Repository navigation

All pull requests

Labels (17)Milestones (0)New pull request

Open

(309)

Closed

(4,905)

Check for sentence boundaries at spaces in SimpleTextAggregator
#6709aconchillo opened Nov 10, 2026Contributor
1
fix(assemblyai): clear language steering when language_codes is updated to None
#6706pylrn opened Nov 10, 2026Contributor
1
Bound eval PCM ingress and caller audio buffers
#6705mattshep opened Nov 10, 2026Contributor
1
Preserve assistant phases when replaying Responses history
#6703Mohammed-AB opened Nov 10, 2026
1
End Thai sentences at spaces in text aggregation
#6701yoyoismee opened Nov 10, 2026
1
`feat: add NVIDIA Nemotron Transcribe STT service
#6700AnandaRaju742 opened Nov 10, 2026Contributor
1
Fix audio eval correctness and modality timing
#6699mattshep opened Nov 10, 2026Contributor
1
Carry selected text metadata from TTS input to spoken text frames
#6695Rutvij97 opened Nov 10, 2026
Settle a cancelled function call with what is known of its outcome
#6693armaan4658 opened Nov 10, 2026Contributor
Downmix stereo input in AudioBufferProcessor recordings
#6685GPT-AGI opened Nov 10, 2026Contributor
2
Keep timed Soniox agent text on its audio connection
#6664GoldArowana opened Nov 9, 2026Contributor
1
Keep the next sentence's words out of the sentence that just ended
#6650GPT-AGI opened Nov 9, 2026Contributor
fix(turns): keep the turn analyzer verdict when a transcript starts the turn late
#6648yakshaverdev opened Nov 9, 2026
1
feat: add support for Qwen3 TTS via faster qwen3-tts
#6633user414 opened Nov 9, 2026
fix(aggregator): flush deferred tool context push on user speech stop
#6631Speedyzyzz opened Nov 9, 2026
1
fix(summarization): keep running async tool calls out of the summary
#6630coder-abu opened Nov 9, 2026
Keep running async tool calls out of context summarization
#6629prassu017 opened Nov 9, 2026Contributor
Fix MoQ playback timing: wall-clock PTS across turn gaps, cut cancelled audio on interruption
#6627vipyne opened Nov 9, 2026Member
5
fix(processors): bound final cleanup and retain task ownership
#6624GoldArowana opened Nov 8, 2026Contributor
1
Tell the model a cancelled sync tool call has an unknown outcome
#6623vishit18 opened Nov 8, 2026
fix(llm): report a direct function without __name__ by its type
#6622lxcxjxhx opened Nov 8, 2026Contributor
1
fix(events): await results from callable event handlers
#6617xuilen opened Nov 8, 2026
2
Don't pad a speech stream at a TTSStoppedFrame
#6601rjp44 opened Nov 7, 2026Contributor
1
Add HeyGen Voice TTS services
#6597spro-heygen opened Nov 7, 2026
1
Install the UIWorker's screen reading on its LLM as an LLM extension
#6592kompfner opened Nov 6, 2026Contributor
7

Navigation Menu

Repository navigation

All pull requests

Check for sentence boundaries at spaces in SimpleTextAggregator

fix(assemblyai): clear language steering when language_codes is updated to None

Bound eval PCM ingress and caller audio buffers

Preserve assistant phases when replaying Responses history

End Thai sentences at spaces in text aggregation

`feat: add NVIDIA Nemotron Transcribe STT service

Fix audio eval correctness and modality timing

Carry selected text metadata from TTS input to spoken text frames

Settle a cancelled function call with what is known of its outcome

Downmix stereo input in AudioBufferProcessor recordings

Keep timed Soniox agent text on its audio connection

Keep the next sentence's words out of the sentence that just ended

fix(turns): keep the turn analyzer verdict when a transcript starts the turn late

feat: add support for Qwen3 TTS via faster qwen3-tts

fix(aggregator): flush deferred tool context push on user speech stop

fix(summarization): keep running async tool calls out of the summary

Keep running async tool calls out of context summarization

Fix MoQ playback timing: wall-clock PTS across turn gaps, cut cancelled audio on interruption

fix(processors): bound final cleanup and retain task ownership

Tell the model a cancelled sync tool call has an unknown outcome

fix(llm): report a direct function without name by its type

fix(events): await results from callable event handlers

Don't pad a speech stream at a TTSStoppedFrame

Add HeyGen Voice TTS services

Install the UIWorker's screen reading on its LLM as an LLM extension

Pull requests