Repository navigation
Conversation
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
🔍 Devin Review: 1 flag
Not posted on this PR by your GitHub settings — view it in Devin Review. (Configure)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
For a user
ChatMessage,metrics.started_speaking_atis the start of the whole user turn. A turn can span several VAD segments and stays open across a false-interruption resume, because STT can still return words for audio heard before the resume. So a turn's start can come before agent speech that played before the turn committed. Transcripts sorted bystarted_speaking_atthen place the user's reply ahead of the agent line it answers (#7063).This is expected behavior, so this PR documents it on
MetricsReport.started_speaking_atand points readers tostopped_speaking_atfor the order in which turns completed. To ignore speech such as a backchannel, filter STT events onSpeechEvent.speech_end_time, as adaptive interruption does.Docstring only; no behavior change.