About the Tool
What does the Sound Latency Test actually measure? The tool measures round-trip audio latency — the total elapsed time in milliseconds between a sound being output by your system and that same sound being captured back by your microphone. It emits a short tone burst through your speakers or headphones, records the microphone input simultaneously, and uses cross-correlation analysis to identify the time offset between the two signals. The result reflects the combined latency of your browser, operating system audio stack, and audio hardware. For the full technical pipeline, see the How It Works page.
Why do I need to give microphone permission? The measurement requires capturing the output tone burst through your microphone to calculate the round-trip delay. Without microphone access the tool has no way to detect when the emitted sound actually arrived at the capture point. The microphone audio is processed entirely within your browser and is never recorded, stored, or transmitted. See the Data Security page for the full explanation of how your audio is handled.
What is a normal latency result? Normal varies significantly by hardware type. As a practical reference: external audio interfaces with ASIO or Core Audio drivers typically produce 5–15 ms; built-in laptop or desktop audio typically produces 30–80 ms; USB class-compliant audio devices typically produce 20–50 ms; Bluetooth headphones or speakers typically produce 70–250 ms depending on the codec. If your result falls significantly outside the range for your hardware type, see the Troubleshooting page.
How accurate is the measurement? The cross-correlation method achieves sample-level theoretical resolution — approximately 0.023 ms per sample at 44,100 Hz. In practice, real-world accuracy on a single test run is ±3–5 ms due to microphone frequency response variation, OS audio processing (AGC, noise suppression), and background noise. Running three or more tests and averaging the results improves practical accuracy to ±1–2 ms.
Can I run the test more than once? Yes, and running multiple tests is recommended. A single test captures one measurement at one moment in time. System load, background processes, and minor timing variation can cause individual results to differ by a few milliseconds. Three or more runs averaged together give a more reliable baseline figure.
Understanding Your Result
My result is 5 ms. Is that correct? A result of 5 ms or lower is achievable with a professional external audio interface using low-latency ASIO drivers on Windows or Core Audio on macOS, combined with a low-buffer-size setting. If you are using built-in laptop audio and getting 5 ms, the result is likely being influenced by hardware-level zero-latency monitoring on your interface, which routes audio before it enters the software stack and cannot be measured by a browser-based tool. In that case the measured figure does not represent true software round-trip latency.
My result is over 200 ms. What is causing that? Results above 200 ms are most commonly caused by Bluetooth audio output. Bluetooth codecs — including SBC, AAC, aptX, and LDAC — introduce encoding and decoding overhead that adds 40–200 ms before the Web Audio API pipeline even begins. If you are using Bluetooth headphones or speakers, switching to wired audio and retesting will produce a dramatically lower result. Other causes of high results include large OS audio buffer sizes and certain browser audio processing settings.
My result changes every time I run the test. Why? Some variation between runs is normal and expected. The OS audio buffer flush timing, system load at the moment of the test, and minor variation in microphone capture timing all contribute to run-to-run differences of 2–10 ms. Variation larger than 10–15 ms between runs on the same hardware setup suggests a system-level audio instability — large OS buffer sizes, conflicting audio drivers, or background applications competing for audio device access.
What does my latency result mean for my DAW? In a DAW, your audio interface buffer size setting determines software monitoring latency. A buffer of 128 samples at 44,100 Hz = approximately 2.9 ms output latency; 256 samples = ~5.8 ms; 512 samples = ~11.6 ms. The round-trip figure measured by this tool is roughly double the one-way figure your DAW uses for latency compensation, because it includes both output and input pipeline delay. Use the measured round-trip figure divided by two as a cross-reference against your DAW’s reported latency compensation value.
Does my result tell me the latency my audience hears in a live stream? Not directly. Streaming introduces additional network transmission latency and encoder/decoder buffering that are entirely separate from the local audio round-trip. The tool measures your local system latency — the delay in your own monitoring chain — not the end-to-end latency between your microphone and a viewer’s speakers.
Accuracy and Limitations
Why does my result differ from what my DAW reports? DAW latency figures are calculated from your audio interface’s reported buffer size and sample rate. Browser-based measurement uses cross-correlation of real audio signals, which captures the full round-trip including OS buffer overhead, browser scheduling latency, and hardware I/O delay — some of which DAWs do not account for in their displayed figures. The browser measurement is typically higher than the DAW figure because it measures the complete system rather than the interface buffer alone.
Can the test measure my audio interface’s hardware latency independently? No. The round-trip figure includes hardware latency, OS latency, and browser latency combined. The tool cannot isolate the contribution of each layer separately. To measure hardware-only latency independently, a dedicated measurement application with direct hardware access — such as a professional acoustic measurement tool — is required.
Does the test work with Bluetooth microphones? Yes, but the result will reflect the combined latency of both Bluetooth output and Bluetooth input, which can produce round-trip figures of 200–400 ms or more. For a meaningful measurement of your audio system’s base latency, wired audio for both output and input is strongly recommended.
Why does the result vary between Chrome, Firefox, and Safari? Each browser implements the Web Audio API’s scheduling and buffer management differently. Chrome and Edge typically report lower baseLatency values than Firefox on the same hardware. Safari on macOS has direct access to Core Audio and often produces lower latency figures than Chrome on the same machine. These differences are real browser-level differences, not measurement errors.
Is browser-based latency measurement as accurate as professional measurement tools? No. Professional acoustic measurement tools with dedicated hardware interfaces, ASIO driver access, and calibrated measurement microphones achieve sub-millisecond accuracy. Browser-based measurement is a practical diagnostic tool — accurate enough to identify whether your latency is in the range appropriate for your use case and to compare setups, but not a substitute for certified professional measurement.
Technical Questions
What is AudioContext.baseLatency and why doesn’t it equal my result? AudioContext.baseLatency is a property exposed by the Web Audio API that estimates the latency introduced by the browser’s audio processing pipeline on the output side only. It captures one component of round-trip latency. It does not include OS audio buffer latency, hardware I/O latency, microphone input latency, or the return journey through the capture pipeline. The full round-trip figure measured by this tool is almost always substantially higher than baseLatency alone.
What is the difference between latency and jitter? Latency is the consistent delay between output and capture — the figure this tool measures. Jitter is variation in that delay over time. High jitter — results that vary by 15 ms or more between runs on the same hardware — indicates an unstable audio pipeline and is often more disruptive than consistent high latency, because it makes timing compensation difficult to set reliably.
Why does lowering my buffer size reduce latency but increase CPU load? Audio buffer size determines how many samples the OS processes in each audio cycle. Smaller buffers mean shorter processing cycles and lower latency, but the CPU must complete each cycle faster to keep up. At very small buffer sizes (32 or 64 samples), many consumer machines cannot complete the audio processing cycle in time and produce dropouts or crackling. The lowest stable buffer size for your hardware is the optimal setting for minimum latency.
About This Website
Who runs SoundLatencyTest.com? SoundLatencyTest.com is built and maintained by Jackson Cooper, an audio systems engineer and web audio latency researcher. Jackson writes all content on the site personally. There are no guest contributors, no sponsored content, and no advertisers. For Jackson’s full background and the reasons this site was built, see the Author page.
Is this site free to use? Yes. The Sound Latency Test and all educational content on SoundLatencyTest.com are completely free. There is no registration, no account required, and no payment of any kind.
How do I report an error or ask a technical question? Use the Contact page with the subject line “Content Correction” for errors, or “Technical Question” for measurement questions. Jackson reviews all submissions personally. Response time for technical questions is 48–72 hours.
Written by Jackson Cooper, founder of SoundLatencyTest.com. Last updated: June 2026.
{ “@context”: “https://schema.org”, “@type”: “FAQPage”, “mainEntity”: [ { “@type”: “Question”, “name”: “What does the Sound Latency Test actually measure?”, “acceptedAnswer”: { “@type”: “Answer”, “text”: “The tool measures round-trip audio latency — the total elapsed time in milliseconds between a sound being output by your system and that same sound being captured back by your microphone. It emits a short tone burst through your speakers or headphones, records the microphone input simultaneously, and uses cross-correlation analysis to identify the time offset between the two signals. The result reflects the combined latency of your browser, operating system audio stack, and audio hardware.” } }, { “@type”: “Question”, “name”: “What is a normal latency result?”, “acceptedAnswer”: { “@type”: “Answer”, “text”: “Normal varies significantly by hardware type. External audio interfaces with ASIO or Core Audio drivers typically produce 5–15 ms. Built-in laptop or desktop audio typically produces 30–80 ms. USB class-compliant audio devices typically produce 20–50 ms. Bluetooth headphones or speakers typically produce 70–250 ms depending on the codec used.” } }, { “@type”: “Question”, “name”: “How accurate is the measurement?”, “acceptedAnswer”: { “@type”: “Answer”, “text”: “The cross-correlation method achieves sample-level theoretical resolution of approximately 0.023 ms per sample at 44,100 Hz. In practice, real-world accuracy on a single test run is plus or minus 3–5 ms due to microphone frequency response variation, OS audio processing such as AGC and noise suppression, and background noise. Running three or more tests and averaging the results improves practical accuracy to plus or minus 1–2 ms.” } }, { “@type”: “Question”, “name”: “Why do I need to give microphone permission?”, “acceptedAnswer”: { “@type”: “Answer”, “text”: “The measurement requires capturing the output tone burst through your microphone to calculate the round-trip delay. Without microphone access the tool has no way to detect when the emitted sound actually arrived at the capture point. The microphone audio is processed entirely within your browser and is never recorded, stored, or transmitted over the network.” } }, { “@type”: “Question”, “name”: “Why does my result differ from what my DAW reports?”, “acceptedAnswer”: { “@type”: “Answer”, “text”: “DAW latency figures are calculated from your audio interface’s reported buffer size and sample rate. Browser-based measurement uses cross-correlation of real audio signals, which captures the full round-trip including OS buffer overhead, browser scheduling latency, and hardware I/O delay — some of which DAWs do not account for in their displayed figures. The browser measurement is typically higher than the DAW figure because it measures the complete system rather than the interface buffer alone.” } }, { “@type”: “Question”, “name”: “Why does the result vary between Chrome, Firefox, and Safari?”, “acceptedAnswer”: { “@type”: “Answer”, “text”: “Each browser implements the Web Audio API scheduling and buffer management differently. Chrome and Edge typically report lower baseLatency values than Firefox on the same hardware. Safari on macOS has direct access to Core Audio and often produces lower latency figures than Chrome on the same machine. These are real browser-level differences, not measurement errors.” } }, { “@type”: “Question”, “name”: “What does my latency result mean for my DAW?”, “acceptedAnswer”: { “@type”: “Answer”, “text”: “In a DAW, your audio interface buffer size setting determines software monitoring latency. A buffer of 256 samples at 44,100 Hz equals approximately 5.8 ms one-way latency. The round-trip figure measured by this tool is roughly double the one-way figure your DAW uses for latency compensation, because it includes both output and input pipeline delay. Use the measured round-trip figure divided by two as a cross-reference against your DAW’s reported latency compensation value.” } }, { “@type”: “Question”, “name”: “Who runs SoundLatencyTest.com?”, “acceptedAnswer”: { “@type”: “Answer”, “text”: “SoundLatencyTest.com is built and maintained by Jackson Cooper, an audio systems engineer and web audio latency researcher. Jackson writes all content on the site personally. There are no guest contributors, no sponsored content, and no advertisers.” } } ] }