Voice & AI

Phone-side cloud ASR/TTS bootstrap, offline glass speech features, AI streaming chat, plus how pieces stitch together end-to-end.

Pages

• Phone SDK init (ASR/TTS)
• Glass TTS & ASR
• Glass AI Chat

Reading order

1. Phone bootstrap — where hosted voice services negotiate credentials.
2. Glass lane — contrasts offline wakewords vs streamed ASR lanes.
3. AI Chat — ties microphones, cloud ASR responses, playback together.

Capability wiring

Feature	Responsibility	Typical host screen
Offline wake-word	Local hot-word triggers	HomeActivity
Offline TTS	Local narration	SendMessageActivity
Cloud ASR	Dictation-grade dictation	SendMessageActivity
Cloud TTS	Hosted speech synthesis	Provisioned phone init; stubs on glass
AI Chat	LLM streamed answers	SdkMediaActivity

End-to-end mental model

sequenceDiagram
    participant U as User
    participant G as Glass client
    participant A as ASR service
    participant AI as AI Chat
    participant T as TTS playback

    U->>G: Speak
    G->>A: Begin speech decode
    A-->>G: Partial/final hypotheses
    G->>AI: Forward user question text
    AI-->>G: Incremental narration tokens
    G->>T: Render reply as audible speech
    T-->>U: Playback