Browser audio recorder

What a browser audio recorder should not do

Recording a few minutes of speech is one of the simplest things a computer can be asked to do, and it is remarkable how much the free tools want in exchange. One of the best-known ones caps the length of the take and offers to lift the cap for a subscription covering fifty other apps. Another stops at five minutes and, if you want to share the result, uploads it to a server that keeps it for three months. A third is not a web page at all but a browser extension you have to install, which is exactly what someone searching for a recorder that runs in a browser was trying to avoid.

None of those limits exists because recording is hard. They exist because a limit is a place to put a price. This page has no cap, no countdown, no account and no upload, and the reason is simply that none of them is needed: your browser can already record, and the audio never has to go anywhere.

There is no time limit, and no hidden one either

The only real constraint is your own device. Audio accumulates in this tab while it records, and the page shows how much has been captured as it grows — a statement of fact, not a warning shot before a wall. If a very long recording is going to be heavy, you are told before it becomes a problem rather than stopped at a number someone picked. Memory, not time, is what actually runs out, and a tab that runs out reloads itself without explaining why, which is precisely why the page says so in advance.

Why the file format depends on your browser

The recording is done by the browser's own recorder, and browsers do not agree on what to write. Chrome and Edge produce WebM containing Opus audio. Safari produces MP4 containing AAC. The code on this page is identical in both cases; the container is the engine's choice, not a setting. Rather than guess from a label, this page reads the first bytes of the finished file and names what it actually is, then uses that for the download name.

That matters if something downstream expects one specific format. When it does, the audio converter takes the file this page produced and writes the format you actually need, also without uploading it.

Why the recorder takes a moment to start

Browsers need a short run-up before the recorder is genuinely capturing, and the delay is not trivial: measured on 2 September 2026, Chromium discarded roughly half a second at the beginning of the first recording in a tab, and about a twentieth of a second on later ones. Safari lost almost nothing. A page that flips its button to "recording" the instant you click it is therefore telling you something that is not yet true, and the cost is your first word.

This page waits. The indicator says it is starting until audio has actually arrived from the recorder, and only then does it say it is recording. The elapsed counter starts from that same moment rather than from the click, and it stops while you are paused. When you stop, the length shown is read from the finished file itself, because that is the number that is true.

What happens to the audio

It stays in the tab. There is no upload endpoint on this site, no account to create, and no server that could hold a copy — the playback control above plays the same bytes the download button gives you, which is a demonstration rather than a promise. When you close the page without downloading, the recording is gone. If you would rather check that than believe it, the guide on verifying local processing shows what to look for in your browser's own network panel, and this page is an easy one to test because it needs no file to work at all.

Recording and editing are different jobs

This page captures and stops there, deliberately. Once a take exists, the work belongs to tools built for files: trimming an exact section out of a longer take, removing the silence at the beginning and end by threshold rather than by eye, or drawing the waveform when the recording has to be shown rather than heard.

And if the take comes back quiet, distorted or picking up the wrong device, that is a capture problem which no edit repairs. The microphone test is the sibling of this page: same permission, opposite job. It judges the microphone and throws the audio away; this one keeps the audio and says nothing about the microphone. Checking there first is usually faster than recording twice.