Pipecat C++ Client SDK 1.3.0
Connect native apps to Pipecat bots
Loading...
Searching...
No Matches
pipecat::rtvi::BotOutputData Struct Reference

#include <pipecat/rtvi/messages.h>

Public Attributes

std::string text
 The text.
 
std::string output_type
 
std::optional< int64_t > segment_id
 ID shared by updates of a text.
 
std::optional< bool > will_be_spoken
 Whether the bot will speak it.
 
std::optional< std::string > spoken_status
 
std::optional< SpokenProgressData > spoken_progress
 How much of the text the bot has spoken. Only for text it speaks.
 
std::optional< std::vector< std::string > > delivery
 

Detailed Description

Something the bot says or outputs, e.g. a sentence.

If the bot speaks the text, it's sent again as the bot speaks it, with the same segment_id and updated spoken_status and spoken_progress.

Member Data Documentation

◆ delivery

std::optional<std::vector<std::string> > pipecat::rtvi::BotOutputData::delivery

How the bot speaks the words it just spoke, e.g. {"warm", "emphasis"}, or empty in its usual voice. It can change from word to word, so it comes as the bot speaks, not with "new". Only from bots with expressive conversation (alpha).

◆ output_type

std::string pipecat::rtvi::BotOutputData::output_type

What kind of text it is, e.g. "sentence", "word", or "backchannel" for a short acknowledgment the bot says while the user talks, like "Mm-hmm.". Bots with expressive conversation (alpha) also send "expression_sound" for a sound the bot makes, with its name as the text, like "laugh", and "expression_pause" for a pause, with its length in milliseconds as the text, like "500", each in a single "completed" message once the bot makes it. Bots can have types of their own, like "code".

◆ spoken_status

std::optional<std::string> pipecat::rtvi::BotOutputData::spoken_status

Where the bot is in speaking it: "new" on the first message of a text, "in-progress" as the bot speaks each word (only with text-to-speech services that report word timings), and "completed" once the bot has spoken it all. Text the bot doesn't speak gets a single message, with "new" from bots on RTVI 2.2 or newer and no status from older ones.


The documentation for this struct was generated from the following file: