Streaming A/V Sync Calibration

A/V Sync Calibration for Live Streaming

A/V sync calibration is a recurring setup task, not a one-time checkbox. Streaming A/V sync calibration is the process of measuring the timing gap between your webcam video and microphone audio before you go live. Human lip-sync tolerance is content-dependent, but broadcast guidance defines detectability and acceptability limits in the same millisecond scale, so even a small fixed gap can make a stream feel wrong before viewers can explain why.¹

Calibrating correctly requires two things: a precise measurement of your specific hardware offset, and the correct location in your streaming software to enter the correction. The measurement changes each time your hardware configuration changes. Consequently, treating calibration as a one-time setup step and never re-measuring means your sync correction can become stale when you add a USB device, update a driver, or switch to a new machine.

Measuring your webcam-to-mic offset

The 5-clap test on this page measures the millisecond gap between your webcam video and your microphone audio under your current conditions. Select your production webcam and microphone in the dropdowns, click Start Test, and clap five times in front of the camera. The tool pairs each audio transient with its corresponding video movement and averages the readings.

Your specific offset depends on your camera model, USB host controller, OS version, and background CPU load. A camera that measures one value on one machine can measure a different value on another because USB timing, drivers, and application scheduling all affect the final path. Measure on the machine you stream from, with the exact USB devices you use during streams.

Run the calibration test with unnecessary streaming apps closed and your browser as the only application accessing the camera and microphone. The browser asks for camera and microphone permission before it can receive those media tracks, so the test should be run in the same permission state you use for production.² A clean machine state gives the browser less competing device access and usually produces a tighter five-reading spread. Measure first, then open your streaming software and enter the correction without changing any USB device connections between the measurement and the entry.

Keeping calibration tied to the live scene

The correction must match the scene you actually stream. If one scene uses a USB headset mic and another uses an interface-fed studio mic, each scene needs its own measured offset because the audio path changed. Label the OBS value with the scene, webcam, microphone, USB port, and resolution so you can restore the right correction after rebuilding a scene.

After rebuilding a scene, do not assume the old value still applies. The new scene may route audio through a different device or a different USB port than the original, which shifts the offset even when the camera is the same. Restoring the labeled value from your notes is only safe when the scene, microphone, USB port, and resolution all match the setup you measured.

Entering the correction in OBS and Streamlabs

In OBS Studio, open Edit > Advanced Audio Properties. Find your microphone in the source list and enter the measured value in the Sync Offset (ms) column. OBS exposes a source API that gets and sets audio sync offset in nanoseconds, while the Advanced Audio Properties UI converts between milliseconds and nanoseconds for the spin box.³

Streamlabs Desktop is OBS-derived and exposes audio-source sync settings through its audio API, including a syncOffset field on audio sources.⁴ In the current Streamlabs interface, use the audio-source sync setting available in your version rather than looking for a separate filter named Audio Delay. Enter the measured value for the microphone source, then verify the result with a short recording.

Choosing OBS or Streamlabs as the correction point

Use the software that produces your final output. OBS and Streamlabs share similar source timing concepts, but the panel names and interface paths are not identical, so the correction must be entered where the stream is encoded. If you use OBS for local recordings and Streamlabs for live streaming, measure once and enter the value in both applications because each one applies its own independent offset to the audio source during output.

Verifying sync in a short test recording

After entering the measured value in OBS or Streamlabs, record a short test clip before going live. Clap once near the middle of the clip. Play back the clip and scrub to the clap frame-by-frame. At the correct sync offset, the clap sound lands on the same frame as the peak hand-impact frame. At 30fps each frame is about 33 ms; at 60fps each frame is about 16.7 ms. If the sound arrives one frame early or late after applying the correction, adjust the sync offset by a small amount and re-record until the alignment is close enough for your stream.

Maintaining accurate sync over time

Re-calibrate whenever you update your webcam driver, add a new USB device to your streaming machine, or notice sync issues in recorded content. USB isochronous transfers are designed for time-dependent data such as audio and video streaming, and the host controller reserves bus intervals for those transfers.⁵ That structure helps steady delivery, but it does not make every device path equal.

Furthermore, re-calibrate after any OS update that touches USB audio scheduling. Windows and macOS updates can change driver behavior or USB scheduling, which shifts the audio path independently of the video path. The 5-clap test takes under two minutes and gives you a fresh correction value every time. Driver updates from USB chipset vendors like Intel, AMD, or ASMedia can also change the host controller's frame delivery timing, which affects the video side of the measurement even when the camera and microphone remain the same.

The safest practice is to re-calibrate at the start of the first streaming session after any major OS update or driver update. Re-calibration confirms your existing OBS or Streamlabs offset is still accurate before viewers see your stream. Keeping a log of your measured offsets alongside the driver versions and OS build numbers helps you correlate future sync shifts with specific updates, making it faster to diagnose whether a recent change caused the problem or whether the shift came from a hardware or USB topology change instead.

When to use this

Watch for a driver update silently breaking your OBS offset before the first stream on a new machine, after any change to your USB device configuration, or any time a viewer reports lip-sync issues on your stream.

Examples

Fixing OBS Audio Sync with a Browser-Based Offset Measurement

The fastest OBS audio sync fix is to measure the actual timing gap instead of guessing. OBS audio sync problems have one root cause: the video and audio paths take different amounts of time. Video from a USB webcam travels through the camera's ISP, a USB host controller, and a browser or OBS video decoder before landing in the timeline. Audio from a USB microphone follows a shorter path through the audio class driver and OBS audio stack. The difference between these two path lengths is the offset you measure and correct.¹

Before adjusting any settings in OBS, measure the actual offset with the 5-clap test on this page. Getting a measured number eliminates the guesswork of trial-and-error OBS adjustments. The browser test uses the camera and microphone inputs you choose through getUserMedia, which prompts for permission before the page can receive media tracks.² Typical USB webcam setups often show audio arriving before video, but the exact value changes by machine, driver, and USB topology.³ Once you have the measured value, entering it in OBS's Sync Offset field takes under 30 seconds.

How to measure the offset before touching OBS

Run the 5-clap test on this page with your actual streaming microphone and webcam selected as the inputs. Each clap generates one timestamp pair. After five pairs the tool displays an average offset with a direction label and a pre-calculated OBS Audio Offset value. Write down or copy that value before opening OBS; the measurement disappears when you navigate away from the page.

For the most accurate reading, run the test with OBS closed. When OBS is open and capturing the same camera, it competes with the browser for the video feed, which can add jitter to individual frame timestamps and inflate the spread across your five clap readings. Measure in the browser alone, then open OBS and apply the value.

The most accurate measurement comes from running the five-clap test with OBS closed. When OBS has your webcam open as a video source, it shares the camera hardware with the browser, adding scheduling overhead to the video frame delivery. The browser must negotiate frame access with OBS on every capture cycle, which increases per-frame variance and widens the spread across your five clap readings. Close OBS, run the browser test, record the value, then open OBS and enter the correction.

Matching OBS source and microphone pairing

Make sure the OBS value you measured corresponds to the same webcam and microphone source you use in the scene. If your scene has multiple microphones, the Sync Offset column applies to one source at a time. Enter the value on the microphone row that feeds your stream, not on desktop audio, game capture audio, or a secondary mic. Taking an extra moment to verify the source name in the browser dropdown matches the OBS source name prevents the common mistake of correcting the wrong audio path and wondering why the sync still looks off.

If your OBS setup has both a webcam audio source and a desktop audio source, the Sync Offset column is per-row, not per-scene. Choosing the desktop audio row by mistake shifts only the system sound, leaving the mic path that users actually hear unchanged. Matching the browser dropdown label to the OBS source name is the quickest way to avoid that mistake.

Entering the offset in OBS Studio

In OBS Studio, open the Edit menu and select Advanced Audio Properties. The table lists every audio source in your current scene collection. Find your microphone source in the list and click the Sync Offset (ms) column cell for that row. Enter the OBS Audio Offset value this tool displayed, then press Tab or click another cell to confirm the value is accepted before closing the dialog.⁴

OBS implements that field as a millisecond sync offset for each audio source, with the source file showing a numeric input range from -20000 ms to 20000 ms and converting the entered value to nanoseconds before applying it.⁵ A positive value delays that audio source, which is the usual correction when audio arrives before video. A negative value lets OBS use buffered audio to move the source earlier relative to the rest of the mix, which is useful when video arrives before audio.⁴

After entering the value, close Advanced Audio Properties and record a 10-second test clip. Play back the clip and watch whether your clap sound aligns with visible hand impact in the footage. If the alignment looks correct in the test clip but your live stream still shows desync, check whether OBS is applying the offset to the recording output but not the streaming output, since OBS maintains separate output paths that can have different sync states.

Confirming the saved OBS offset

After entering the OBS Audio Offset value, verify it saved correctly by closing Advanced Audio Properties and reopening it. OBS saves the scene collection when that dialog closes, but confirming the value on reopen protects you from a silent entry mistake.⁶ If the field shows the correct value on reopening, the correction is active. Record a 10-second test clip, play it back, and clap once visibly and audibly in the clip to check alignment.

When to re-measure instead of reusing an old value

Your OBS sync offset is not permanent. Re-measure whenever you add or remove a USB device that shares the webcam's host controller, because USB bus load changes video frame timing.³ Driver updates from Logitech, Elgato, or your USB chipset vendor can also shift the pipeline latency, so verify the old value after any driver or firmware update.

Consequently, the safest workflow is to measure at the start of each streaming session rather than relying on a value from a previous session. The 5-clap test takes under two minutes and gives you confidence that your stream will be in sync from the first minute of content, not just the first session after you set up the offset.

When to re-measure after USB or driver changes

Several events require re-measuring rather than reusing the saved OBS offset. Adding or removing any USB device that shares the webcam's host controller changes the bus load and shifts the video frame timing. Updating the webcam firmware through Logitech G HUB, Elgato Camera Hub, or Razer Synapse can shift the pipeline latency, so get a fresh measurement before you trust the saved value. Switching to a different USB port on the same machine also changes the offset if the new port is served by a different host controller. The five-clap test takes under two minutes; run it again whenever any of these changes occur to keep your OBS correction accurate.

When to use this

Confirm whether OBS needs a positive or negative Sync Offset when recordings show audible lip-sync gap, when a viewer reports your audio is slightly ahead of or behind your mouth movement, or when you have changed your streaming hardware setup and want to verify the sync before going live.

Examples

Streamlabs OBS Audio Sync Setup Guide

Start in Streamlabs with the same microphone source your stream actually uses. Streamlabs OBS uses the same source-level audio delay concept as OBS Studio, but the control is reached through Advanced Audio Settings instead of the OBS Studio Advanced Audio Properties table. In OBS Studio, you enter the value in the Sync Offset column; in Streamlabs OBS, you open Advanced Audio Settings from the audio mixer and enter the value in the Sync Offset column there. Knowing this distinction prevents the common mistake of looking for the control in the wrong panel and wondering why the recording is still out of sync.¹

Measure your offset with the 5-clap test before changing Streamlabs settings. The measurement gives you a millisecond value and a direction; when audio arrives before the matching video frame, enter that value as the Streamlabs Sync Offset for your microphone source. That positive value goes into the Streamlabs Sync Offset field for your microphone source.

Where to find the sync control in Streamlabs OBS

In Streamlabs OBS, click the settings icon near the top right of the Audio Mixer panel. Open Advanced Audio Settings, find the Sync Offset column for your microphone or other audio source, and enter your measured millisecond value. A positive Sync Offset delays the audio source by that number of milliseconds, compensating for a video stream that arrives later than audio.¹ Applying it to the microphone source rather than to a background music or game audio source ensures only the voice track receives the delay correction.

Adding Sync Offset in Streamlabs

To add Sync Offset, click the settings icon near the top right of the Audio Mixer panel, then open Advanced Audio Settings. Find the Sync Offset column for the audio source you want to delay, enter your measured value from the five-clap test, and close the settings window. The Sync Offset value is active immediately without restarting Streamlabs. You can verify the correction by recording a short clap test clip and checking whether the sound aligns with the visible hand impact before you go live.

Keep the settings window path in mind when you switch machines. The control sits in Advanced Audio Settings rather than the main settings menu, so a fresh Streamlabs install can hide it if you look in the wrong place. The fastest confirmation is still a short clap clip, because the alignment you see there is the same sync your live viewers receive from the Streamlabs output.

Common sync issues specific to Streamlabs

Streamlabs OBS and OBS Studio should be verified separately, so a value that works in one app may need a small adjustment in the other. If an OBS Studio offset looks correct in recordings but the same value in Streamlabs still shows slight desync, try nearby values in the Streamlabs Sync Offset column and verify with a test recording. Part of that difference is the capture source configuration: OBS exposes a Windows-only Use Device Timestamps option on audio input sources, described as an attempt to use device timing information to prevent desync, and it is off by default.²

The second common issue is noise suppression. OBS documents Noise Suppression as an audio-source filter with methods such as RNNoise, Speex, and NVIDIA Noise Removal.³ RNNoise's source defines a 480-sample processing frame, which is why noise suppression should be treated as an active processing stage rather than a passive setting.⁴ Any noise-suppression filter changes the audio processing path, so measure the offset with it active to capture the actual production offset including the suppression overhead. Disabling noise suppression for the measurement and re-enabling it for streaming produces an offset that is smaller than the actual production delay, resulting in a correction that under-corrects and leaves a visible sync gap. Run the five-clap test in the same software configuration your stream uses.

Noise suppression belongs in the measured path

If you use noise suppression, gate, compressor, or limiter filters on the microphone source, measure with RNNoise or Speex still switched on. Those filters add processing stages before the final Streamlabs output, so the measured offset must include them. Turning a filter off for the measurement and back on for the actual stream produces a correction that is too small, because the filter's own processing delay is missing from the number you entered. Keeping every production filter active while you clap gives you the true end-to-end delay that Streamlabs will apply to your live audio.

Verifying the sync in a Streamlabs test recording

After applying the Sync Offset value, record a 10-second test clip in Streamlabs. Use the built-in recording mode rather than a third-party recording application; recording through Streamlabs verifies the sync as it applies to your Streamlabs output, not to a separate capture. Play back the test clip and clap once during playback while watching the frame.

Building on this, the clap sound should arrive at the exact frame where your hands visibly contact. If the sound arrives a few frames before or after the visible impact, adjust the Sync Offset by a small amount and record again. A small number of iterations usually produces frame-accurate sync for most setups. CapyToolkit allows re-running the 5-clap test any time without closing the browser; refresh the page to reset the measurement.

Iterating the Sync Offset value between test recordings is the fastest path to frame-accurate sync in Streamlabs. After applying the initial measured value, record 10 seconds and clap once at the midpoint. If the sound arrives one frame early at 30 fps, add about one frame duration to the Sync Offset value; if it arrives one frame late, subtract about one frame duration. At 30 fps, one frame is about 33 ms, and at 60 fps it is about 17 ms.⁵ More typically the initial measurement produces residual error under one frame, and a small adjustment in the correct direction achieves alignment within the same session.

Fine-tuning Streamlabs after the initial measurement

The first measured value is the starting point, not the end of the process. After applying it, record a short clip and clap once at the midpoint. If the sound arrives one frame early at 30 fps, add about one frame duration to the Sync Offset value; if it arrives one frame late, subtract about one frame duration. At 30 fps, one frame is about 33 ms, and at 60 fps it is about 17 ms. More typically the initial measurement produces residual error under one frame, and a small adjustment in the correct direction achieves alignment within the same session.

When to use this

Run this test before first streaming with a Streamlabs OBS setup, after adding a noise suppression plugin to your audio source, after updating Streamlabs or its audio drivers, or when Twitch VOD review shows lip-sync drift that was not visible during the live broadcast.

Examples

Elgato Cam Link 4K A/V Sync and OBS Setup

Measure the full HDMI capture stack before entering any OBS delay. Cam Link 4K connects a DSLR or mirrorless camera to your computer through HDMI, then presents that camera feed as a USB webcam for streaming, video calls, and recording.¹

That capture stage changes how you should measure sync. Instead of applying a generic webcam offset, run the 5-clap test with the exact camera, HDMI mode, USB port, microphone, and recording software you plan to use. Then enter the measured value in OBS Sync Offset for the audio source you selected during the test.²

Why Cam Link A/V sync differs from USB webcam sync

A native USB webcam sends its own processed video stream directly to the computer. Cam Link 4K takes a camera's HDMI output, converts it through the capture device, and exposes the result as a USB video source.¹ Windows also provides a system-supplied USB Video Class driver for UVC devices, which is why class-compliant capture devices can appear as webcam-style video sources without a custom driver.³ The browser cannot know how much time your camera spent preparing the HDMI feed before the capture device converted it, so the full stack needs one measurement.

The audio path matters just as much as the video path. If your microphone is a separate USB device, the browser compares that audio input against the Cam Link video input. If you choose HDMI audio from the camera instead, you are measuring a different audio path for the same video feed. Keep that audio path consistent between measurement and correction.

Audio path selection before OBS

This is why a Cam Link setup should not borrow an offset from another creator's setup. A different camera body, HDMI output mode, USB port, or audio input can change the timing relationship, even when the same capture card remains in the chain. Selecting the wrong audio input for the measurement produces a correction that targets a different pipeline than the one OBS actually records, which can make the sync worse instead of better.

Pick one audio input and keep it for both the measurement and the correction. If the browser dropdown shows both HDMI audio and a USB microphone, the choice decides which pipeline the measured number describes. Switching inputs between the two steps silently changes the audio side of the offset, so the OBS correction ends up applied to a path the recording never used.

Measuring the offset for a Cam Link setup

Select "Elgato Game Capture HD" or "Cam Link 4K" as the video input in the browser dropdown. Choose the microphone you will actually use, whether that is a separate USB microphone or the camera's HDMI audio input. Run the 5-clap test with the camera connected to the Cam Link and the Cam Link connected to a USB 3.0 port.⁴

Use the fastest shutter speed your lighting allows before the test. A fast shutter freezes the clap motion and makes the luminance edge easier to timestamp, while a slow shutter spreads hand movement across the exposure and softens the visual transient.⁵ If 1/250s is too dark, use the fastest practical setting and keep it fixed while you repeat the five claps.

After the first run, repeat the test once or twice without changing the camera, USB port, microphone, or OBS settings. If the spread is wide, check focus, exposure, and hand motion before trusting the average. A stable average is more useful than a single measurement that happened to land cleanly.

Applying the correction in OBS

After measuring, enter the offset in OBS Advanced Audio Properties for the audio source you tested. OBS exposes the sync offset control as a millisecond spin box and converts the entered value to nanoseconds when applying it to the source. The conversion from milliseconds to nanoseconds happens internally, so you always think and enter values in milliseconds regardless of the internal precision OBS uses for the actual audio delay calculation.²

Because Cam Link sits between the camera and the browser, the correction should match the complete capture stack. Do not subtract a generic USB webcam offset or reuse a value from a different camera body. If you swap cameras, change HDMI output settings, move the Cam Link to another USB port, or change the audio input, run the 5-clap test again and update the OBS value. The Cam Link's FPGA encoding stage adds a consistent base latency on top of whatever the camera's HDMI output contributes, so the total offset is always the sum of both stages and cannot be estimated from camera specs alone.

Re-measuring after switching cameras on the Cam Link

Swapping camera bodies on the same Cam Link requires re-measuring the A/V offset. The capture card may be the same, but the camera can output a different processed HDMI signal with different internal encoding latency. Record each camera's measured offset so you can restore the correct OBS Sync Offset quickly after a swap. Label each value with the camera model and HDMI output mode so you can pull up the right number without re-running the test every time you change bodies.

HDMI audio versus external microphone pairing

If you use HDMI audio from the camera, measure with that HDMI audio input selected. If you use a separate USB microphone, measure with that microphone selected. Those are different audio paths, and one OBS value cannot correct both pairings accurately. Keep notes for each pairing so you can restore the right correction when you switch between HDMI audio and an external mic.

When to use this

Run this test when setting up an Elgato Cam Link 4K or similar HDMI capture card for the first time, when switching to a new camera body in an existing Cam Link setup, or when OBS recordings show a lip-sync gap that a standard USB webcam offset correction did not fix.

Examples