While working on another side project, I noticed how limited and fragmented transcript APIs are for developers. Many platforms make it unnecessarily hard to scrape transcripts, and some don't even offer content with native captions.

I built a single API that:

Fetches video metadata and audio

Performs noise reduction and voice activity detection (VAD)

Generates word-level transcripts at scale

It can transcribe entire channels, playlists, or individual videos in minutes, even when no captions exist.

I’ve also built a web playground for non-devs who just want to play around, translate a video and watch it directly on the site, with captions overlaying inline on the video player.

Outputs are available in multiple formats: plain text (with or without timestamps), SRT, VTT, and JSON. The API supports transcription in any language and translation to 100+ languages with ~95% accuracy.