How shubhgit-hub/ai-voice-assistant-theraphy works
VoiceRx is a single-file Streamlit speech therapy application. Its main workflow takes a user recording through audio preparation, Whisper transcription, pronunciation scoring and feedback; the app also provides voice analysis, visualizations, adaptive and tongue-twister exercises, synthesized playback, progress tracking and AI coaching. The source excerpt is partial, so links for unsampled UI wiring and state handling are omitted; feature coverage also reflects the README.
Main entry points: ai_assistant.py Streamlit application · temp.wav Temporary audio
Components 16
Practice and capture
- Audio recording
ai_assistant.py - Adaptive exercises
ai_assistant.py - Tongue twisters
ai_assistant.py - Exercise playback
ai_assistant.py
Speech analysis
- Audio preparation
ai_assistant.py - Whisper transcription
ai_assistant.py - Pronunciation scoring
ai_assistant.py - Pitch and energy
ai_assistant.py - Audio visualizations
ai_assistant.py
Coaching and progress
- Speech feedback
ai_assistant.py - Session progress
ai_assistant.py - AI coach
ai_assistant.py
Application interface
- Streamlit application
ai_assistant.py
Other components
- Speaker
- Anthropic API
- Temporary audio
temp.wav
Connections 16
- Speaker → Streamlit application: uses
- Streamlit application → Audio recording: captures speech
- Audio recording → Audio preparation: prepares audio
- Audio preparation → Whisper transcription: transcribes
- Whisper transcription → Pronunciation scoring: compares transcript
- Pronunciation scoring → Speech feedback: scores pronunciation
- Streamlit application → Pitch and energy: analyzes voice
- Streamlit application → Audio visualizations: renders plots
- Streamlit application → Adaptive exercises: selects practice
- Streamlit application → Tongue twisters: offers practice
- Streamlit application → Exercise playback: plays exercise
- Streamlit application → Session progress: tracks sessions
- Streamlit application → AI coach: requests advice
- AI coach → Session progress: reads history
- AI coach → Anthropic API: requests coaching
- Streamlit application → Temporary audio: stores audio