Added
-
Added
OpenAITTSService. -
Allow passing
output_formatandmodel_idtoCartesiaTTSServiceto change audio sample format and the model to use. -
Added
DailyRESTHelperwhich helps you create Daily rooms and tokens in an easy way. -
PipelineTasknow has ahas_finished()method to indicate if the task has completed. If a task is never ranhas_finished()will return False. -
PipelineRunnernow supports SIGTERM. If received, the runner will be canceled.
Fixed
-
Fixed an issue where
BaseInputTransportandBaseOutputTransportwhere stopping push tasks before pushingEndFrameframes could cause the bots to get stuck. -
Fixed an error closing local audio transports.
-
Fixed an issue with Deepgram TTS that was introduced in the previous release.
-
Fixed
AnthropicLLMServiceinterruptions. If an interruption occurred, ausermessage could be appended after the previoususermessage. Anthropic does not allow that because it requires alternateuserandassistantmessages.
Performance
-
The
BaseInputTransportdoes not pull audio frames from sub-classes any more. Instead, sub-classes now push audio frames into a queue in the base class. Also,DailyInputTransportnow pushes audio frames every 20ms instead of 10ms. -
Remove redundant camera input thread from
DailyInputTransport. This should improve performance a little bit when processing participant videos. -
Load Cartesia voice on startup.