Drop your audio here
or click to browse — MP3, WAV, M4A, OGG, WEBM
🔒 Private by design: transcription runs 100% in your browser. Your audio never leaves this device.
The model downloads once (Turbo ≈ 810 MB, Small ≈ 250 MB) and is cached in your browser afterwards. Turbo is recommended for Nepali accuracy.
…
Transcript
Click a time code to jump · double-click text to edit
How it works
- Upload a Nepali audio clip (interview, speech, song, voice note).
- Transcribe with OpenAI's Whisper running locally in your browser via WebGPU.
- Review time-coded segments, click any time code to jump, edit text inline.
- Export as SRT/VTT subtitles, timestamped TXT, or JSON.
Long recordings are processed in 30-second windows. Very long files (1 hr+) are best on a desktop with WebGPU; the smaller model is faster on modest devices. Nothing is uploaded anywhere.