How azure-samples/voicelive-api-salescoach works
Voice Live API Sales Coach is a browser-based sales-practice application. Users choose or create a scenario, start a voice conversation with an AI customer, and review performance and pronunciation feedback. The frontend manages scenario selection, audio capture, realtime messaging, and results; the Flask backend provides scenario and agent APIs, proxies voice sessions, and performs analysis. The README documents Azure AI and Speech integrations, including optional Foundry agents. Some backend route wiring and frontend audio internals were not sampled, so uncertain links are omitted.
Main entry points: App.tsx Practice interface · app.py Flask API · analyzers.py Conversation analysis · api.ts HTTP client
Components 22
Practice frontend
- Practice interface
frontend/src/app/App.tsx - Scenario selection
frontend/src/components/ScenarioList.tsx - Scenario editor
frontend/src/components/CustomScenarioEditor.tsx - Scenario state
frontend/src/hooks/useScenarios.ts - Custom scenarios
frontend/src/services/customScenarios.ts - HTTP client
frontend/src/services/api.ts - Audio capture
frontend/src/hooks/useRecorder.ts - Realtime client
frontend/src/hooks/useRealtime.ts - Conversation display
frontend/src/components/ChatPanel.tsx - Avatar display
frontend/src/components/VideoPanel.tsx - Assessment panel
frontend/src/components/AssessmentPanel.tsx
Application backend
- Flask API
backend/src/app.py - Runtime configuration
backend/src/config.py - Scenario and agent management
backend/src/services/managers.py - Graph scenario generation
backend/src/services/graph_scenario_generator.py - Conversation analysis
backend/src/services/analyzers.py - Scenario utilities
backend/src/services/scenario_utils.py - Voice session proxy
backend/src/services/websocket_handler.py
AI services
- Azure Voice Live
- Azure AI models
- Azure Speech
Other components
- Sales trainee
Connections 26
- Sales trainee → Practice interface: practices in
README.md - Practice interface → Scenario selection: renders
frontend/src/app/App.tsx - Practice interface → Conversation display: renders
frontend/src/app/App.tsx - Practice interface → Avatar display: renders
frontend/src/app/App.tsx - Practice interface → Realtime client: starts session
frontend/src/app/App.tsx - Practice interface → Assessment panel: shows feedback
README.md - Scenario selection → Scenario editor: opens
frontend/src/components/ScenarioList.tsx - Practice interface → Scenario state: loads scenarios
frontend/src/app/App.tsx - Scenario state → HTTP client: fetches scenarios
frontend/src/hooks/useScenarios.ts - Scenario state → Custom scenarios: manages locally
frontend/src/hooks/useScenarios.ts - Scenario editor → Custom scenarios: edits scenarios
frontend/src/components/CustomScenarioEditor.tsx - Scenario selection → HTTP client: requests scenarios
frontend/src/components/ScenarioList.tsx - HTTP client → Flask API: calls API
frontend/src/services/api.ts - Flask API → Scenario and agent management: uses managers
backend/src/app.py - Flask API → Conversation analysis: runs analysis
backend/src/app.py - Flask API → Voice session proxy: routes voice socket
backend/src/app.py - Scenario and agent management → Graph scenario generation: uses generator
backend/src/services/managers.py - Conversation analysis → Scenario utilities: uses utilities
backend/src/services/analyzers.py - Voice session proxy → Scenario and agent management: looks up agent
backend/src/services/websocket_handler.py - Realtime client → Voice session proxy: streams messages
README.md - Audio capture → Realtime client: provides audio
- Voice session proxy → Azure Voice Live: proxies session
README.md - Conversation analysis → Azure AI models: scores conversation
README.md - Conversation analysis → Azure Speech: assesses pronunciation
README.md - Scenario and agent management → Azure AI models: creates optional agents
README.md - HTTP client → Assessment panel: returns assessment