Speech to Text
Transcribe audio and video with AI.
Transcript studio
Audio in, text out.
Drop audio to transcribe
Interviews, calls, lectures and voice notes. Choose the language, model and output before you run.
MP3, WAV, M4A or OGG · up to 50 MB
The file stays here until you start the cloud transcription.
A simpler workflow
How Speech to Text works
- 1
Add an audio file. The waveform and duration load locally.
- 2
Choose the language, model, timestamps and speaker labels.
- 3
Sign in and transcribe. Edit the result, then download TXT, SRT or VTT.
Made to stay out of your way
- Word-level timestamps and speaker labels for interviews and meetings.
- Editable transcript output in TXT, SRT or VTT format.
- Clear pricing at 2 credits per started minute, with failed runs refunded.
Frequently asked questions
How much does transcription cost?
2 credits per started minute, with a minimum of 2 credits per run. A 90-second clip costs 4 credits. The exact cost appears before sign-in.
Which formats and limits are supported?
MP3, WAV, M4A and OGG, up to 50 MB and 30 minutes per file. Longer recordings can be split into parts.
Can I get timestamps and speaker labels?
Yes. Turn on timestamps for timecoded lines and speaker labels to separate voices. SRT and VTT always include timestamps.
What can I export?
TXT for plain transcript text, SRT for video subtitles, or VTT for the web. The transcript stays editable before download.
Which languages are supported?
Auto-detect plus English, Hindi, Spanish and French. More languages can be added with the transcription engine.
Is my audio uploaded?
The file stays in your browser until the run starts. Server processing uses private temporary storage, and files are deleted within one hour.
How accurate is the transcript?
Accuracy is strongest on clear speech with little background noise. Choose the Precise model for accents, calls or noisy recordings, and review names and numbers.
Last updated October 2026