Fall 2022
This repository is a fork of openai/whisper
with some added tools and scripts related to speaker diarization.
Files authored by us: clustering/clustering_diarizer.py
, data_processing/process_ami_annotations.py
The code we wrote to perform all dataset pre-processing and run our clustering experiments is in this Colab notebook.
The code we used to fine-tune the Whisper model (adapted from this fine-tuning notebook and edited heavily) is in this Colab notebook.
###Setup Instructions
sudo apt update && sudo apt install ffmpeg
(dependency for the Whisper package)- Clone this repo:
git clone https://github.com/melaniezhang/whisper-diarization.git
pip install -r requirements.txt
pip install librosa
python clustering/clustering_diarizer.py