Transcribe audio and video files to text with timestamps. Paste direct MP3, MP4, M4A, WAV or WEBM links and get the full transcript, timed segments, spoken language and duration as clean JSON. Whisper runs inside the Actor: no API key, no signup, nothing to install.
Priced: $0.01 per run, plus $0.009 per minute of audio, rounded up, with a minimum of one minute
per file. A file that fails, has no audio track, or does not fit your run's maximum charge is not
charged.
Set a maximum charge and the run stops before a file it cannot pay for. That file's row says so,
and nothing is charged for it.
The example input is now a short clip with speech in it, so the first run shows a real
transcript.
0.1 - 2026-09-13
First build. Paste direct links to audio or video files (MP3, MP4, M4A, WAV, WEBM, MOV) and get
one row per file: the full transcript, timed segments with start and end seconds, the spoken
language, the duration and a word count.
Whisper base runs inside the Actor. No API key and no signup.
One required field, mediaUrls. language is optional (leave it empty to detect the language),
and maxFiles stops a run at 10 files by default, up to 100.
Every file the run attempts gets a row, including the ones that fail: a status, a reason from
a fixed list your pipeline can branch on, and a sentence saying what happened.
Limits: https links only, up to 200 MB and 10 minutes per file. A link to a private or internal
address is refused before anything is fetched, and again on every redirect.
If the transcription model cannot load, the run fails and says so, instead of returning a failed
row for every file.
