This tool uses WhisperX to transcribe all audio and video files in a folder. Additionally, it offers a guide for installation and CUDA and CUDNN version management.
- Install Python==3.11.
- Find out your CUDA version with
nvidia-smiand CUDNN withpython get_cudnn_version.py. - CUDA and PyTorch installation: Follow the instructions on the PyTorch website to install the correct version of CUDA and PyTorch for your system.
- With your CUDA and CUDNN version, head over to faster-whisper | Requirements, check for the correct version of
ctranslate2, based on the version of CUDA and CUDNN you have, and install it. - Install the required Python packages;
pip install -r requirements.txtIn the transcribe.py file, update the folder_path variable to the path of the folder containing and pass the model name or path to the AudioTranscriber class. Then, run the script:
python transcribe.py