What's Changed
- azure speech sdk version upgrade (>=1.41.0 to >=1.43.0) by @jayeshp19 in #2008
- fix: resolve type mismatch between _TurnDetector protocol and plugins by @kechako in #2002
- feat: enhance openai LLM metrics to include cached prompt tokens by @theomonnom in #2013
- delete livekit-plugins-rag by @theomonnom in #1995
- fix: add missing FunctionToolsExecutedEvent export by @mateuszkulpa in #2016
- detect leaked tasks on tests by @theomonnom in #1993
- toxic proxy & bring back tts tests by @theomonnom in #1982
- fix 11labs timeout & tests by @theomonnom in #2017
- Emit errors from open ai realtime model by @Shubhrakanti in #1968
- fix prompt_cached_tokens in llm metrics by @longcw in #2024
- simplify google tts voice params by @jayeshp19 in #2025
- fix gemini live interruption by @jayeshp19 in #2029
- support with_azure method in openai stt/tts by @jayeshp19 in #2030
- add bey avatar plugin by @longcw in #2031
- avoid duplicate room io initialization by @longcw in #2037
- add wait_for_participant as a function by @longcw in #2039
- fix realtime default turn detection and transcription for azure openai by @jayeshp19 in #2041
- allow overriding eou threshold by @jeradf in #2035
- Speechify TTS Plugin by @chaerla in #2044
- Readme updates by @davidzhao in #2045
- Fix dangling websocket connections by @ChenghaoMou in #2027
- move wait_for_participant to utils by @longcw in #2047
- add recovery method for OAI realtime text response by @longcw in #2015
- fix Speechify tests by @theomonnom in #2051
- fix openai timeout & tests by @theomonnom in #2020
- Speechify TTS Plugin Improvements by @chaerla in #2056
- support raw function descriptions by @theomonnom in #2055
- add bithuman avatar plugin by @longcw in #2054
- Misc fixes by @davidzhao in #2049
- fix discarded jobs when launching processes by @theomonnom in #2059
- fix generate_reply instructions by @theomonnom in #2058
- keep exception chaining by @theomonnom in #1990
- Use ogg by default for speechify by @davidzhao in #2068
- fix(gemini live): avoid duplicating input by @davidzhao in #2069
- fix realtime interruption for VAD turn detection by @longcw in #2072
- reset stt when clear user turn by @longcw in #2070
- Bug Fix - wrong transition to SPEAKING when the agent is in fact listening by @MajorTal in #2075
- support default arguments on function tools by @theomonnom in #2076
- store audio-synchronized transcript in chat_ctx when interrupted by @theomonnom in #2071
- support RunContext on raw function tools by @theomonnom in #2077
- Bug fix in agent_activity.py: user_state events were not fired by @MajorTal in #2078
- Support gemini thinking budget. by @pmaldonado in #2060
- publish legacy transcription event for avatar by @longcw in #2074
- merge anyOf optional unions to type list by @theomonnom in #2079
- fix optional arguments on function tools by @theomonnom in #2080
New plugins
- livekit-plugins-speechify — https://speechify.com/text-to-speech-api/
- livekit-plugins-bithuman — https://bithuman.mintlify.app/api-reference/sdk/quick-start#1-livekit-agent
- livekit-plugins-bey — https://docs.bey.dev/integration/livekit
New Contributors
- @kechako made their first contribution in #2002
- @mateuszkulpa made their first contribution in #2016
- @chaerla made their first contribution in #2044
- @pmaldonado made their first contribution in #2060
Full Changelog: https://github.com/livekit/agents/compare/livekit-agents@1.0.13...livekit-agents@1.0.14