The 'no audio frame push' error in LiveKit occurs specifically in the Text-to-Speech (TTS) system when the synthesis process completes but no audio data is generated or pushed to the output. This error is raised in the TTS

ChunkedStream

class under the following conditions: • The TTS synthesis process runs to completion. • The system waits for all audio frames to be pushed and processed. • The final check shows that

pushed_duration() <= 0.0

, meaning no audio data was produced. Potential Causes • The TTS service failed to generate audio output for the provided text. • Network or connectivity issues prevented audio data from being received. • Problems with the TTS provider's response or audio encoding. • Issues with the audio processing pipeline that prevented frames from being properly pushed. Error Handling • If retries are available, the system will attempt the synthesis again after a retry interval. • If it’s the final retry attempt, the error is raised as an

APIError

with the message "no audio frames were pushed." This error is typically recoverable through retries, as it often represents temporary issues with the TTS synthesis process rather than permanent configuration problems. Sources: livekit-agents/livekit/agents/tts/tts.py