TesterAudio Latency Tester
Free browser audio latency tester. Reports Web Audio base + output latency in ms, sample rate and buffer size, with CSV/JSON export and pass/fail check.
Have feedback? Report bugs, suggest features, or share your thoughts — we read them allAbout Audio Latency Tester
Audio Latency Tester helps you measure the delay between when audio playback is triggered and when sound actually plays through your speakers or headphones. This is crucial for music production (DAW recording), gaming (audio-visual sync), live streaming, video editing, and any application where timing is critical. Lower latency means better real-time responsiveness.
- Click Start Test to initialize the audio system and grant browser permissions
- The tool will run multiple tests automatically with audible beeps
- View real-time results including average latency and range
- Check the test log for detailed results of each measurement
- Run tests multiple times for more accurate results
Understanding Audio Latency
- Base Latency: Time between audio context and audio hardware buffer (the browser's reported estimate)
- Output Latency: Additional buffering delay in the output device — Safari does not expose this, so it shows N/A
- Total estimate: base + output latency in ms — this is the browser's estimate, not a true acoustic round-trip
- Buffer Size: derived frames = base latency x sample rate; smaller is lower latency but riskier
- Good latency: < 10ms (imperceptible), Acceptable: 10-30ms, High: > 30ms
- Lower buffer sizes reduce latency but may cause audio glitches
- Use headphones for more accurate testing vs. speakers with room delay
- Close background apps to reduce system load and improve latency
- Professional audio interfaces typically have lower latency than built-in sound cards
- Gaming and music production require the lowest possible latency
- Video streaming can tolerate higher latency (50-100ms is acceptable)
Frequently Asked Questions
Audio latency is the delay between when a sound is requested by software and when it actually reaches the listener's ear. This tester measures the round-trip path inside the browser: the time between calling the Web Audio API to schedule a sound and the moment the audio engine reports the buffer is ready for playback. It captures buffer size (in frames), sample rate (typically 44.1 or 48 kHz), and the computed latency in milliseconds. It does not measure the additional latency added by your DAC, USB audio interface, Bluetooth codec, or external speakers — those add their own delays which the browser cannot inspect. For full system latency, you would need to record the output with a microphone and compare timing.
Latency is invisible until it crosses your threshold of perception, then it becomes intolerable. Musicians playing virtual instruments need under 10 ms total round-trip or notes feel disconnected from key presses. Live monitoring during recording needs under 5 ms to avoid distracting "echo" between the performer's ears and headphones. Gamers benefit from under 30 ms because footsteps and gunfire need to align with visual cues for accurate positioning. Streamers and content creators need audio and video to stay in sync within 40 ms or viewers notice lip-sync drift. Video calls tolerate higher latency (100–200 ms) because conversation is bidirectional and humans naturally adapt, but anything over 250 ms causes talk-overs and conversational confusion.
For browser audio specifically: under 20 ms is excellent and suitable for any musical performance; 20–50 ms is good for media playback and most gaming; 50–100 ms is acceptable for passive streaming and video; above 100 ms will be noticeable as a delay between user actions and audio response. Be aware that Bluetooth headphones add 40–300 ms on top of whatever this tool reports — Bluetooth Classic A2DP can be 200 ms, aptX Low Latency reaches around 40 ms, AirPods Pro is roughly 130 ms but improves to 80 ms on Apple devices. Wired headphones add essentially zero latency. USB audio interfaces typically add 3–10 ms.
Modern browsers throttle background tabs to save battery and CPU. When this tab loses focus, the audio worklet thread may run at lower priority, the timer resolution drops from 1 ms to as much as 1 second, and buffer underruns become common. You may also see latency jump when the OS engages power-saving modes (laptops on battery, mobile in low-power mode), when another high-priority app like a video call claims the audio device exclusively, or when the audio driver renegotiates its buffer size in response to load. For consistent measurements, run this test on a fresh tab in the foreground, plug in your laptop, close other apps that use audio (Zoom, Spotify, Discord) and disable browser extensions that capture audio.
Buffer size is the number of audio samples processed per callback cycle. Smaller buffers mean less audio waiting in the queue and therefore lower latency, but they also mean more frequent callbacks and tighter CPU deadlines — miss one and you get an audible glitch (a crack or pop). At 48 kHz sample rate, a 256-sample buffer is 256/48000 = 5.33 ms one-way latency; a 128-sample buffer halves that to 2.67 ms but doubles the CPU pressure. Professional audio interfaces often run at 32 or 64 samples for live monitoring, while consumer cards default to 512–1024 samples for stability. The browser typically selects buffer size based on the device and current load — you cannot override it from JavaScript without renegotiating the AudioContext.
Bluetooth audio adds substantial latency through three layers: codec encoding, radio transmission, and codec decoding. The dominant factor is the codec. SBC (the universal default) typically adds 200–300 ms because it processes large frames and uses aggressive forward error correction. AAC (Apple's preferred) is around 150–200 ms with optimizations on iOS. aptX is around 70 ms. aptX Low Latency is around 32–40 ms but requires both transmitter and receiver to support it. aptX Adaptive and LC3 (Bluetooth 5.2 LE Audio) get down to 20–40 ms. Apple's AirPods use AAC with proprietary tweaks to reach about 130 ms on iPhones. To minimize Bluetooth latency: use wired connections when possible, pair devices supporting the same low-latency codec, and keep firmware updated.
Sample rate is how many audio samples per second the system processes. Higher rates capture finer temporal detail (Nyquist frequency = sample_rate / 2, so 48 kHz captures up to 24 kHz of audio) and reduce the per-sample buffer time linearly: at 48 kHz a 256-sample buffer is 5.33 ms, at 96 kHz the same buffer is 2.67 ms. However, doubling sample rate doubles CPU load, increases file size, and beyond 48 kHz produces no audible difference for human listeners since 20 kHz is the upper hearing limit. Pro audio uses 96 kHz to give signal processing headroom for nonlinear plugins. Most consumer audio is 44.1 kHz (CD heritage) or 48 kHz (video heritage). Web Audio defaults to the device's preferred rate, usually 48 kHz on modern hardware.
Professional audio recommendations come from several sources. ITU-R BS.1116 specifies under 1 ms timing accuracy for double-blind sound quality tests. Live concert monitoring follows AES recommendations of under 10 ms for in-ear monitors and under 20 ms for floor wedges. ASIO drivers on Windows can achieve 1–5 ms round-trip with proper hardware. CoreAudio on macOS commonly achieves 2–10 ms. Browser Web Audio with the AudioWorklet API can reach 5–20 ms on desktop and 30–100 ms on mobile. The browser stack adds overhead from JavaScript execution, message passing between threads, and Chromium's media pipeline. For latency-critical professional work, native DAWs (Logic, Reaper, Pro Tools) remain the standard; browsers are catching up but not yet equivalent.
This tool reads AudioContext.baseLatency and AudioContext.outputLatency. Chrome and Edge expose both on desktop and Android. Firefox exposes baseLatency and recent versions also outputLatency. Safari (macOS and iOS) implements baseLatency but does NOT expose outputLatency, so the Output Latency card will show N/A and the total estimate reflects base latency only. No microphone permission is required — the tool only plays a short beep and reads the context's reported numbers; nothing is recorded. Because it uses the Web Audio API, it requires a secure context (HTTPS); on an insecure origin it shows an unsupported message. Everything runs locally in your browser: no audio, no measurements and no exported file ever leave your device — the CSV/JSON download is generated client-side.
Base Latency is the browser's reported delay between scheduling audio and the hardware buffer; Output Latency is the extra buffering the OS/device adds before sound leaves the speaker. The headline total is base + output latency in milliseconds and is the browser's ESTIMATE — it is not a microphone-verified, true acoustic round-trip, and it does not include your DAC, USB interface, Bluetooth codec or speaker delay. Safari shows N/A for output latency, so treat its total as base-only. For a ground-truth loopback measurement, play a sharp click and record it back through a microphone placed at the speaker, then cross-correlate the recorded onset against the trigger time; the time difference is the real end-to-end latency. Use this tool's base/output estimate plus the derived buffer size and CSV/JSON export to log and compare devices, and reserve the mic-loopback method when you need a certified absolute number.
DIAGNOSTICS11
AUDIO TOOLS40
WUTOOLS