GitHub - MrBinit/Speaker-embedding-and-Clustering: This project uses a pre-trained diarization pipeline and embedding model to detect and cluster speakers from audio, assigning unique IDs per speaker. Speaker embeddings are clustered with DBSCAN based on cosine similarity, and audio metadata is managed via SQLAlchemy for database interactions.

Code Overview

Diarization Pipeline: Uses the pre-trained pyannote/speaker-diarization-3.1 pipeline for diarization.
Embedding Model: Uses the pre-trained pyannote/embedding model to extract speaker embeddings.
Clustering: Embedding vectors are clustered using DBSCAN based on cosine similarity.
Database Interaction: The script interacts with a database to retrieve and update audio file metadata using SQLAlchemy.

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
readme.md		readme.md
requirements.txt		requirements.txt
speaker_embedding.py		speaker_embedding.py