Skip to content

[voice agent][Fix] for non-EOU ASR models#15467

Closed
stevehuang52 wants to merge 2 commits into
mainfrom
stevehuang52-patch-1
Closed

[voice agent][Fix] for non-EOU ASR models#15467
stevehuang52 wants to merge 2 commits into
mainfrom
stevehuang52-patch-1

Conversation

@stevehuang52
Copy link
Copy Markdown
Collaborator

@stevehuang52 stevehuang52 commented Mar 5, 2026

Important

The Update branch button must only be pressed in very rare occassions.
An outdated branch is never blocking the merge of a PR.
Please reach out to the automation team before pressing that button.

What does this PR do ?

Previously some additional logic was added to let VAD create user frames for non-EOU ASR models (e.g., nvidia/nemotron-speech-streaming-en-0.6b), which will cause the pipeline to send two UserStartedSpeakingFrame/UserStoppedSpeakingFrame, one by VAD and the other by TurnTakingService.

In this fix, we force to only use TurnTakingService to handle those signals, since the non-EOU ASR models can be treated as an ASR-EOU model with 100% miss rate, which is already handled in the TurnTakingService.

Signed-off-by: He Huang (Steve) <105218074+stevehuang52@users.noreply.github.com>
@stevehuang52 stevehuang52 requested a review from tango4j March 5, 2026 15:54
Signed-off-by: He Huang (Steve) <105218074+stevehuang52@users.noreply.github.com>
@stevehuang52 stevehuang52 deleted the stevehuang52-patch-1 branch March 5, 2026 19:22
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant