How tsmdt/whisply works
Main Components Identification
The system should be represented with these key components:
a) Core Processing Pipeline:
• Input Handler (handles files, folders, URLs, and .list files)
• Transcription Engine (manages different Whisper implementations)
• Speaker Annotation System (WhisperX integration)
• Post-processing Module (corrections and enhancements)
• Output Generator (multiple format support)
b) Integration Components:
• FFmpeg Integration
• HuggingFace Integration
• Device Management System
[Input Sources] → [Input Handler] → [Device Manager] → [Processing Pipeline] → [Output Generator] → [Export Formats]
Specific Implementation Details
• The automatic device selection logic
• The parallel processing capabilities
• The integration points with external services
• The configuration management system
• The post-processing pipeline
Component Relationships
Key relationships:
• How the CLI (cli.py) and App (app.py) interface with the core system
• The interaction between diarize_utils.py and the speaker annotation system
• How enhance_annotations.py fits into the processing pipeline
• The role of post_correction.py in the output generation
Technical Details
Annotations for:
• Supported Whisper models
• Hardware requirements for each processing mode
• File format support
• Configuration options