Event-driven channels

Hume AI

Hume AI builds empathic voice and emotional AI models. The platform exposes four published APIs: Voices, Octave Text-to-Speech, Empathic Voice Interface (EVI / speech-to-speech), and Expression Measurement (multimodal emotion analysis). REST + WebSocket interfaces are documented with public OpenAPI

3 channels across 1 AsyncAPI spec · Provider profile

Channels

/chat

Real-time EVI chat. Client sends audio and control messages; server streams transcripts, assistant text, synthesized audio and tool events. Connection URL: `wss://api.hume.ai/v0/evi/chat`.

Hume AI WebSocket APIs
/chat/{chat_id}/connect

Secondary connection to an in-progress EVI chat. The original chat must have been opened with `allow_connection=true`. The secondary connection can send the same control-plane messages as `/chat` exce

Hume AI WebSocket APIs
/models

Streaming multimodal expression measurement inference. Connection URL: `wss://api.hume.ai/v0/stream/models`. Each client message includes a `models` configuration and the `data` payload (Base64-encode

Hume AI WebSocket APIs