Skip to main content

Outbound: Tavus sends each streaming chunk so you can show what is being said in real time as the accumulated text grows during the turn.

Contrast with conversation.utterance, which delivers the full turn text at once rather than streaming chunks.

For PAL turns, duplicate app-messages are sent with properties.role of "pal" and legacy "replica" (same payload otherwise).

​
message_type
string

conversation for conversation-layer app messages.

Example:

"conversation"

​
event_type
string

conversation.utterance.streaming.

Example:

"conversation.utterance.streaming"

​
timestamp
number

Unix timestamp (seconds since epoch) indicating when this event was created. Use this to build timestamped transcripts or reconstruct the timeline of a conversation.

Example:

1746000000.123

​
seq
integer

Monotonic per event; a higher seq was emitted later. Use with turn_idx as in Event ordering and turn tracking.

Example:

42

​
conversation_id
string

The unique identifier for the conversation.

Example:

"c123456"

​
inference_id
string

A unique identifier for the PAL's current inference turn. Can be used to correlate with other events such as conversation.utterance and conversation.started_speaking.

Example:

"83294d9f-8306-491b-a284-791f56c8383f"

​
turn_idx
integer

Turn index; increments when conversation.respond is processed. Groups utterances, tool calls, and speaking-state events for the same turn. See Event ordering and turn tracking.

Example:

3

​
properties
object

PAL and user streaming share this shape; is_interrupted appears on PAL messages when relevant.