Skip to content

Assistant

Services
Assistant

Assistant

Image: Assistant / voice AI (placeholder).

The Assistant service powers the voice AI pipeline: STT (speech-to-text), LLM (language model), TTS (text-to-speech), optional image generator, and MCP (Model Context Protocol) servers. Users choose engines and models in settings; the assistant listens, sends to the LLM, and speaks replies.

What you see

  • State (AssistantState) — Is active; selected STT, LLM, TTS, image generator; selected model; Ollama model download state; MCP servers list; providers list.
  • Actions — AssistantSetIsActiveAction, AssistantSetSelectedSTTAction, AssistantSetSelectedLLMAction, AssistantSetSelectedTTSAction, AssistantSetSelectedImageGeneratorAction, AssistantSetSelectedModelAction, AssistantDownloadOllamaModelAction; AssistantStartListeningAction, AssistantStopListeningAction, AssistantToggleListeningAction; AssistantUpdateProvidersAction, AssistantAddMcpServerAction, AssistantToggleMcpServerAction, AssistantDeleteMcpServerAction, AssistantSyncMcpServersAction; AssistantReportAction for internal events.
  • Implementation — ubo_app/services/090-assistant/: engines_registry, mcp_servers, reducer, setup, ubo_handle; subproject ubo-service/ contains the voice pipeline (Piper, Vosk, LLM, tools). Uses engines from ubo_app/engines/.