Audio to text
Drop an audio file. It goes straight to our transcription worker, a speech model writes it down, and the text comes back here with timestamps, ready to copy or download as TXT, SRT or VTT.
A real run, word for word
We made this clip ourselves (voice by Kokoro, pictures by FLUX) so it can be published here and every word checked. It was not taken from any platform.
0:00Notes from the Thursday planning call, present were Priya, Tom and me.
0:05First, the website launch moves from the 3rd to the 10th, because the payment page still needs testing on phones.
0:12Second, Tom will send the price list to the printer by Monday, Priya owns the newsletter and wants the final copy by Wednesday.
0:20Open question, do we keep the free trial at 7 days or cut it to 3, we decide on the next call.
Checked against the script that was spoken: all 77 words match.
4 numbers were written as digits (“third” → “3rd”); that is formatting, not counted.
Written down by whisper-large-v3 on fal, 2026-10-31. The live tool runs whisper-large-v3-turbo on Groq; this sample was recorded before launch with Whisper large v3, the larger model of the same family.
How do you turn an audio file into text?
Drop the MP3, M4A, WAV or OGG above. Your browser reads its length, sends the file to our transcription worker, and Whisper large v3 turbo writes it down. The text comes back with timestamps as plain text, SRT or VTT; the file is deleted within two hours. Three runs a day are free without an account for recordings up to 3 minutes; after that it is 3 credits a minute, up to 10 minutes and 100 MB.
- Files
- MP3, M4A, WAV, OGG, OPUS, FLAC, AAC
- Size
- Up to 100 MB
- Free
- 3 a day, up to 3 min
- Then
- 3 credits a minute, up to 10 min
How to transcribe an audio file
Drop the file
A voice memo, a recording, a podcast clip: MP3, M4A, WAV, OGG, OPUS, FLAC or AAC, up to 100 MB. Your browser reads how long it is before anything is sent.
Pick the language and start
Name the language when you know it; it helps with short recordings. The upload goes straight to the worker, with a progress bar.
Copy or download
Read the text with or without timestamps, copy it, or download .txt, .srt or .vtt.
Audio to text questions
Is this audio to text tool free?
Three runs a day are free without an account, for recordings up to 3 minutes. After that you sign in and pay 3 credits a minute, rounded up; 1,500 credits are $15, which is 500 minutes. A run that fails is not charged.
What happens to my recording?
It is sent to our transcription worker, read once by the speech model and deleted within two hours; only the transcript is kept, for you. Nothing is published.
Can it handle a recording longer than 10 minutes?
Not yet. Cut it into parts of 10 minutes or less with any audio editor and upload them one by one.
Does it tell speakers apart?
No. The transcript is one voice after another without speaker labels. Timestamps help you find where each person starts.
How accurate is it?
For the voice memo above, every word matched what was said, with only the numbers written as digits. That was a clear voice close to the microphone, which is where speech models do best; echo, music and several people talking at once are where they slip, and we have not measured those recordings yet.
Which languages does it handle?
The model is multilingual. Leave the language on automatic or pick one of the 17 in the menu; naming the language helps when the first seconds are silence.
Transcribe a recording
Drop an audio file. The first three today are free.
More tools
Explore each tool’s instructions, output limits and available recorded examples.