Super-simple subtitles with WhisperX

Usage

python transcription.py [-h] [--model MODEL] [--diarize] [--save_to_path SAVE_TO_PATH] media_path language

Alternatively, ./run.sh is mapped to python transcription.py.

Arguments

positional arguments:
  media_path            Path to the media file
  language              Language for translation

options:
  -h, --help            show this help message and exit
  --model MODEL         Model for transcription
  --diarize             Whether to diarize the audio
  --save                Whether to render a video + subtitles

Example

• English (with diarization and re-rendering with subtitles):
  • ./run.sh https://www.youtube.com/watch\?v\=Zl_5LT2fzak en --model=base --diarize --save
• Norwegian (no diarization or re-rendering):
  • ./run.sh https://tv.nrk.no/serie/munter-mat/2023/KOID33006222/avspiller no --model=NbAiLabBeta/nb-whisper-small

Setup and installation

• Find the proper PyTorch 2.0 installation for your system:
  • https://pytorch.org/get-started/previous-versions/#v200
• run make

Diarization

To support speaker diarization, you need to save your huggingface token in secrets.yml in the root dir:

1. Create a huggingface account

2. Accept the terms at https://huggingface.co/pyannote/speaker-diarization-3.1

3. Create a token from https://huggingface.co/settings/tokens

4. Create secrets.yml in the root dir with the following content:

   HF: <your-token>