Turn any recording into words.
Drop in audio or video files and read them back in minutes, as plain text or with the time each line was said.
Free, with no account. Each recording is deleted the moment its words are ready.
Recent files
| Sent | Name | Length | Speed | Status | Delete |
|---|
This list is kept in this browser only. Each transcript is deleted 24 hours after it is made.
What happens to your recording
- It is held in memory and never written to a disk.
- It is deleted the moment its words are ready, or the moment you press Stop.
- The transcript stays at a private link for 24 hours, then it is deleted too. You can delete it sooner.
- The link is the only key. There are no accounts, and nobody can look a transcript up by name.
- Transcription runs on our own server with open source software. Nothing is sent to another company.
- No tracking, no ads, no cookies. This browser keeps a list of links to your own transcripts, so it can show your recent files. On a shared computer, delete them when you are done.
- To keep the daily limit fair, your internet address is held in memory for a day. We keep no log of it.
How it works
- Choose one recording or several, or drop them on the page.
- Watch the words arrive while each one is read.
- Copy the text, or export it as Word, text, subtitles or PDF.
The limits
- File size
- Up to 200 MB
- Length
- Up to 60 minutes
- Files a day
- 30 for each visitor
- Waiting
- Recordings are read one after another, so yours may wait a few minutes when the page is busy.
Good to know
Which files can I send?
Most audio and video files work: mp3, m4a, wav, ogg, flac, mp4, mov, webm and more. If a file has sound in it, it will usually be read. For a long video, export only the audio first to keep the file small.
Can I send several files at once?
Yes. Choose or drop as many as you like. They are sent 2 at a time and read one after another, so keep the page open until the last one is sent.
Which languages does it understand?
About a hundred, including English, Urdu, Hindi, Arabic, Spanish and French. The language is detected from the first half minute. If the guess is wrong, choose the language yourself and send the file again.
How accurate is it?
Clear speech in a quiet room comes back close to word perfect. Noise, people talking over each other, strong accents and phone lines lower that. Names and numbers are where a machine slips most, so check those against the recording before you rely on them.
What does unsure mean?
When the model had low confidence in a stretch of speech, that stretch is marked unsure. Treat it as a best guess.
Does it tell speakers apart?
No. The transcript shows what was said and when, not who said it.
Which formats can I export?
Word (.docx), plain text (.txt), subtitles (.srt), and PDF through your browser's print window. Word and text come as plain paragraphs, or with the time of every line when timestamps are on.
Where is my list of recent files kept?
In this browser only. It is a list of links to your own transcripts, and it empties itself as they pass 24 hours. A transcript's link also works on another device until then.
How is this free?
It runs on open source software called Whisper, on a server Webly Studio already owns, so there is no cost per minute. The limits keep it fair for everyone.