-video in streaming software

In OBS Studio, open Edit > Advanced Audio Properties and enter the negative measured value in the Sync Offset (ms) column for your audio source. A value of −45 means OBS will delay your audio source by 45 ms before mixing it with the video output, aligning the two tracks.

For Bluetooth microphone users, the most reliable fix is switching to a wired USB microphone for the recording session. Bluetooth codec delay can vary by tens of milliseconds as connection quality changes, making consistent calibration difficult.⁷ Wired USB microphones deliver consistent, low-latency audio that produces stable positive offsets easy to correct with a static OBS value.

Switching from a Bluetooth microphone to a wired USB microphone eliminates codec delay as a variable and usually produces a positive offset in the normal range. A wired USB condenser microphone with direct monitoring has a predictable, low-latency audio path.²⁸ Paired with most USB webcams, which have 60–110 ms video path latency, the result is a stable positive offset between +40 and +90 ms. Positive offsets are easier to correct consistently than Bluetooth offsets, which drift between sessions due to codec fallback from interference. That is why switching off Bluetooth flips the offset positive and worth confirming with a fresh five-clap reading.

When to use this

Run this test when recordings show your words arriving before your lips move, when you are troubleshooting a Bluetooth headset or wireless microphone setup that seems to arrive ahead of the camera, or when you switch to a new camera with a faster processing pipeline and suspect the polarity of your sync offset has changed.

Examples