Skip to content

improve should_discard check #5676

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
chenghao-mou wants to merge 4 commits into
base: main
Choose a base branch
from
+28 −5
Open

improve should_discard check #5676

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
a8ce986
improve should_discard check
chenghao-mou May 7, 2026
36948ee
add a warning
chenghao-mou May 7, 2026
c44eab3
move warning to activity instead
chenghao-mou May 8, 2026
de46f74
move warning to activity instead
chenghao-mou May 8, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
33 changes: 28 additions & 5 deletions livekit-agents/livekit/agents/voice/agent_activity.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -219,6 +219,9 @@ def __init__(self, agent: Agent, sess: AgentSession) -> None:
# speeches that audio playout finished but not done because of tool calls
self._background_speeches: set[SpeechHandle] = set()

# skip_stt/discard_audio_if_uninterruptible warning
self._skip_stt_warning_started: bool = False

def _validate_turn_detection(
self, turn_detection: TurnDetectionMode | None
) -> TurnDetectionMode | None:
Expand Down Expand Up @@ -1003,20 +1006,40 @@ def push_audio(self, frame: rtc.AudioFrame) -> None:
if not self._started:
return

should_discard = (
self._current_speech
and not self._current_speech.allow_interruptions
and self._session.options.interruption["discard_audio_if_uninterruptible"]
) or (
aec_warmup_active: bool = (
self._session.agent_state == "speaking"
and self._session._aec_warmup_remaining > 0
and self._session._aec_warmup_timer is not None
)

speech_active: bool = (
self._current_speech is not None
and not self._current_speech.done()
and not self._current_speech.interrupted
and not self._current_speech.allow_interruptions
and self._session.options.interruption["discard_audio_if_uninterruptible"]
)

should_discard: bool = aec_warmup_active or speech_active

if not should_discard:
if self._rt_session is not None:
self._rt_session.push_audio(frame)

# warn once per continuous window
if should_discard and not self._skip_stt_warning_started:
self._skip_stt_warning_started = True
logger.info(
"stt audio discarding started",
extra={
"aec_warmup_active": aec_warmup_active,
"speech_active": speech_active,
},
)
elif not should_discard and self._skip_stt_warning_started:
self._skip_stt_warning_started = False
logger.info("stt audio discarding stopped")

# Always forward to _audio_recognition for VAD, even when discarding STT/LLM
# VAD needs frames to detect speech end and update user state correctly
if self._audio_recognition is not None:
Expand Down
Loading