Now with trainable voice profiles

From a 5 GB video to a
perfect transcript in minutes.

Drop in a video, get downloadable MP3 and M4A, plus a word‑timed transcript with speaker diarization. Voice Lab learns each speaker as you correct it — so the next file recognises them instantly.

Up to 5 GB videos

Resumable chunked upload to private storage. HEVC, MP4, MOV, MKV.

Smart MP3 + M4A extraction

Smaller desktop files extract audio locally in your browser. Larger files and phones hand off to the server for the encode — no failed loads on big videos.

Word‑accurate transcripts

WhisperX gives word‑level timestamps. Gemini does a second pass to fix typos and punctuation.

Speaker diarization

pyannote separates voices in any file. Each speaker gets a chip you can rename in one click.

Trainable voice profiles

Every name you assign teaches the model. Embeddings are matched by cosine similarity across files.

Exports everywhere

Download MP3, M4A, plain text, SRT subtitles, and JSON with timestamps + speaker IDs.