Audio to Text
Turn an MP3, WAV or voice recording into English text — transcribed privately in your browser.
Drop your audio here
or browse files · max 50 MB
First use downloads the language model (~75 MB, one time — then it's cached and works offline). MP3 · WAV · M4A · OGG · up to 10 minutes.
Frequently Asked Questions
Which languages are supported?
English only, by design — the tool uses an English-optimised speech model, which keeps it small enough to run in your browser and noticeably more accurate on English than a general multilingual model of the same size.
Is my audio uploaded anywhere?
No. The transcription runs inside your browser. On first use, the tool downloads its language model (about 75 MB, one time — it is then cached), but your audio file itself never leaves your device.
How accurate is it?
Very good on clear speech — podcasts, voice notes, lectures recorded close to the microphone. Heavy background noise, crosstalk, strong accents, or distant microphones reduce accuracy. It adds punctuation automatically.
Is there a length limit?
Up to 10 minutes of audio per file. Longer audio is processed in 30-second chunks internally, but beyond ~10 minutes browser memory and time become impractical, especially on phones. For longer recordings, split the file first.
Which audio formats work?
MP3, WAV, M4A and OGG. If you have a video, run it through our Video to MP3 tool first, then transcribe the MP3 here.
Key Terms
Whisper
An open speech-recognition model family. This tool runs the compact English-only variant fully in your browser via WebAssembly — no server involved.
One-time model download
The first transcription downloads the model file (~75 MB) and caches it in your browser. Later uses load from cache and even work offline.
16 kHz mono
Speech models listen to audio as a single 16,000-samples-per-second channel. The tool converts your file to this format automatically before transcribing.