How sevask/ecoute works
Main Components Identification
a) Input/Output Layer:
• Microphone Input Handler
• Speaker Output Handler
• User Interface (Textbox Display)
b) Processing Layer:
• Audio Recording System (AudioRecorder.py)
• Transcription System (AudioTranscriber.py)
• GPT Response Generation (GPTResponder.py)
c) External Services:
• OpenAI GPT-3.5 API
• OpenAI Whisper API (optional)
Component Relationships
a) Primary Flow:
GPTResponder → Response Display
b) API Interactions:
• AudioTranscriber ↔ Whisper API (when --api flag is used)
• GPTResponder ↔ OpenAI GPT-3.5 API
Default Path
• Local Whisper model (tiny.en.pt)
• English-only processing
API Path (--api flag)
• Connection to Whisper API
• Multi-language support
Technical Details
Annotations for:
• Python >=3.8.0 requirement
• FFmpeg dependency
• Windows OS compatibility
• Real-time processing indicators
• Default audio device dependencies
Additional Notes
A notes section highlighting:
• Real-time nature of the system
• Warm-up period requirement
• Language limitations
• Default device dependencies
• API credit usage considerations