How saigayathrigudla-1184/fluency-net works
Fluency-Net is a speech-therapy web application. The user records or streams speech through the browser; the backend transcribes it, combines text with acoustic cues, and uses an adaptive AI agent to produce clinical analysis and fluent speech output. The README documents multilingual transcription, SOAP notes, session persistence, and Kokoro/Edge TTS. The backend is concentrated in main.py; the excerpts show selected integrations and prompt handling, not every internal call or route.
Main entry points: main.py FastAPI Backend · index.html Browser UI · sessions.db Session Storage
Components 16
User Interaction
- Browser UI
index.html - FastAPI Backend
main.py
Speech Processing
- Speech Transcription
main.py - Acoustic Analysis
main.py - Language Handling
main.py
Therapy Intelligence
- Agent State
main.py - Therapy Strategy
main.py - Clinical Analysis
main.py
Persistence & Output
- Fluent Regeneration
main.py - TTS Router
main.py - Session Storage
data/sessions.db
Other components
- User
- Ollama LLM
- Deepgram Fallback
- Kokoro TTS
- Edge TTS
Connections 16
- User → Browser UI: records speech
README.md - Browser UI → FastAPI Backend: streams audio
README.md - FastAPI Backend → Speech Transcription: transcribes audio
README.md - FastAPI Backend → Acoustic Analysis: extracts cues
README.md - FastAPI Backend → Language Handling: handles languages
README.md - Speech Transcription → Deepgram Fallback: fallback transcription
main.py - FastAPI Backend → Agent State: updates state
main.py - FastAPI Backend → Therapy Strategy: selects tactic
README.md - FastAPI Backend → Clinical Analysis: requests analysis
main.py - Clinical Analysis → Ollama LLM: infers analysis
main.py - FastAPI Backend → Fluent Regeneration: reconstructs speech
README.md - Fluent Regeneration → TTS Router: routes synthesis
main.py - TTS Router → Kokoro TTS: synthesizes English
README.md - TTS Router → Edge TTS: synthesizes multilingual
README.md - FastAPI Backend → Session Storage: persists progress
README.md - FastAPI Backend → Browser UI: returns results
README.md