Audio to text: free transcription in your browser

How to transcribe audio to text

1. Choose the spoken language, or leave “Detect automatically”.
2. Press the microphone to record, then stop. Or choose a file: MP3, M4A, WAV, OGG, FLAC, WebM, or a video (MP4, MOV, WebM). You can also drag the file onto the page.
3. The first time, your browser downloads the speech model (about 77 MB for Fast, 249 MB for Accurate). Next time it starts right away.
4. The text appears while it is being transcribed. Edit it if needed, then copy it or download it as .txt, or as .srt / .vtt subtitles with timestamps.

Private by design: your audio stays on your device

Most transcription sites upload your recording to their servers, and many limit you to a few minutes a day unless you pay. This tool runs OpenAI’s Whisper speech recognition model directly in your browser. Your audio and the transcript never leave your computer or phone, so it is suitable for interviews, meetings, lectures, voice memos and other recordings you would rather not upload.

Fast or Accurate?

• Fast (Whisper base, ~77 MB) is quick and works well for clear speech in widely spoken languages. On a typical laptop it transcribes a minute of audio in about 15 seconds.
• Accurate (Whisper small, ~249 MB) makes fewer mistakes with accents, background noise, names, and languages such as Japanese, Korean or Chinese. It is about three times slower. On older phones, use Fast.

Supported files and languages

Any audio or video format your browser can play works: MP3, M4A (iPhone voice memos), WAV, OGG, Opus (WhatsApp voice messages), FLAC, AAC, WebM, MP4 and MOV. For video files, only the sound is used. Whisper recognizes 99 languages, including English, Spanish, Portuguese, French, German, Italian, Japanese, Korean, Chinese, Russian, Arabic, Hindi and Turkish. Tick Translate to English to get an English translation of speech in any of these languages.

Make subtitles (SRT and VTT)

Every transcript keeps timestamps, so you can download it as an SRT or WebVTT subtitle file and add it to YouTube, a video editor or a media player. Turn on Timestamps to see when each sentence starts.

Frequently asked questions

Is it really free?

Yes. There is no account, no watermark and no limit on the number of files or minutes. The page is supported by ads.

How accurate is it?

With clear speech, Whisper usually gets more than 9 words out of 10 right; the Accurate setting does better with noise and strong accents. Check names, numbers and technical terms before you use the text. Music, overlapping voices and very quiet recordings reduce accuracy.

How long can a recording be?

There is no fixed limit. Transcription runs on your device, so long files simply take longer: with Fast, a one-hour recording takes roughly 15 minutes on a recent laptop. Keep the tab open until it finishes.

Why is the first transcription slower?

Your browser first downloads the speech model and stores it. After that, the model loads from your browser’s storage, even offline.

Does it work on my phone?

Yes, in recent versions of Chrome, Safari, Edge and Firefox on Android and iPhone. Phones are slower than computers, so the Fast setting is the better choice there.

Does it tell different speakers apart?

No. The transcript is a single text with timestamps; it does not label who is speaking.

Can I record first and transcribe later?

Yes. Record with the voice recorder, download the MP3, then drop the file here whenever you want the text.