WhisperX - Real WhisperX transcription, word timestamps and speaker labels in your browser

WhisperX is an independent, unofficial hosted service for people who want to use real WhisperX without installing Python or operating a GPU server. Upload an audio or video file, receive a transcript with word timestamps and speaker labels, and review it in a browser editor. The editor combines media playback with a transcript so users can listen, jump to individual words, correct text, rename speakers, save changes and export their work. Supported exports include TXT, SRT, VTT and JSON. Edited text and original model output are kept distinct; edits are not presented as freshly aligned words. Private uploads require an account. Visitors can explore clearly labeled examples before signing in. The service is useful for interview transcripts, meeting notes, video captions and speaker-aware transcription. It offers dedicated audio-to-text, video-to-text, subtitle and speaker diarization pages using the shared editor. A verified account receives a one-time five-minute trial. Paid plans start at $9 per month for 120 minutes; a $29 monthly plan includes 600 minutes, with annual billing options also available. Initial transcription support is for English recordings. Machine transcription and speaker attribution can contain errors and should be reviewed. WhisperX links to the upstream WhisperX project and is not affiliated with or endorsed by its authors, OpenAI or pyannote.