This program fixes a subtitle file timings by syncing them to the audio of a movie using offline speech recognition.
- Currently, only supports windows, but this can be extended by adding vosk and ffmpeg binaries for linux and mac-os and fixing the file paths.
- Improve heuristic for matching the existing subtitle text to the generated subtitle timings. Have more focus on keeping the subtitles in the same order.
- clone the repository
- make sure you have NodeJS installed (v16.x.x)
- run
npm install
- download the vosk model and store it under
./vosk-speech-model
(so thatvosk-speech-model/am/final.mdl
exists) - run
npm run start:example1
- run
npm run build
- the exe will be created in the
dist/
folder