Fields

FieldTypeDescription
message typeStringIndicates the product this event is associated with. For this use case, the message_type will be set to conversation.
conversation idStringThe unique identifier for the conversation.
event typeStringSpecifies the type of event being sent. This field is included in all events and helps distinguish between different event types.
modalityStringIndicates the input type for this event. Possible values: audio, text.
textStringIf modality is set to text, this field contains the text that the replica will speak out loud.
audioStringIf modality is set to audio, this field contains the base64-encoded audio that the replica will speak out loud. The encoded audio must be in 16-bit PCM 16000 format.
inference idStringA unique identifier for a given utterance. Multiple messages with the same inference_id indicate they belong to the same utterance.
doneBooleanIndicates if all audio chunks for an utterance have been sent. If done is not set to True, audio chunks will continue to stream until a final event with done set to True is received.
contextStringSpecifies the new conversational context that will be used by the LLM (Language Model).

Returned Values

This action does not return any fields.