Audio Recorder
Record from any microphone and download as WebM, WAV, MP3, M4A, or OGG. Optional local Whisper transcription. Browser-only, no upload.
published
- [FREE]
- [NO_SIGNUP]
- [NO_UPLOAD]
A browser audio recorder that captures your microphone and lets you download the result in the format you actually need — without installing anything, creating an account, or sending a single byte of audio to a server.
What it does
- Records from any microphone your browser can see, with a device picker, live level meter, timer, and pause/resume.
- Downloads the clip as the original WebM/Opus (instant), WAV (instant, uncompressed), or MP3 / M4A / OGG (via a local FFmpeg).
- Transcribes the recording to text with a local Whisper model — plain text, timestamped text, SRT, or WebVTT — without leaving the page.
Recording format vs. download format
Browsers don’t let a web page record straight to MP3. MediaRecorder writes WebM with the Opus codec in Chrome, Edge, and Firefox, and MP4 with AAC in Safari. Opus is a modern codec that sounds better than MP3 at the same bitrate, so the original recording is usually the best file to keep.
When you need a specific format for another app, convert:
| Format | How it’s made | Best for |
|---|---|---|
| WebM / MP4 (original) | native MediaRecorder | smallest, highest quality, instant |
| WAV | decoded + PCM-encoded in-page | editing in a DAW, maximum compatibility, instant |
| MP3 | FFmpeg libmp3lame (q≈2, ~190 kbps VBR) | universal playback, podcasts |
| M4A | FFmpeg AAC 192 kbps | Apple ecosystem, smaller than MP3 |
| OGG | FFmpeg libvorbis q5 | open-format playback, web |
WAV and WebM need no download. MP3 / M4A / OGG load FFmpeg compiled to WebAssembly (~30 MB) on first use — the binary is the codec, not a server round-trip.
Built-in transcription
Stop a recording and a transcription panel appears. It reuses the same locally-running Whisper engine as the Audio to Text tool: WebGPU when your browser supports it, WebAssembly otherwise. Pick Tiny for speed or Small for accuracy, choose a language (or auto-detect), and export the transcript as TXT, timestamped text, SRT, or WebVTT.
Privacy
The microphone stream is captured by the browser and held in page memory. There is no upload endpoint to hit — there’s no backend at all. The codec and model downloads (FFmpeg from jsDelivr, Whisper weights from the Hugging Face CDN) are one-way fetches of public binaries, cached after the first use. Verify it: open DevTools → Network, record and export, and watch for any request carrying your audio. There is none.
Related tools
- Audio to Text Transcriber — transcribe existing audio or video files, plus an SRT/VTT subtitle converter.
- Video Compressor — shrink video (with its audio track) in your browser.
- Video Splitter & Joiner — cut or merge clips before exporting.
Frequently asked questions
Is my recording uploaded anywhere?
No. The microphone stream is captured by your browser's built-in MediaRecorder and stays in the page's memory. WAV is encoded in-page, MP3/M4A/OGG are transcoded by an in-browser FFmpeg, and transcription runs a local Whisper model. The only network traffic is a one-time codec/model download from a CDN — those contain no audio. Open DevTools → Network while recording and exporting: nothing carries your audio.
Which formats can I download?
The original recording (WebM/Opus in Chrome and Firefox, MP4/AAC in Safari) downloads instantly. You can also convert to WAV (uncompressed 16-bit PCM, encoded instantly in-page), or to MP3, M4A (AAC), or OGG (Vorbis) via the in-browser FFmpeg. WAV is best for editing; MP3 is the most universally playable compressed format.
Why does MP3 / M4A / OGG take a moment the first time?
Those formats are produced by FFmpeg compiled to WebAssembly. The first conversion in a session downloads the FFmpeg core (~30 MB) from the jsDelivr CDN and caches it; later conversions are quick. WebM and WAV need no download at all. The FFmpeg binary is a codec, not a server — your audio never leaves the tab.
Can I pick which microphone to use?
Yes. The Microphone dropdown lists every audio input the browser exposes. Device names appear after you grant microphone permission the first time (browsers hide labels until then). Pick a device before recording; "System default" follows your OS setting.
How do I transcribe what I recorded?
After you stop recording, the "Transcribe this recording" panel appears. Choose a Whisper model (Tiny is fastest, Small is most accurate) and a language, then click Transcribe. The result can be copied or downloaded as plain text, timestamped text, SRT, or WebVTT. It runs entirely on your device.
Is there a time limit on recordings?
No hard limit, but the recording lives in memory, so very long sessions use more RAM (roughly a few MB per minute of Opus). For multi-hour recording, download periodically. Transcribing very long clips on the CPU also takes a while — WebGPU (Chrome/Edge) is much faster.
Can I pause and resume?
Yes. Pause holds the recording without ending it; Resume continues into the same file. Stop finalizes the clip so you can play it back, download it, or transcribe it.
Does it work on my phone?
Mostly. Recording works in mobile Chrome and Safari. Safari records to MP4/AAC rather than WebM, which the format converter and transcriber both handle. Microphone permission is required, and the page must be served over HTTPS (this site is).
Why is the recording WebM and not MP3 by default?
Browsers' MediaRecorder records to WebM/Opus (or MP4/AAC on Safari) natively — Opus is an excellent, modern codec. MP3 isn't a native recording target in any browser, so producing it requires the FFmpeg transcode step. The original WebM/Opus is the smallest, highest-quality option; convert only if you need MP3/M4A/OGG for a specific app.