What's Changed
- Upgrading Cartesia TTS default to Sonic 3 by @chongzluong in #4922
- (google stt): add denoiser support and explicit adaptation param by @tinalenguyen in #4918
- feat: add Telnyx STT and TTS plugins by @fmv1992 in #4665
- feat: add livekit-plugins-sambanova with LLM support by @mahimairaja in #4910
- skip adding run event when run result is done by @longcw in #4925
- guard against RuntimeError when restoring allow_interruptions in AgentTask by @longcw in #4930
- Add support for Gradium pronunciation ids. by @LaurentMazare in #4932
- feat: optimize wav decoding by @davidzhao in #4905
- fix: drain buffered log records before closing LogQueueListener by @longcw in #4928
- fix(voice): return ToolError for unknown function calls instead of si… by @yusuf-eren in #4935
- Update readme to include mcp and skill information by @Topherhindman in #4937
- fix: migrate HttpServer to AppRunner for proper connection lifecycle by @longcw in #4945
- ignore unknown tools from xai realtime by @longcw in #4941
- soniox stt: populate timing and confidence from token metadata by @longcw in #4939
- fix(openai): preserve non-instruction system messages in update_chat_ctx for realtime models by @longcw in #4942
- feat(openai): add gpt-realtime-1.5 to RealtimeModels by @yusuf-eren in #4947
- standardize language handling by @davidzhao in #4926
- fix: avoid blocking event loop with unconditional psutil call in _load_task by @msaelices in #4946
- add AEC warmup to suppress false interruptions on first speech by @longcw in #4813
- initial by @dhruvladia-sarvam in #4923
- fix asyncio.Future crash in console mode by @davidzhao in #4952
- fix(11labs): Default to original alignment for CJK scripts by @chenghao-mou in #4968
- support openai responses websocket mode by @tinalenguyen in #4931
- Keyframe Labs Plugin by @kradkfl in #4950
- hotfix: import issue in
agent_worker.pyby @kradkfl in #4970 - feat(stt): add keyterms parameter in Elevenlabs STT plugin by @Arjun-A-I in #4967
- feat(elevenlabs): report STT audio duration via RECOGNITION_USAGE events by @BkSouX in #4953
- Fix/sarvam tts update options language code by @yusuf-eren in #4957
- Fix: call playback started in sound device callback (console mode) by @chenghao-mou in #4958
- fix: close duplex wrapper and log listener on process start failure by @longcw in #4977
- feat(assemblyai): add u3-rt-pro model plus mid-stream updates, SpeechStarted, and ForceEndpoint support by @gsharp-aai in #4965
- feat(stt): add support for AssemblyAI u3-rt-pro model and mid-session updates by @russellmartin-livekit in #4961
- rename Language to LanguageCode by @theomonnom in #4981
- livekit-agents 1.4.4 by @theomonnom in #4982
New Contributors
- @fmv1992 made their first contribution in #4665
- @mahimairaja made their first contribution in #4910
- @yusuf-eren made their first contribution in #4935
- @Topherhindman made their first contribution in #4937
- @kradkfl made their first contribution in #4950
- @Arjun-A-I made their first contribution in #4967
- @BkSouX made their first contribution in #4953
- @gsharp-aai made their first contribution in #4965
Full Changelog: https://github.com/livekit/agents/compare/livekit-agents@1.4.3...livekit-agents@1.4.4