A/V Sync by Webcam Model

A/V Sync by Webcam Model

The offset you measure belongs to a configuration, not to a brand. Move a camera to another USB port, change its resolution, load a different software preset and the number moves with it, which is why a figure copied from a review or from someone else's desk rarely lands. What the model does determine is which settings you should fix before measuring, and what tends to destabilise the reading once you start.

Each section below covers one brand and its models. The pattern is the same throughout: settle the camera into the mode you actually stream in, close or match the vendor software, then run the five-clap test and enter the result in your app's sync field.

Logitech

Logitech's cameras run as standard USB video devices, so nothing needs installing before the browser can see one and you can start measuring.¹ The differences between the models come from frame rate and from how much work the camera does on every frame, which means the first thing to settle is the resolution and frame rate you actually stream with rather than whatever the camera happens to default to when a page opens it. Hold the cabling still for the whole measurement too, because moving a device between ports changes the bus load and can shift frame delivery by enough to move the reading; run the five claps, then run them again without touching anything and compare the two spreads before you trust the average.

Razer

The two Kiyo models here sit at opposite ends of the range: one has a ring light that changes how the clap itself is detected, and the other carries a large sensor that changes how much data every frame moves through the USB link. Both are sensitive to the state of Razer's software while you measure, so decide before you start whether you want a hardware baseline with the software closed or a production reading with your own profile loaded, and then keep that choice fixed. Take the reading from the position the camera will hold during the stream too, because both models autofocus and both can settle differently after a small move; if you change a profile or a ring-light setting between runs, measure again instead of reusing the old value.

Elgato

Both Facecams apply their image processing through Camera Hub, and both send far more video data than a webcam that compresses harder, which makes USB conditions unusually visible in the reading.² Treat the connection as part of the measurement rather than as an afterthought, because a busy bus shows up as a wider spread across the five claps rather than as a steady error you could correct with a single number. Run the test with the preset your scene uses, since every Camera Hub control changes the work done on each frame; if the cluster tightens after you move the camera to its own controller, the earlier spread was bandwidth rather than anything about the camera's own timing.

EMEET

One model sits in this group, and its own settings decide most of the reading, because the resolution it starts in and the microphone you choose both belong to the camera rather than to your machine. EMEET's camera can begin a session in a lower mode than its sensor supports, so confirm what the browser is actually receiving before you treat a value as final. Fix the resolution, the audio device and the USB port before the first clap, then change only the sync value in your streaming app, because nothing else about the setup should move while the five readings are being collected.

OBSBOT

This camera moves, which is the whole reason it needs a different routine from a fixed webcam. OBSBOT's software controls tracking, zoom and imaging settings, and any of them can change the frame while the test is timing a clap, so the order of work matters more here than the value you end up with. Park the camera, close the software, and only then measure; if your stream uses tracking, measure the parked position anyway and expect the number to describe the camera rather than the framing it will produce once tracking starts moving it around.

Logitech C920x A/V Sync & Latency Test

C920x delivers 1080p/30fps over USB-A and Logitech lists the C920 family as a USB 2.0 UVC webcam with 78° diagonal field of view and autofocus.¹ Connect the cable, open this page in Chrome or Edge, click "Enable Camera & Mic", and the browser requests camera and microphone tracks through getUserMedia before it can read either stream.²

The C920x video path usually carries more processing than the microphone path because the camera sends video frames while the browser receives audio samples as a separate MediaStream track. The measured offset can be positive or negative depending on the USB controller, operating system, and browser state, so use your five-clap result rather than a generic published figure. When the result says Audio Delay, enter the positive value in OBS Studio's microphone Sync Offset field to delay the audio until it matches the video frame.

Specifications¹

Connecting the C920x without a hub

Connect the C920x to a USB port directly on your computer rather than through a hub. USB hubs and shared host controllers can change bandwidth availability, and Microsoft describes USB bandwidth as a shared controller resource that varies by system and device topology.³ Chrome and Edge both support the getUserMedia API the tool requires; grant camera and microphone permissions when prompted. After granting access, select your microphone from the audio input dropdown if multiple devices appear.

Hot-plugging is supported. Disconnect and reconnect the cable at any time and the browser device selector refreshes automatically without reloading the page. If the camera does not appear in the video input list, confirm the browser has camera permission in your operating system settings, not just in the browser's site permission dialogue.

Camera and microphone permission checklist

Before the test session, trace the C920x cable to confirm it runs directly into a rear motherboard port rather than into a USB extension cable or hub. Extension cables and hubs add extra device topology between the camera and the host controller, which can widen the spread across your five clap readings. A tighter spread gives you more confidence in each individual reading, so move the C920x to a port on a separate controller if the readings jump around.

USB bandwidth contention on the C920x

USB bandwidth contention is the most common source of elevated offset on the C920x. When other high-bandwidth USB devices share the same USB host controller, the camera's video stream and the microphone stream compete with the rest of the device tree for controller time. Microsoft notes that USB bandwidth competition comes from both hardware and software, so moving the C920x to a different USB port can test whether those ports sit behind a separate controller.³

Keep room lighting stable if your five readings spread apart, because changing brightness can make the luminance spike harder for the detector to timestamp cleanly. Inconsistent lighting forces the camera to adjust exposure between claps, which changes the baseline luminance level and can shift the apparent peak of the transient by one frame in either direction.

USB host controller position and offset variance

Device Manager shows which host controller serves each USB port on your machine. Open Device Manager, expand Universal Serial Bus controllers, and look for multiple USB host controller entries; most modern desktop boards list two to four. Right-clicking a host controller entry and selecting Properties then the Devices tab shows which ports share that controller. Moving the C920x from a port on a busy controller to a port on a less-loaded controller reduces contention, tightening the spread across your five measurements without changing your average offset.

Reading the C920x result and entering it in OBS

The tool clocks a visible luminance transient in the video frame against a simultaneous amplitude transient in the audio stream. It reads canvas pixel data with getImageData to find the video spike, then uses Web Audio time-domain data to find the matching audio spike.⁴⁵ Inside the browser, both timestamps use the same performance.now() clock, so the measurement is a direct comparison rather than an estimate.⁶ Five clap pairs average out individual frame jitter and timing outliers.

Beyond the raw offset number, the result screen shows the direction label: Audio Delay (positive value), Audio Leads (negative value), or In Sync (within ±5 ms). Building on this, copy the OBS Audio Offset value and paste it directly into OBS Studio under Edit > Advanced Audio Properties for your microphone source. OBS forum guidance identifies that dialog as the place to enter Sync Offset in milliseconds, where 1 second equals 1000 ms.⁷

Using OBS Audio Offset correctly

Copying the OBS Audio Offset value immediately after the result displays avoids re-running the test when you open OBS. The result screen does not persist after page navigation. Paste the value directly into the Sync Offset field in OBS Advanced Audio Properties while the conditions that produced the measurement still apply: the same USB devices connected, the same background processes running. An offset value captured and applied within the same session reflects the same USB bus state that produced the measurement.

Apply the value while the page stays open. The result screen does not reload, so the offset you copied reflects the USB bus state that produced it. Treat a reading inside ±5 ms as In Sync, but still apply it if viewers report drift, because small residual gaps surface only during fast motion. Close background apps and retest your C920x after a USB port swap, since that shift can change the average offset enough to matter.

Logitech Brio 500 A/V Sync & Latency Test

For a Brio 500, the important sync number is the one measured on your desk, not the marketing latency figure. Brio 500 is a Full HD 1080p USB-C webcam with RightLight 4 HDR, RightSight auto-framing, and Show Mode.¹ Connect it to a USB-C port, grant camera and microphone access in Chrome, and the clap test starts after the browser receives both media tracks through getUserMedia.²

Logi Tune can customize Brio 500 settings, including AI-optimized framing, so disable RightSight before testing if you want a fixed framing state across all five clap readings.³ Measure on your specific machine because Windows audio buffers and driver settings can change capture timing; driver versions and USB topology affect the actual offset more than a published camera figure.

Specifications¹

USB-C, adapters and Logi Tune before you measure

Connect the Brio 500 via USB-C before opening this page. If your machine lacks a USB-C port, the included USB-C to USB-A adapter can still provide basic plug-and-play camera access, but moving the camera to a different USB topology can change available bandwidth and timing.⁴ Chrome recognizes the camera after you grant the getUserMedia permission; no Logitech software installation is needed for basic camera access.

Disabling RightSight auto-framing before testing produces more consistent results. Open Logi Tune, navigate to the Brio 500 camera settings, and toggle RightSight off. With auto-framing disabled, the camera delivers a fixed framing state instead of changing the crop while it detects and tracks your face. Leaving RightSight active during the test can cause the camera to reframe between claps, which changes the luminance distribution in each frame and makes the detector work harder to find a clean transient.

Stabilizing Logi Tune and framing settings

Set the Brio 500 to the same framing state before every test. If you use Show Mode, HDR, or a custom Logi Tune preset, keep that preset active during all five claps. Switching presets after the test can change processing time and make the saved OBS value less reliable for the setup you actually used. Keeping a written note of the active preset alongside the measured offset helps you reproduce the same conditions when you re-enter the value in OBS after a session break.

Verifying camera permissions in Windows 11

If the Brio 500 does not appear in the video input dropdown after clicking Enable Camera and Mic, check both browser permissions and OS-level permissions. Windows 11's Privacy and Security settings include separate camera permission toggles for apps and for browsers; if the browser-level toggle is off, Chrome cannot enumerate any webcam regardless of its internal permission status.⁵ Open Settings > Privacy and Security > Camera on Windows 11 and confirm that browser-level camera access is on before reloading the page and trying again.

Driver and OS updates that move the Brio 500 offset

Windows audio latency depends on the driver and buffer configuration, so a Windows or webcam driver update can change the audio side of your Brio 500 measurement.⁶ If you measured the Brio 500 before a major OS or driver update and entered that value in OBS, re-measure now. A changed audio buffer path can shift the apparent offset even when the camera hardware has not changed.

RightSight active is the second common issue. When enabled, RightSight changes the active framing state while it detects and tracks your face. This can make individual clap readings spread further apart across five tests because the camera crop shifts between claps, altering the luminance baseline the detector uses for each measurement. Disable it in Logi Tune and retest if your five measurements show more than 25 ms spread, and wait three seconds after disabling before starting the test so the camera fully settles into its fixed framing state.

Run the five-clap test after any major Windows update to confirm the offset has not shifted. Keep the same USB port, same microphone, same Logi Tune settings, and same browser between measurements so the result reflects the setup you are about to use for recording or streaming. Treat the measured offset as a snapshot of your current configuration rather than a permanent property of the camera alone, and rerun the test after any system change that might touch the audio or video path.

Logitech's own webcam comparison chart lists RightSight auto-framing and HDR as standard on the Brio 500, features the simpler Brio 300 in the same lineup does not carry.⁷ That extra per-frame processing is consistent with the Brio 500 typically reading higher than a minimal-processing camera like the C920x.

Applying the Brio 500 result and checking a test clip

Inside the browser, the tool samples the video frame canvas for a luminance spike and the Web Audio AnalyserNode for a transient above the calibrated noise floor. Both use the same performance.now() clock, which is a high-resolution monotonic timer that is not subject to system clock adjustments. Five paired timestamps produce an averaged offset with direction, and the five-reading design helps cancel out any single-frame timing anomaly caused by a momentary CPU spike or USB scheduling delay.

For the Brio 500, the offset direction is almost always positive (Audio Delay), because video processing runs longer than audio. Building on this, the result panel displays an OBS-ready value labeled "OBS Audio Offset". Enter that number in OBS Studio under Edit > Advanced Audio Properties > Sync Offset (ms) for your microphone source. Logi Tune and the Windows Camera app do not expose a sync offset control, so OBS or Streamlabs is where the correction must be applied. The Brio 500's RightLight 4 HDR pipeline adds more in-camera processing than the older C920x, which is why the measured offset tends to be slightly higher on the same USB port and microphone combination.

Confirming the offset in a test clip

After entering the OBS Audio Offset value, record a 10-second test clip and clap once clearly in front of the camera. The clap sound should align with the frame where your hands visibly contact. Brio 500 recordings corrected with the measured offset typically show alignment within one or two frames at 30fps, which is 33–67 ms of residual error and below the perceptible threshold for passive viewers. If the sound arrives noticeably before or after the visible clap in the test clip, adjust the OBS Sync Offset by 10 ms in either direction and record again.

During the test clip, match the conditions to your real session and account for the Brio 500's HDR processing lag. Keep the same USB port, microphone, and Logi Tune preset you used for the measurement, because the Brio 500's extra in-camera HDR processing makes the offset sensitive to those choices. A positive result means Audio Delay, so the value you entered pushes audio later until it meets the frame. If viewers still notice a gap after the first correction, nudge the Sync Offset another 10 ms and re-record until the clap lines up.

Logitech Brio 300 A/V Sync & Latency Test

Measure the Brio 300 on your own USB path before relying on a generic sync value. Brio 300 is Logitech's USB-C webcam for 1080p video calls. It delivers 1080p/30fps over USB-C, and Windows can use its system UVC driver without a Logitech driver.¹ Logitech lists RightLight 2 auto light correction and a noise-reducing mic, while the support specs list fixed focus and an integrated privacy shutter.

Chrome receives the camera and microphone tracks after you grant getUserMedia permission, then this tool compares the luminance and audio transients from those local tracks.² Because Logitech does not publish an HDR or auto-framing pipeline for the Brio 300, treat it as a simple fixed-focus UVC camera rather than a camera with active processing. Measure on your machine, because host controller, browser state, and USB topology affect the actual offset.

Specifications³

Plug-and-play setup and fixed focus on the Brio 300

Plug the Brio 300 into a USB-C port before navigating to this page. Windows can use the system UVC driver for compliant USB video devices, so the camera can enumerate without a Logitech driver installation.⁴ Click "Enable Camera & Mic", approve the browser permission prompt, and the test starts after Chrome receives both media tracks through getUserMedia. Fixed focus means there is no autofocus hunt during the test; sit at your normal monitor-top distance for the sharpest clap detection image.

No Logi Tune configuration is needed or recommended before testing. The Brio 300 has no software-adjustable processing settings that would affect the A/V offset. This simplicity is actually an advantage for sync measurement because there are fewer variables that could shift between sessions, making the Brio 300 one of the more consistent cameras for repeated five-clap tests across different days and system states.

Fixed focus and stable working distance

The Brio 300's USB-C connector works without any Logi Tune configuration before the test. Because the camera has no auto-framing, HDR pipeline, or adjustable processing settings, there is nothing to disable before measuring. Staying at your normal working distance ensures the fixed-focus lens delivers a sharp image, which sharpens the luminance transient the clap creates and gives the detector the best possible frame-to-frame contrast for an accurate timestamp.

Auto-exposure and changing room light

Automatic exposure adjustment is the main processing step to watch with the Brio 300. In a bright room with consistent light, auto-exposure stabilizes quickly and produces clean clap detection. In a room where light changes frequently, the exposure algorithm adjusts frame-by-frame and can smooth out the luminance transient the clap creates. Clap near the camera in such conditions to maximize the luminance change detected.

Fixed focus also means the camera delivers a blurry image when you sit closer than its target distance. Blurry images reduce the sharpness of the luminance change at the clap boundary, which can cause missed detection or shift the detected transient by one frame. Sit at your normal working distance rather than leaning in, and verify that your hands are clearly visible in the video preview before starting the test so you know the detector has a sharp image to analyze.

Stable lighting for repeatable claps

Maintaining consistent room lighting throughout the test run reduces auto-exposure adjustments that change the frame brightness between claps. A window with direct sunlight shifting behind clouds during the test causes the sensor to open and close the exposure within the test window, creating inconsistent frame brightness at the clap boundary. Running the test in a room with stable artificial lighting, or with the window blinds drawn, produces five readings with tighter spread and a more reliable average.

Why Brio 300 readings cluster tightly

Five luminance and audio transient pairs produce an averaged offset value with a direction label. The Brio 300's simplified pipeline means the five readings cluster tightly, giving you high confidence in the average even before seeing all five individually. Because the camera lacks HDR, auto-framing, and other active processing stages, the video path latency stays consistent from clap to clap, which is why the spread across five readings is typically narrower than on more feature-rich cameras.

After the test, copy the displayed OBS Audio Offset value and paste it into your streaming app's audio sync or delay control. In OBS Studio, open Advanced Audio Properties for the microphone source and use the Sync Offset field in milliseconds.⁵ In Streamlabs Desktop, open Advanced Audio Properties from the Audio Mixer when you need source-level audio controls, then verify the result with a short recording.⁶

Verifying the correction

Record a short clip after applying the offset, then play it back and confirm the clap sound arrives at the frame of visible hand impact. If the image still leads the audio, increase the audio delay; if the sound leads the image, reduce the delay and test again. Watching the clip frame by frame in a video editor gives you the most precise way to confirm alignment, since stepping through individual frames at 30fps lets you see whether the sound and motion share the same 33 ms window.

Logitech's own Brio 300 specification lists RightLight 2 automatic light correction with no HDR or auto-framing entry³, confirming there is no active per-frame processing stage to add latency beyond the basic USB video path. That puts the Brio 300 on the same minimal-processing footing as the C920x, so expect an offset close to that tight baseline rather than the wider range HDR- or AI-processing cameras in this series show.

Before you stream, record the clip under the conditions you actually use and keep the same working distance the fixed-focus lens needs. The Brio 300's narrow five-reading spread means a single corrected clip should match the average closely, but only if the same USB port, microphone source, and browser stay in place. OBS Studio and Streamlabs Desktop both expose the Sync Offset field in milliseconds, so enter the captured value in whichever app applies the correction. If the clip shows the sound and motion sharing that 33 ms window, the correction is holding. Start your session with confidence.

Logitech MX Brio 4K A/V Sync & Latency Test

Start with the production settings you actually use, then measure the MX Brio from that same state. MX Brio is Logitech's 4K Ultra HD collaboration and streaming webcam, with 4K at 30fps or 1080p at 60fps from its image sensor.¹ Logitech describes AI image enhancement, Show Mode, and advanced controls through Logi Options+ or G HUB, so disable variable framing or presentation features before you measure a stable A/V offset.

This tool compares local video and audio transients after the browser grants the requested media tracks, then reports the offset for the resolution and device path you actually used. Use the five-clap result from your own USB topology and app settings rather than a generic MX Brio offset figure.

Specifications²

Show Mode and AI framing off before testing

Disable Show Mode and any variable AI framing features before opening this page. Those features can change crop, lighting, or enhancement behavior between claps, which makes the five readings less consistent. The MX Brio's AI image enhancement pipeline can also adjust sharpness and noise reduction on a per-frame basis, so a variable scene with moving hands and changing light creates a less predictable processing load than a static scene.

After granting camera and microphone access, verify the preview resolution and frame rate before you start; the reported offset applies to that exact browser stream.³ Connect the MX Brio to a USB port with enough available bandwidth for the resolution you want to test. USB bandwidth is shared across the host controller and device tree, so another high-bandwidth device can change what the camera can sustain during the test.⁴

The MX Brio is a USB video device, and Windows can use the system UVC driver for compliant USB video devices instead of a Logitech driver.⁵ Close conferencing apps or other camera-control software before testing if the preview stutters, drops frames, or shows a different resolution than expected.

Matching the production stream profile

Keep the browser preview on the same resolution and frame rate you intend to use in OBS or Streamlabs. A 4K measurement can differ from a 1080p/60 measurement because the camera, USB path, and browser processing path are no longer identical. If you switch between 4K and 1080p during the week, measure both profiles and label the OBS values separately.

AI image enhancement and reading spread

AI image enhancement is the main feature to control on the MX Brio. With variable enhancement active, processing can change from frame to frame, which can widen the spread between clap readings. Disable Show Mode and variable framing features before testing so the camera stays in a consistent state. The MX Brio's 4K sensor captures significantly more data per frame than a 1080p camera, and when AI enhancement processes each frame at full resolution, the per-frame processing time can vary depending on scene complexity, introducing small timing fluctuations that show up as wider spread across your five clap measurements.

Frame drops are the other common issue to watch. They can come from USB bandwidth pressure, another app holding the camera, or power-management behavior on the host. The five-reading average helps expose inconsistent sessions, but it cannot turn a dropped-frame clap into a reliable timestamp. If one reading sits far outside the cluster, close camera-control software, keep the same USB topology, and rerun the test.

Outlier readings and how to handle them

Outlier readings usually mean the clap was not measured under the same conditions as the other four. A dropped frame, a sudden exposure change, or another app changing the camera mode can move one timestamp away from the rest. Retest before copying the offset into OBS or Streamlabs when the readings do not cluster tightly. Comparing the spread between your five readings against the average tells you whether the outlier pulled the final number away from the true offset for your setup.

What the MX Brio average applies to

Five clap pairs produce an averaged millisecond offset and a direction label. The value applies to the exact conditions you measured: same resolution, same machine load, same USB port, and same camera settings. Changing any variable requires re-measurement. Because the MX Brio supports multiple resolution and frame-rate combinations, each combination can produce a different offset; a 4K/30fps measurement should not be applied to a 1080p/60fps production scene because the camera's internal processing path changes between modes.

OBS source timing is based on audio and video timestamps, so the sync value you enter should match the measured browser stream conditions as closely as possible.⁶ Enter the value in OBS under Edit > Advanced Audio Properties > Sync Offset (ms) for your microphone source, then record a short test clip to verify the clap sound aligns with the visible hand impact.

Rechecking after Logitech app changes

Re-running the test after changing Logi Options+, G HUB, or camera firmware settings confirms whether the new settings changed the offset. Open the relevant Logitech app after each update, check the MX Brio image and framing features, disable any variable features you do not want during production, and run the five-clap test again to get the updated offset. A before-and-after comparison of the two measured values quantifies exactly how much the Logitech software change shifted your pipeline, which helps you decide whether the new settings are worth the sync trade-off.

Logitech markets the MX Brio around AI-powered face-based image enhancement that runs continuously as you present¹, a heavier per-frame processing load than a fixed-focus, non-HDR camera in this series carries. Expect the MX Brio's typical offset to sit closer to other active-processing cameras than to the C920x's minimal-pipeline baseline.

Because the offset only holds for the conditions you measured, keep in mind the 4K-versus-1080p60 offset gap on the MX Brio and start from the production state you intend to keep. The value you capture applies to the exact resolution, machine load, USB port, and camera settings you used, so re-running after a firmware or app change protects the number you paste into OBS. A 4K/30fps reading should never be applied to a 1080p/60fps scene, because the MX Brio's internal processing path differs between those modes. Keep the before-and-after numbers together so you can see how much a single setting shift moved your pipeline. Measure once per mode.

Logitech C922 Pro Stream A/V Sync & Latency Test

For C922 Pro Stream sync work, the capture mode matters as much as the camera model. C922 Pro Stream is a streaming webcam rated for 1080p/30fps and 720p/60fps, with a 78° diagonal field of view and USB 2.0 connection.¹ Logitech's support specs list UVC 1.1 support, a tripod mounting option, and 720p/60fps as the camera's maximum high-frame-rate mode.

The key variable for streamers is which resolution the camera runs during production. At 720p/60fps the camera delivers frames twice as often as at 1080p/30fps, reducing the inter-frame window between the clap motion and the first frame that captures the peak luminance. Test in the same resolution your OBS scene uses, because the browser-delivered track settings and camera mode affect the measured offset.

Specifications²

Port choice for stable C922 readings

Connect the C922 Pro Stream directly to a USB-A port on your computer. Sharing a USB hub with another high-bandwidth device can increase contention on the same USB topology, which can widen the spread between readings. Grant camera and microphone access in Chrome or Edge; the Media Capture and Streams API requests local media tracks from the browser and exposes the final width, height, and frameRate settings on the delivered tracks.³ The C922 can run through Windows' system UVC driver, so basic camera access does not require a Logitech-specific driver.⁴

For 720p/60fps testing, keep the camera mode and browser-delivered track settings the same when you compare readings. This tool measures the stream that the browser actually delivers, so test in the same resolution and frame rate your OBS scene captures. The C922's MJPEG compression at 720p/60fps uses more USB bandwidth than 1080p/30fps, which means the camera is more sensitive to USB controller contention at the higher frame rate; a port that delivers clean 1080p/30fps readings may show wider spread at 720p/60fps if other devices share the same controller.

Choosing the right USB port for stable C922 readings

For the most stable five-reading cluster, connect the C922 directly to a USB-A port on your computer instead of a shared hub with other active USB devices. Microsoft documents USB bandwidth as a shared resource across the host controller and device tree, with hubs and other clients consuming part of the available bandwidth.⁵ On laptops, moving between ports may not change the underlying internal hub, so focus on keeping the topology and competing devices consistent while measuring.

720p60 MJPEG bandwidth on a shared controller

MJPEG is a UVC compressed video format, and the C922's 720p/60fps mode uses more bandwidth than lower-frame-rate modes.⁴ If multiple USB video devices are active on the same host controller, they compete for the same USB bandwidth pool. Moving the C922 to a port that maps to a different controller can reduce contention-related variance, but the important test is to keep the same USB topology while measuring.⁵

Frame-rate matching is the other common sync issue. OBS exposes Resolution/FPS Type and FPS settings for video capture devices, and the source will not display if the selected frame rate is unsupported by the device.⁶ Set OBS capture to the same camera mode you test here, then use the five-clap result for that mode. A mismatch between the browser test frame rate and the OBS capture frame rate is one of the most common reasons a measured correction fails to fix the problem in the final recording, so double-check both settings before entering the value.

Matching OBS mode to your test result

Match the OBS capture mode to the browser-delivered track settings before you trust a sync offset. If the C922 delivers 720p/60fps in the browser but OBS records 1080p/30fps, the camera may use a different internal path and the measured correction can shift. Keep the mode fixed, measure, enter the offset, then verify with a short recording.

Long recordings can still drift if audio and video clocks run at slightly different rates, but that is a separate issue from the static offset this tool measures. The clap test gives you the initial correction for the current browser stream; matching your capture settings keeps that correction relevant across your recordings. Sample rate mismatch is the usual culprit behind progressive drift, so set all audio devices to the same rate, typically 48 kHz, before relying on a static correction for sessions longer than 30 minutes.

Browser offset versus the OBS mode you stream in

The clap test measures the browser-level offset: the time gap between a luminance transient in the video stream and an amplitude transient in the audio stream, both sampled against the same browser clock. At 1080p/30fps on the C922, this offset reflects the camera mode, browser stream settings, and USB scheduling added to the video path.

Measure the C922 in the mode OBS actually captures, because OBS aligns sources by their timestamps and a 720p60 measurement says little about a 1080p30 scene.⁷ Enter the result in the microphone's Sync Offset field, then confirm it with a short test clip.⁸

Multi-camera OBS scenes

When you use the C922 as one of several cameras in a multi-camera OBS scene, measure the C922-to-microphone offset separately from any other webcam-to-microphone offsets in your rig. Each camera has its own pipeline latency, and the Sync Offset field applies to the selected audio source for the active scene. Isolating the C922's offset in a multi-cam rig means re-measuring after any USB configuration change, driver update, or change in OBS capture resolution to keep the correction current.

Razer Kiyo A/V Sync & Latency Test

The Razer Kiyo needs a steady lighting state before the clap test can be trusted. Kiyo ships with a concentric ring light that encircles the 1080p lens.¹ For A/V sync calibration, the ring light affects the luminance detector more than on cameras without a built-in light source. On-axis illumination from the ring fills your face evenly, which means the clap luminance change is smaller relative to the bright ambient fill. The clap hands reflect ring light, so the per-frame luminance delta from the hand motion is reduced compared to a less-lit room.

Set the ring light to a medium brightness before running the test. At maximum brightness, the high ambient fill reduces the clap contrast enough to cause intermittent missed detections. At zero brightness, the darker scene causes the camera's auto-exposure to raise analog gain, increasing noise and reducing clap sharpness. Medium ring brightness with sharp claps at arm's length gives the most consistent five-reading results.

Specifications¹

Ring light setting before you grant permissions

Connect the Kiyo directly to a USB-A port before opening this page. Adjust the ring light dial to a medium setting, roughly 50% of maximum brightness. Grant camera and microphone permissions in Chrome or Edge when prompted.² The camera appears in the video input dropdown as "Razer Kiyo" on Windows and macOS.

The Kiyo uses autofocus rather than fixed focus, so sit at your normal working distance and let the camera settle before testing. If the clap motion looks soft, move closer or adjust the camera position until your hands are sharp in the preview. Soft images weaken the clap luminance change, which can make missed detections more likely.

Set the ring light to your streaming brightness before opening the test page and leave it at that level throughout the measurement. Running the test at a different brightness than your actual stream produces an offset measured under different exposure conditions. The ring light changes the camera's exposure time: at maximum brightness the exposure shortens, at zero brightness the exposure lengthens. Exposure time affects how sharp the clap luminance transient appears in each frame, and a sharp transient gives the detector a more accurate timestamp at the frame boundary.

Ring brightness before browser permissions

Keep the ring at the same brightness while granting browser permissions and collecting all five readings. If you adjust the dial after the first clap, the browser may capture a different exposure state, and the average will no longer describe one stable camera mode. Writing down the ring brightness percentage alongside your measured offset gives you a complete picture of the test conditions, making it easier to reproduce the same result in future sessions.

If your ring light gets bumped mid-test, stop and restart the five-clap sequence from the beginning. Even a small brightness change alters the camera's automatic gain, which shifts the exposure baseline and can average into a value that does not represent the streaming setup you actually plan to use live.

Ring brightness and clap detection

Ring light brightness interacts directly with clap detection reliability. At full ring brightness, the evenly lit scene provides minimal luminance contrast for the detector when hands enter and exit the frame center. Moving your hands to clap higher than your face (above the ring's illumination zone) increases the luminance change from the hand shadows, improving detection. Yet moving higher introduces the risk of clapping outside the optimal camera frame for luminance measurement.

Consequently, the most reliable calibration posture with the Kiyo is: ring at 50% brightness, clap directly in front of your chest at arm's length from the camera, with your hands entering from the sides rather than from above. This clap path maximizes the frame-to-frame luminance delta the detection algorithm measures, and practicing the motion two or three times before starting the five-clap pass helps you hit the same position consistently for each reading.

For the most reliable clap detection with the Kiyo's ring light active, position yourself so the ring light illuminates your face and chest evenly. Clap at chest height about arm's length from the camera, with hands entering the frame from both sides rather than from above or below. This path creates the largest possible luminance change in the center of the frame, where the luminance detector applies the highest weight. Clapping too close to the lens increases motion blur at the frame boundary and can overwhelm the sensor with a brief overexposure flash that the detector mistakenly timestamps at the wrong frame edge.

Ring brightness and clap detection reliability

Use the same ring brightness you plan to stream with, but avoid the extremes if the detector misses claps. Full brightness can flatten the hand shadow, while zero brightness can force the camera into a slower exposure. If your readings are inconsistent, set the ring to a middle position, clap sharply at chest height, and rerun the five-clap pass before copying the OBS value.

Why the ring light does not change the Kiyo offset

On the Kiyo, the measured offset reflects the active 1080p/30fps MJPEG pipeline and your selected microphone path. Because the ring light does not change the USB or ISP pipeline latency, only clap detection reliability is affected by the light level, not the final averaged offset value once five valid detections complete.³

Copy the OBS Audio Offset value after the test and enter it in OBS Studio under Edit > Advanced Audio Properties.⁴ Building on this, if the Kiyo is your only light source in a dark room, the ring light is essential for stream quality; calibrate with the ring at your streaming brightness so the measured offset matches your production conditions.

Applying the OBS correction at streaming ring brightness

Once you enter the measured OBS Audio Offset value, verify the correction by recording a 10-second test clip with the ring light at your streaming brightness. Play back the clip and look for the clap you performed during calibration. At the correct sync offset, the clap sound arrives in the same frame as the visible hand impact. Entering the correct value in OBS Advanced Audio Properties helps eliminate the lip-sync gap that viewers would otherwise notice during your stream.

Independent review testing confirms the Kiyo works as a plug-and-play webcam with no driver or software required for basic operation; Razer Synapse only adds optional brightness, contrast, and color-profile adjustments on top of that baseline path.⁵ Leaving Synapse's color adjustments off is why the Kiyo's offset tracks closer to a basic MJPEG camera's baseline than to a camera whose active processing runs by default.

Razer Kiyo Pro Ultra A/V Sync & Latency Test

A large-sensor webcam can look excellent and still need a specific sync correction on your machine. Kiyo Pro Ultra houses a 1/1.2-inch Sony STARVIS 2 sensor and an f/1.7 aperture lens, with Razer positioning it around large-sensor webcam capture and HDR support.¹ That physical size gives more low-light headroom, but it also means each frame carries more data through the camera processor and USB link. In practice, the Kiyo Pro Ultra should be tested in the same capture mode you plan to stream with, because HDR, Synapse processing, frame-rate, and exposure settings can all change the production offset.

Measure the Kiyo Pro Ultra in your actual capture configuration instead of relying on a generic offset range. The useful number is the one you get with your browser, USB connection, Synapse profile, lighting, and OBS frame-rate settings active.

Specifications¹

USB 3.0 port and Synapse profile

Connect the Kiyo Pro Ultra to a USB 3.0 port before opening the test page. Grant camera and microphone permissions in Chrome when prompted; the browser Media Capture API is designed to request local audio and video devices from the platform.² USB video devices can also work through the system UVC driver without a proprietary camera driver.³

For the most accurate sync measurement, close Razer Synapse while you capture a clean baseline, then reopen it and configure the Kiyo Pro Ultra's Camera and Processing settings before testing your production setup. Razer's Synapse support workflow exposes those camera controls, including zoom and lens-distortion processing options, so the baseline and production runs should be treated as two different configurations.⁴

Synapse profile changes that shift the offset

Do not mix the clean baseline with the production profile when copying values into OBS. If the baseline is measured with Synapse closed and your stream uses Synapse HDR or lens processing, the baseline number will not represent the live setup. Run a second five-clap test after enabling the production profile, then use that second result for the stream. The difference between the baseline and production values tells you exactly how much Synapse processing adds to your video path, which is useful information when diagnosing future sync issues after software updates.

Synapse image settings and varying readings

Synapse image settings are the first things to check when Kiyo Pro Ultra readings vary. The product is designed around large-sensor capture and Synapse-driven customization, so your measured offset should match the exact profile you stream with rather than a default lab-style setup. Small changes to sharpness, saturation, or lens-distortion processing in Synapse can shift the video pipeline timing enough to move the offset by several milliseconds, which is why the production profile always matters more than a generic baseline.

The second common issue is changing light. The Media Capture and Streams specification notes that source conditions can change, such as when a camera switches to a lower frame rate because of low light, and that can affect the timing window used by the detector. Set the room to consistent, bright lighting before testing so the measurement reflects your capture pipeline instead of exposure changes. The Kiyo Pro Ultra's large 1/1.2-inch sensor gathers more light than smaller webcam sensors, which helps in dim rooms, but the camera's auto-exposure algorithm can still introduce frame-to-frame brightness adjustments that soften the clap transient if the lighting is uneven or slowly shifting.

Lighting changes that affect clap detection

In a dim room, add a desk lamp facing you before running the test. Brighter ambient light lets the sensor use a faster shutter speed, which means the clap motion appears in a shorter exposure window and creates a sharper luminance edge at the frame boundary. Sharper edges make individual clap detection more reliable and tighten the spread across five readings. Return to your streaming light setup after measuring and re-enable HDR if your production uses it.

Frame timing at 30 and 60 fps on the Kiyo Pro Ultra

The luminance transient detector samples each video frame drawn to a canvas element. At 60fps, each frame represents about 16.7 ms; at 30fps, each frame represents about 33.3 ms, so the detector has a smaller timing window when the capture path is running at 60fps.⁵ This makes the five-clap average more useful when your browser and OBS scene are both using the same frame-rate target.

Building on this, the tool reports the average offset across your five clap pairs. OBS lets you choose the resolution and frame-rate combination for a Video Capture Device source, so set that source to the same capture mode you use in your scene before copying the result into production.⁶ For the Kiyo Pro Ultra used as an OBS webcam source, enter the displayed OBS Audio Offset value in Edit > Advanced Audio Properties for your microphone; the Sync Offset column accepts a delay in milliseconds.⁷

Re-measuring after changing capture settings

The measured offset is specific to the capture configuration you tested with. HDR, Synapse processing settings, lighting, browser permissions, and OBS frame-rate settings can all change the production value, so re-run the five-clap test any time you change one of those settings. Update the OBS Sync Offset value to match the new measurement immediately, before starting a stream or recording session that uses the changed configuration. An outdated offset produces visible lip-sync gap from the very first minute of the session.

Independent review testing found that enabling HDR recording on the Kiyo Pro Ultra measurably reduces its frame rate, confirming the Synapse-controlled processing options change the camera's real-time performance rather than just its stored image settings.⁸ That heavier, configurable processing load is why this camera's typical offset runs higher than a basic fixed-focus, non-HDR webcam in this series.

Match the capture mode to your OBS scene before you trust the number. At 60fps each frame is about 16.7 ms and at 30fps about 33.3 ms, so the detector works inside a tighter timing window at the higher frame rate and the same HDR or Synapse change can read differently across modes. Set the Video Capture Device source in OBS to the exact resolution and frame rate you measured, then paste the offset into the Sync Offset column. Re-checking how enabling HDR changes the Kiyo Pro Ultra's frame rate after any single setting change keeps the first minute of your stream clean. Update the value before you go live.

Elgato Facecam A/V Sync and Latency Test

Use your real streaming preset before copying any Facecam sync value into OBS. Facecam delivers 1080p at a native 60fps, and Elgato's Camera Hub software applies color correction, contrast, and sharpening per frame.¹ At 60fps, the detector receives a new frame every 16.7 ms, which keeps the clap transient inside a shorter inter-frame window than a 30fps camera.² The five-clap average gives you a practical OBS sync value for your current Camera Hub preset and microphone setup. Measured offsets vary by machine, preset, and Camera Hub state, so use your production configuration when you record the final value.

Specifications³

USB 3.0 for uncompressed 1080p60

Connect the Facecam to a USB 3.0 port before opening this page. Facecam uses uncompressed 1080p/60fps video, and USB-IF defines USB 3.2's inherited 5Gbps tier as part of the USB 3.x family, which gives high-bandwidth cameras more headroom than USB 2.0.⁴ Grant camera and microphone permissions in Chrome when prompted.

If Elgato Camera Hub is running, test with your production preset active, then close it and rerun the page when you need a cleaner baseline. Camera Hub applies its image processing pipeline in software on the host computer rather than in the camera firmware, so the processing load on your CPU can affect frame delivery timing; a busy system may show slightly different offsets with Camera Hub active versus closed.

Camera Hub baseline versus production preset

Closing Camera Hub entirely before the test gives you a baseline that reflects the camera hardware, USB transfer, and browser decode stage. Reopen Camera Hub with your streaming preset active for the production value, then compare the two readings to see how much your image pipeline changes the offset. Write down both values if you use both a baseline and a streaming preset, because the difference tells you whether a future update or preset change shifted the Facecam path.

OBS verification after applying Facecam offset

After entering the OBS Audio Offset value, record a short test clip and clap once near the center of the frame. Play the clip back and confirm that the clap sound lands on the frame where your palms meet. If the sound still leads the image, increase the Sync Offset by 5–10 ms; if the sound trails the image, reduce it by the same amount.

Camera Hub presets and the Facecam offset

Camera Hub presets can change Facecam offset measurements because each preset applies different image processing. Elgato lists Camera Hub controls for saturation, sharpness, contrast, shutter speed, exposure, and color temperature, so confirm which preset your OBS scene uses and test with that preset active. A heavy processing preset with multiple adjustments enabled can add a few more milliseconds of host-side computation than a flat uncorrected profile, which is why the active preset should always match the scene you stream with.¹

The other key setting is OBS capture frame rate. OBS lets you choose the resolution and framerate combination for a video capture device, and it will not display the source if the device does not support the selected frame rate.⁵ Match OBS capture to the Facecam's native 60fps output to keep the timing window consistent with the browser test. Running OBS at 30fps while the Facecam delivers 60fps means OBS drops every other frame, and the specific frame the browser measured at 60fps may be one that OBS discarded, making the correction unreliable for the recorded output.

OBS capture frame rate matching Facecam native output

The Facecam delivers 1080p/60fps natively. At 30fps capture, the inter-frame interval is 33.3 ms instead of 16.7 ms, which widens the timing window for clap detection.² Set your OBS video source capture frame rate to 60fps to match the Facecam's native output and keep the browser measurement aligned with your recording pipeline. Running the browser test at 60fps while OBS captures at 30fps introduces a mismatch that makes the measured offset unreliable for your production setup.

While you measure, keep the Camera Hub preset and capture frame rate fixed at the values you stream with. The Facecam's native 60fps delivery puts a new frame in front of the detector every 16.7 ms, so the clap transient sits inside a shorter inter-frame window than on a 30fps camera. Set the OBS video source to 60fps before you paste the value, because a 30fps capture drops every other frame and discards the one the browser timed. Rerun the test before going live after any Camera Hub or firmware update so the OBS value matches your current pipeline. Test again after every update.

Clap timing at 60 frames per second

At 1080p/60fps, the video path samples the clap at a higher temporal resolution than 30fps cameras because each frame is spaced 16.7 ms apart. The five-clap average helps expose outliers and gives you a practical value for your current production pipeline. Because the Facecam's uncompressed video stream demands more USB bandwidth than a compressed MJPEG camera, the measurement is more sensitive to USB bus conditions; a clean USB 3.0 connection with no competing devices produces the most reliable five-reading cluster.

The OBS Audio Offset value in the result panel applies to Camera Hub, your OBS capture frame rate, and your microphone. Enter the value in OBS under Edit > Advanced Audio Properties for your microphone source; the Sync Offset field accepts audio delay in milliseconds, and 1000 ms equals one second.⁶

Because the Facecam runs at 60fps, the five-reading average should be more stable than a 30fps measurement when your capture settings match the camera output. If Elgato releases a firmware or Camera Hub update between sessions, rerun the test before going live so the OBS value reflects your current pipeline.

Independent review testing found the Facecam's array of shutter speed, ISO, and color adjustments in Camera Hub give it a broad active-processing feature set even at 1080p, which reviewers noted alongside its generally strong latency reputation compared to other webcams in its class.⁷ Weighing that active Camera Hub processing against a bare C920x baseline explains why the Facecam's measured offset sits noticeably higher even without HDR in the mix.

Elgato Facecam Pro 4K A/V Sync and Latency Test

The Facecam Pro offset belongs to your exact capture state, not the camera box. Facecam Pro captures 4K/60fps from a 1/1.8-inch Sony STARVIS CMOS sensor, and Camera Hub lets you control exposure, white balance, focus, contrast, saturation, and other image settings.¹ If your browser or app requests 1080p, the camera can deliver a 1080p stream while still using the full sensor workflow; the W3C Media Capture spec says user agents may use downsampling to satisfy track constraints.² At 60fps, each frame is spaced 16.7 ms apart, so the five-clap average gives you a practical sync value for the exact resolution, Camera Hub preset, cable, and port you test.³ Measured offsets vary by machine and configuration, so use your production setup when you record the final OBS value.

Specifications⁴

USB 3.0 Type-C and Camera Hub for a 4K baseline

Connect the Facecam Pro to a USB 3.0 or better Type-C port before opening the test page. USB-IF identifies 5Gbps as one USB 3.x transfer tier, which gives high-bandwidth 4K video more headroom than USB 2.0.⁵ Close Elgato Camera Hub during the test if you want a baseline measurement, or leave it open with your production preset active if you want the production-accurate offset.

Grant camera and microphone permissions when Chrome prompts. The Facecam Pro registers as "Elgato Facecam Pro" in the dropdown. If two entries appear, select the primary hardware entry rather than any virtual device registered by Camera Hub. The virtual device adds an extra software layer between the camera and the browser, which can introduce additional latency that does not reflect the hardware camera path OBS captures directly.

Run the test twice if you want both a baseline and a production offset. For the baseline, close Camera Hub before measuring. For the production offset, open Camera Hub, select your streaming preset, and run the five-clap test again. The production offset is what you enter in OBS. Keeping both values in a note lets you diagnose future changes when you change Camera Hub settings, resolution, cable, or port.

4K baseline versus browser-delivered stream

The baseline value is useful for diagnosing the camera path, but it may not match what OBS receives if OBS is configured for 4K capture while the browser defaults to 1080p. The value you use for live production should come from the same resolution and Camera Hub profile you intend to stream with. When you keep a side-by-side record of baseline and production offsets, you build a reference that makes future troubleshooting faster because you can isolate whether a sync shift came from the camera hardware or from the software processing layer.

4K processing and the capture configuration

4K image processing is the main reason to measure the Facecam Pro in its actual capture configuration. Elgato lists 4K60 output, H.264 and MJPEG formats up to 2160p60, and Camera Hub controls for focus, exposure, white balance, and image effects, so your preset and resolution can change the pipeline you are measuring.¹

The other common issue is USB bandwidth contention. If other high-bandwidth devices share the same controller, frame delivery can become less consistent and the five-reading spread can widen. Disconnect unnecessary USB devices and retest on a direct USB 3.0 or better port if the spread looks unstable. The Facecam Pro's 4K/60fps uncompressed stream requires substantially more bandwidth than a 1080p camera, so even a single external SSD on the same USB controller can introduce enough contention to widen your five-clap spread by 10 to 20 ms.

USB bandwidth and laptop port conflicts

Laptop USB ports often share bandwidth with other internal devices. When a dock, external drive, capture device, or another high-bandwidth peripheral competes for the same controller, the Facecam Pro may show more jitter in the five-clap readings. Disconnect competing peripherals during the measurement and reconnect them after capturing the OBS value. If your laptop has both USB 2.0 and USB 3.0 ports, always choose a USB 3.0 port for the Facecam Pro because the higher bandwidth tier reduces the chance of frame delivery contention affecting your clap timestamps.

Measuring the exact 4K setup you stream with

Five clap pairs measure the video-to-audio offset under your current cable, port, resolution, and Camera Hub configuration. The Facecam Pro's 4K pipeline makes it especially important to measure the exact setup you stream with. Because the camera processes 4K frames through its image engine before delivering them over USB, the processing time per frame is higher than a native 1080p camera, and any change in resolution or Camera Hub settings can shift the offset by a noticeable amount.

Copy the OBS Audio Offset value and enter it in OBS Studio under Edit > Advanced Audio Properties for your microphone. The Sync Offset field accepts audio delay in milliseconds, and 1000 ms equals one second.⁶ After entering the OBS sync offset, record a short test clip and clap once near the middle. OBS exposes resolution and FPS settings for video capture devices, so keep your test resolution aligned with your OBS source.⁷ At 60fps, one frame is 16.7 ms, so use frame-by-frame playback to check whether the clap sound and visible hand impact are aligned.³ If the alignment is off, adjust the OBS Sync Offset value in small increments, record another clip, and keep the value that matches your production capture settings.

Verifying Facecam Pro alignment frame by frame

Frame-by-frame playback matters most with Facecam Pro because 4K/60fps makes sync errors easier to spot. If the clap sound lands one frame early or late after applying the measured offset, adjust the OBS Sync Offset by 10 to 15 ms and record again until the sound and visible impact share the same frame. At 60fps each frame represents only 16.7 ms, so even a single-frame misalignment is visible to attentive viewers during slow-motion playback, making this level of precision worthwhile for professional streams.

Independent review testing confirms the Facecam Pro is the first webcam to capture natively at 4K/60fps, with Camera Hub software handling exposure and color adjustments the sensor itself does not apply automatically.⁸ That combination of a larger 4K/60fps frame and active Camera Hub processing is why this camera's typical offset runs higher than the standard 1080p Facecam.

At the resolution you actually stream with, record the check clip after confirming why your browser caps the Facecam Pro at 1080p by default. OBS exposes resolution and FPS settings for its video capture devices, so keep the test resolution aligned with your OBS source before you paste the offset. The Facecam Pro's 4K pipeline makes the exact setup you stream with the only valid one to measure, because a change in resolution or Camera Hub preset can shift the offset by a noticeable amount. Adjust the Sync Offset in small increments if the clip is still off, and keep the value that matches your production capture settings. Measure once per setup.

EMEET Nova 4K A/V Sync & Latency Test

Confirm the active resolution before you trust any Nova 4K sync number. EMEET Nova 4K is a 4K USB-A webcam with PDAF autofocus, dual noise-cancelling microphones, and a 73° field of view. It can default to 1080p on some systems, so confirm the active resolution before you treat the offset as production-ready.¹ Measure the camera-to-microphone pair you actually use, because the built-in mic and an external USB mic follow different audio paths.

The five-clap test gives you a browser-level offset for the active camera mode, audio endpoint, and USB topology. Use that measured value instead of a generic Nova 4K offset, because resolution, app processing, and the selected microphone can change the timing path.

Specifications¹

USB-A topology and permissions for the Nova 4K

Connect the Nova 4K to a USB-A port and keep the same USB topology for all five readings. The Nova 4K is documented as a 4K/30fps PDAF webcam with a 73° field of view and USB-A connectivity.² Grant camera and microphone permissions in Chrome when prompted; the browser receives local media tracks only after you approve the request.³

Select the camera video input and choose either the Nova 4K's built-in mic or your external microphone in the audio input field, depending on which pairing you use for streaming. If the same camera is moved to a different USB controller, hub, or port, run the five-clap test again because USB bandwidth is shared across the host controller and device tree.⁴ The Nova 4K's USB-A connector means it can plug into any standard port, but on machines where all USB-A ports share a single internal host controller, moving the cable between ports may not actually change the bandwidth situation.

The Nova 4K works as a UVC device with the system-supplied driver, so the browser can access the camera without a vendor-specific driver.⁵ If you use EMEETLINK or another processing app, measure with the same video and audio settings that your streaming setup will use. Running the test with the full software chain active, including any vendor processing app, keeps the measured offset honest about the total delay your actual pipeline introduces from capture to browser delivery.

Built-in mic versus external mic pairing

The microphone selection changes the audio side of the measurement. If you stream with a USB microphone, select that microphone during the test and use that value in OBS. If you stream with the Nova 4K's built-in microphone, select the Nova 4K audio entry instead. Do not measure the built-in mic and then apply the result to an external mic path; those are two different audio endpoints.

If you switch microphones between claps, the reading toggles between the built-in mic and the external mic path. Each path has a different delay, so the average no longer represents either endpoint cleanly. Stick to the same input for all five claps and use that specific offset entry in OBS.

Verifying the Nova 4K OBS correction

After entering the OBS Audio Offset value, record a short clip and clap once near the center of the frame. The clap sound should land on the visible hand impact. If the sound leads the image, increase the Sync Offset by 5 to 10 ms; if the sound trails the image, reduce it and test again. Repeating this verify-and-adjust cycle two or three times typically brings the residual error below one frame at 30fps, which is the practical target for production streaming.

Capture mode and microphone choice change the offset

The Nova 4K's image and audio paths are sensitive to the active capture mode. A 4K stream, a 1080p stream, the built-in microphone, and an external USB mic can each produce a different offset, so keep those settings fixed while you collect the five readings. Changing any of these variables between claps introduces a new variable into the measurement, which makes the five-reading average less meaningful as a correction value for your specific streaming configuration.

If the readings jump around, check the active resolution, USB connection, and audio endpoint before changing OBS. The five-reading average is useful only when each reading comes from the same camera mode and the same microphone path. Re-run the test after switching resolution, moving the USB cable, or changing the selected audio device. The Nova 4K's PDAF autofocus system can also introduce small timing variations if the focus position shifts between claps, so allow the autofocus to settle before starting the test and avoid moving forward or backward during the five-clap sequence.

Interpreting a wide five-reading spread

A wide spread usually means the measurement conditions changed between claps. Look for autofocus movement, resolution changes, background CPU load, or an audio endpoint switch before you treat the average as final. If the spread persists, record two five-reading sessions and compare their averages before copying one value into OBS. A consistent average across two separate sessions gives you much higher confidence that the offset reflects your true hardware timing rather than a single anomalous reading influenced by a momentary CPU spike or USB contention event.

What the Nova 4K average represents

Five clap pairs average the video-to-audio offset for the active Nova 4K capture mode and the selected microphone. The average is useful when the camera resolution, USB connection, and audio endpoint stay the same across all five readings. Because the Nova 4K can default to 1080p on some systems even though it is a 4K camera, always verify the active resolution in the browser preview before trusting the measurement; a 1080p browser stream and a 4K OBS capture represent two different processing paths through the camera.

Building on this, the OBS Audio Offset value displayed reflects either the external mic or the built-in mic pair you selected. The two pairings can produce different offsets, so use the measurement that matches your streaming setup.⁶ The Nova 4K's built-in dual noise-cancelling microphones follow a different internal audio path than an external USB microphone, and the built-in mic signal passes through the camera's own audio processing before reaching the browser, which can add a few milliseconds compared to a direct USB microphone connection.

Keep two OBS Sync Offset entries in your notes if you switch between the Nova 4K's built-in microphone and an external USB microphone. Label each saved offset with the pairing it came from, then verify the correction with a short recording before going live. If you use the built-in mic for quick calls and an external USB mic for streaming, having both values saved means you can switch between setups without re-measuring each time.

EMEET's own product listing notes that the Nova 4K's video resolution defaults to 1080p, with 4K requiring a manual switch in camera software or in an app such as OBS.⁷ Ruling out the Nova 4K's silent 1080p fallback matters as much as closing EMEETLINK for a clean baseline, since either one alone can throw off the number you paste into OBS.

OBSBOT Tiny 3 A/V Sync & Latency Test

Start by deciding whether your stream uses the Tiny 3 as a fixed camera or an AI-tracked camera. Tiny 3 is a 4K AI-powered PTZ webcam built around a 1/1.28" 50MP CMOS sensor, with 4K at 30fps and 1080p at 120fps video modes.¹ Its AI Tracking 2.0, Voice Tracking, object tracking, and motorized gimbal can keep you framed while you move, but those features add moving parts to an A/V sync test.

For calibration, use the Tiny 3 as a fixed camera. Disable tracking and let the gimbal park before opening this browser test, because the browser receives the video stream that your selected camera mode is producing at that moment.² A stable setup gives you one number to enter in OBS instead of a reading that changes whenever the camera reframes during the clap.

Specifications¹

Parking the gimbal before you measure

Open OBSBOT Center and turn AI tracking off before connecting the camera to the browser. Once the camera is parked in a fixed position, close Center before loading this page so its controls do not change tracking, zoom, or framing while you measure. Connect the Tiny 3 with the USB-C to USB-A 3.0 cable or another USB-C 3.0 path, then grant camera and microphone permissions in the browser.³

The camera registers as "OBSBOT Tiny 3" in the browser video input dropdown. If the video dropdown shows more than one OBSBOT entry, keep the same camera mode for all five claps and re-run the test after any tracking, zoom, or resolution change. The browser test measures the active camera stream, not a separate Center preview.

Tracking-off calibration posture

After disabling tracking in OBSBOT Center, close the Center app entirely before loading the test page. This keeps the camera in the tracking-off state and prevents accidental zoom or preset changes during the measurement. With Center closed, the browser is the only application reading from the Tiny 3 camera for the calibration run, which gives you the cleanest signal path for the OBS value you copy from the result.

Tracking-on verification for live streams

If your actual stream uses Voice Tracking or object tracking, run a second five-clap pass with tracking active after recording the tracking-off baseline. The tracking-on value is the one to enter in OBS for AI-tracked broadcasts, while the tracking-off value remains useful for fixed-camera scenes. After applying the OBS offset, record a short clip with the same tracking mode and confirm the clap sound lands on the visible hand impact.

AI tracking as the calibration variable

AI tracking is the defining calibration variable for the Tiny 3. OBSBOT's tracking features can detect people or objects, use Voice Tracking, and move the gimbal to keep the subject framed. If tracking is active during the clap, the video transient can include both your hand motion and the camera's reframing motion, which makes the luminance detector harder to interpret.

Capture mode and Center app state

The second issue is the capture mode you leave active. OBSBOT Center can control imaging settings, avatars, multi-camera setups, and production tools, while the browser receives the active camera stream through the platform media API. Calibrate with the same tracking, zoom, and resolution settings you plan to stream with, then copy the resulting OBS value into the matching OBS Video Capture Device setup.⁴

During the five-clap run, keep the camera mode fixed. The browser test measures the active camera stream, not a separate Center preview, so any change in tracking, zoom, or resolution mid-run shifts the reading you copy into OBS. Close OBSBOT Center before loading the page so the gimbal stays parked and no preset change interrupts the measurement. Use the tracking-off value for fixed-camera streams and the tracking-on value for broadcasts where AI Tracking 2.0 stays active. Match the mode to the stream.

Tracking-off and tracking-on offsets

With tracking disabled, the Tiny 3 provides a fixed camera stream for the luminance detector. Five clap pairs produce an average offset that represents the camera-to-browser video path plus your selected microphone path. The result is a single millisecond value you enter as a negative Audio Offset in OBS, which shifts the microphone track earlier to align with the delayed video frames.

If you stream with AI tracking enabled, capture a second measurement with tracking active after recording the tracking-off baseline. The difference between the two readings tells you how much your particular tracking setup changes the measured offset. Use the tracking-off OBS value for fixed-camera streams and the tracking-on value for streams where you deliberately leave AI Tracking 2.0 active during the broadcast.⁵

Independent review testing found OBSBOT Center exposes extensive per-frame image processing, including separate manual curves for shadows, mid-tones, and highlights, on top of the gimbal-based AI tracking itself.⁶ Even with tracking off, OBSBOT Center's shadow and highlight curve processing keeps the Tiny 3 above a minimal-pipeline webcam during calibration.