A Python script that performs speech diarization (speaker identification) in the current REAPER project, automatically splitting a single mixed recording onto new, colour-coded tracks for each unique speaker.

This script uses the senko library for diarization and reapy to interact with REAPER.

1. REAPER Integration: Automatically connects to a running REAPER instance and targets the first media item on the first track.
2. Format Check & Conversion: Uses ffmpeg-python to check if the source audio is in the required 16kHz, mono, 16-bit PCM WAV format. If not, it converts it to a temporary file before processing, ensuring compatibility.
3. Individual Tracks: Creates a new, unique track for every detected speaker and assigns a distinct colour.
4. Item Splitting: Splits the original media item and moves the resulting segments to the corresponding speaker tracks.
5. Caching: Supports saving and loading diarization results to a JSON file to avoid re-running the heavy diarization process.

• REAPER, with reapy set up
• uv for Python virtual environment management
• ffmpeg installed in your $PATH

1. Clone the Repository:

   git clone https://github.com/atmosfar/reaper_speech_diarizer.git
   cd reaper_speech_diarizer

2. Create and Activate Virtual Environment:

   uv venv --python 3.11.13
   source .venv/bin/activate

3. Install Dependencies:

   uv pip install -r requirements.txt

The script must be run from your terminal/command line outside of REAPER, but while a REAPER project is open. It will automatically connect to REAPER using the reapy library.

1. Prepare Your REAPER Project:

• Ensure the audio you want to diarize is the first item on the first track of your active REAPER project.
• Save your REAPER project.

2. Run the Script:

The primary script is named reaper_speech_diarizer.py.

• To Run Full Diarization:

  python reaper_speech_diarizer.py

• To Run and Save Results to a JSON Cache:

  python reaper_speech_diarizer.py --json_output results/my_diarization_cache.json

• To Load Results from a JSON Cache (skips diarization):

  python reaper_speech_diarizer.py --json_input results/my_diarization_cache.json