Transcribing Indian languages: a practical review guide

What the transcription gives you

Clipy extracts the audio from your video locally as WAV and sends it for transcription; the video itself stays on your device. You get timestamped text you can edit line by line while the video plays beside it.

No transcript of mixed Hindi and English speech is perfect. Names, numbers, code-switched words and fast conversation are where errors collect, so plan a short review pass rather than trusting the first output.

A review pass that catches real errors

1. Add your video, up to 250 MB and 10 minutes, and generate the transcript.
2. Play the video from the first subtitle and read along line by line.
3. Fix names, places and numbers first; they are hardest to guess back from context.
4. Check code-switched words are spelled the way your audience writes them.
5. Adjust start and end times where a line starts late or runs long.
6. Export SRT or VTT, or burn the captions into an MP4 with the bundled Noto Sans Devanagari font.

Troubleshooting the output

If a whole stretch is missing, the speech may be under music or noise; try a cleaner section or trim to the spoken part first. If Devanagari shows boxes in another player, your player lacks the font, so burn the captions in instead.

Keep the original transcript before you edit. If a fix goes wrong, you can start the review again from the untouched text.

Common questions