WhisperX Studio logo

WhisperX Studio

Replicate-powered transcription cockpit

Basic configuration

Keep it light or open the advanced console when needed.

Stored locally in your browser
Publicly accessible audio file link
Use None to enable language detection.
Align output
Auto detect|Diarization disabled

Results console

Live status, transcripts, and structured JSON output.

idle
Model page
Outputs stream here once complete.

How to use WhisperX Studio

WhisperX Studio runs on Replicate. You bring your own API token, choose a public audio file URL, and we orchestrate the prediction run.

Generate a Replicate API token and paste it into the token field. It stays in your browser.
Provide a direct audio file URL (MP3, WAV, etc). The Replicate API must be able to reach it.
Open Advanced to tune diarization, alignment, VAD, and language detection.
Hit Run to start. Status updates and the transcript appear in the results panel.

Set language to None to enable auto-detection. If you enable diarization, add a HuggingFace access token.

Advanced configuration

Fine-tune WhisperX input parameters.

Diarization
Required for diarization and stored locally.
Debug