AI Video & Audio Transcriber

Import a recording, choose its spoken language and get the full editable transcript.

Free browser-local processing · tested on desktop Chrome
Supported formats and file limits

Try MP3, WAV, M4A, OGG or MP4/WebM with a browser-supported audio codec. MKV, MOV, AVI, AAC and FLAC depend on your browser. If decoding fails, convert to WAV/MP3 or use the local backend version. Browser limit: 250 MB / 60 minutes per file. Start with a 1-minute sample on phones. Chrome desktop has been tested; Firefox, Safari, Edge, phone and embedded chat browsers are not verified.

Select Transcript Language

No language is selected automatically. Select the language being spoken, not a translation target. Hindi works best with Small. Bengali returned the wrong script in our tests and is not verified. Mixed-language speech and scripts may need correction.

Model, processing and limits
First use downloads a free model, possibly hundreds of MB. Processing uses your device memory and battery and may be slower than the recording. Keep this tab open and the screen awake. Completed chunks are saved locally; uploaded audio is not saved. After a reload, reselect the file to resume.

Full transcript

ASR can miss or invent words. Listen and review before relying on it. Mark unclear parts as [Unclear Audio] in the editor. This browser engine does not provide reliable confidence scores; automatic uncertainty detection is not available.

Full: complete model output. Verbatim: same raw output, without editorial rewriting; fillers are not guaranteed. Clean: whitespace cleanup only, no summary or sentence removal.

Add a file to begin.

TXT/DOCX use the editor above. SRT/VTT use timed segment text below. Edit segments there to keep subtitle timing accurate. Editing the plain transcript does not invent new subtitle timings.

Edit timed subtitle segments