Crate atomr_agents_tts_voice 

Conversation — bidirectional voice session built on top of the atomr_agents_stt_core::SpeechToText and atomr_agents_tts_core::TextToSpeech traits.

Two operating modes:

• ConversationMode::TurnBased — run STT on caller-supplied PCM, hand the transcript to the user-supplied ConversationAgent, synthesise the reply via TTS. Works with any STT + TTS pair.
• ConversationMode::UnifiedRealtime — open one atomr_agents_tts_core::RealtimeSession (e.g. OpenAI Realtime, Gemini Live, ElevenLabs Conversational AI) that both transcribes inbound audio and emits assistant audio. The agent is consulted only if the backend surfaces an [InboundTranscript] event and the conversation is configured for client-side responses.

This crate intentionally has no platform deps: mic capture and speaker playback live in atomr-agents-stt-audio and atomr-agents-tts-audio respectively. Consumers wire those up around Conversation.

Structs

Conversation

Bidirectional voice conversation.

ConversationOptions

NoopAgent

Enums

ConversationEvent

ConversationMode

Traits

ConversationAgent