An editable transcript in more than 90 languages, made in your browser. Export it as plain text, or with a time on every line. Your file never leaves your device.
Keep this page open. Switching apps pauses it until you're back.
Free, no watermark, no account. Works offline once loaded.
Four steps, nothing to install and no account to make. It works in your browser, on a computer or a phone.
Drop a file on the page, or pick one from your device. It's read where it is — nothing is uploaded, and nothing is sent anywhere.
Speech becomes text right in your browser, with every word timed, in whichever of more than 90 languages is spoken. The first time, it fetches the speech model; after that it starts straight away.
Lines arrive as they're written, and you can edit any of them while the rest is still being transcribed. Press a line's time to hear that part again. Words the transcriber was unsure of are flagged.
Save it as plain text, or with the time each line starts — useful for interviews and meetings, where you need to find the moment again.
Most transcription services want your file on their servers before they'll give you a word of it. That means waiting for it to arrive, trusting someone else with a recording that might be a private interview, a client call, a medical appointment or a recorded lecture — and usually paying by the minute once you're past a free trial.
This runs the speech model on your own machine instead. Nothing is sent anywhere, so there's nothing to wait for and nothing to meter. A two-hour recording costs the same as a two-minute one, which is to say nothing at all.
Once you have the words, you can add captions to the video from the same transcript, or trim the video down to the part worth keeping.
Yes. There's no watermark, no account, no sign-in and no limit on how many files you transcribe. The work happens on your own device, so there's no server to pay for and nothing to meter.
No. Your video or audio is read on your device and never leaves it. The transcript is made in the browser and stays there until you export it. That's why it works offline once the page has loaded.
The same speech recognition sits behind both. This page gives you the words as a document you can read and edit. The captions tool adds the timing and styling you need for subtitles, and exports SRT, VTT or a video with the captions burned in.
Yes. Export offers the transcript as plain text, or with the start time in front of every line, which is what interview and meeting transcripts are usually wanted with.
It uses OpenAI's Whisper, the same model behind many paid transcription services. On clean speech it's close to word for word. Accents, crosstalk, background noise and specialist vocabulary are where any transcriber slips, which is why every line is editable and uncertain words are flagged for you.
More than 90. It detects the spoken language on its own, or you can pick one before you start. Chinese, Japanese, Thai, Lao, Burmese, Khmer and Tibetan aren't supported yet, because the transcript is built around words separated by spaces.
Up to three hours on a computer, or an hour on a phone. There's no limit on file size — only on length, because transcribing is the slow part.
Yes. Every line can be typed over, and the timings follow your edits, so the times stay right against the video. Your work is kept in the browser if you close the page by accident.