StellaScript Documentation

StellaScript is a Python application designed for audio transcription and speaker diarization. Its primary goal is to provide an accurate and efficient tool for converting audio streams, whether pre-recorded or captured live, into structured text while identifying the different speakers.

The system is based on a modular architecture and integrates several state-of-the-art machine learning models for its key features:

• Speech Recognition: Utilizes OpenAI’s Whisper model, through the whisperx library for optimized performance, to ensure accurate transcription.

• Speaker Diarization: Integrates the pyannote.audio pipeline for audio segmentation and turn-taking identification.

• Speaker Identification: Generates voice embeddings with SpeechBrain to differentiate and track speakers consistently.

This documentation aims to provide a technical overview of the project, its architecture, and its API.

Contents

• Core Concepts
  • Visual Overview of the Pipeline
  • Whisper Model Selection
  • Optimizing Transcription Quality with Whisper
  • Bibliography
• Technical Section
  • StellaScript Project Architecture
• API Reference
  • Project Modules

Indices and tables

• Index

• Module Index

• Search Page