meshagent-livekit 0.6.1__py3-none-any.whl → 0.6.3__py3-none-any.whl

meshagent/livekit/agents/meeting_transcriber.py

@@ -15,6 +15,9 @@ from livekit.agents import (
     llm,
     utils,
 )
+from livekit.agents.stt import STT
+
+from meshagent.openai.proxy import get_client
 from livekit.plugins import openai, silero
 from meshagent.api import MeshDocument, SchemaRegistration, SchemaRegistry
 from meshagent.agents import SingleRoomAgent
@@ -83,7 +86,12 @@ class StopTranscriptionTool(Tool):
 
 
 class MeetingTranscriber(SingleRoomAgent):
-    def __init__(self, name: str, requires: Optional[list[Requirement]] = None):
+    def __init__(
+        self,
+        name: str,
+        requires: Optional[list[Requirement]] = None,
+        stt: Optional[STT] = None,
+    ):
         super().__init__(
             name=name,
             requires=requires,
@@ -95,6 +103,7 @@ class MeetingTranscriber(SingleRoomAgent):
                 StopTranscriptionTool(transcriber=self),
             ],
         )
+        self.stt = stt
         self._vad = None
         self._transcription_tasks = dict[str, tuple[asyncio.Task, asyncio.Future]]()
 
@@ -120,7 +129,12 @@ class MeetingTranscriber(SingleRoomAgent):
                 ) as conn:
                     doc = await self.room.sync.open(path=path, create=True)
 
-                    transcriber = MultiUserTranscriber(conn, doc, self._vad)
+                    stt = self.stt
+                    if stt is None:
+                        openai_client = get_client(room=self.room)
+                        stt = openai.STT(client=openai_client)
+
+                    transcriber = MultiUserTranscriber(conn, doc, self._vad, stt)
                     transcriber.start()
 
                     for participant in conn.livekit_room.remote_participants.values():
@@ -180,10 +194,10 @@ class TranscriptRegistry(SchemaRegistry):
 
 
 class Transcriber(Agent):
-    def __init__(self, *, participant, doc: MeshDocument):
+    def __init__(self, *, participant, doc: MeshDocument, stt):
         super().__init__(
             instructions="not-needed",
-            stt=openai.STT(),
+            stt=stt,
         )
         self.doc = doc
         self.participant = participant
@@ -205,10 +219,11 @@ class Transcriber(Agent):
 
 
 class MultiUserTranscriber:
-    def __init__(self, ctx: VoiceConnection, doc: MeshDocument, vad):
+    def __init__(self, ctx: VoiceConnection, doc: MeshDocument, vad, stt):
         self.ctx = ctx
         self.doc = doc
         self.vad = vad
+        self.stt = stt
         self._sessions: dict[str, AgentSession] = {}
         self._tasks: set[asyncio.Task] = set()
 
@@ -285,6 +300,7 @@ class MultiUserTranscriber:
             agent=Transcriber(
                 participant=participant,
                 doc=self.doc,
+                stt=self.stt,
             )
         )
         return session

meshagent/livekit/agents/voice.py

@@ -18,9 +18,15 @@ from meshagent.tools import ToolContext, Toolkit
 from livekit.agents import Agent, AgentSession, ChatContext
 from livekit.agents.llm import RawFunctionTool, ToolError, function_tool
 
-from openai import AsyncOpenAI
+from meshagent.openai.proxy import get_client
 from meshagent.agents import AgentChatContext
-from livekit.agents import BackgroundAudioPlayer, AudioConfig, BuiltinAudioClip
+from livekit.agents import (
+    BackgroundAudioPlayer,
+    AudioConfig,
+    BuiltinAudioClip,
+    RoomInputOptions,
+    RoomOutputOptions,
+)
 
 from livekit.plugins import openai, silero
 # from livekit.plugins.turn_detector.multilingual import MultilingualModel
@@ -32,7 +38,6 @@ from typing import Any
 
 
 from livekit import rtc
-from livekit.agents import RunContext
 
 from typing import Optional
 
@@ -270,13 +275,6 @@ class VoiceBot(SingleRoomAgent):
         )
 
     async def create_agent(self, *, context: ToolContext, session: AgentSession):
-        @function_tool
-        async def say(context: RunContext, text: str):
-            "says something out loud to the user"
-            logger.info(f"saying: {text}")
-            session.say(text)
-            return "success"
-
         ctx = ChatContext()
 
         initial_context = await self.init_chat_context()
@@ -287,7 +285,7 @@ class VoiceBot(SingleRoomAgent):
             chat_ctx=ctx,
             instructions="\n".join(self.rules),
             allow_interruptions=True,
-            tools=[*await self.make_function_tools(context=context), say],
+            tools=[*await self.make_function_tools(context=context)],
         )
 
         # agent = Agent(
@@ -305,23 +303,14 @@ class VoiceBot(SingleRoomAgent):
         return AgentChatContext()
 
     def create_session(self, *, context: ToolContext) -> AgentSession:
-        token: str = context.room.protocol.token
-        url: str = context.room.room_url
-
-        room_proxy_url = f"{url}/v1"
-
-        oaiclient = AsyncOpenAI(
-            api_key=token,
-            base_url=room_proxy_url,
-            default_headers={"Meshagent-Session": context.room.session_id},
-        )
+        oaiclient = get_client(room=self.room)
 
         session = AgentSession(
             max_tool_steps=50,
             allow_interruptions=True,
             vad=silero.VAD.load(),
             stt=openai.STT(client=oaiclient),
-            tts=openai.TTS(client=oaiclient, voice="echo"),
+            tts=openai.TTS(client=oaiclient),
             llm=openai.LLM(client=oaiclient),
             # turn_detection=MultilingualModel(),
         )
@@ -355,7 +344,18 @@ class VoiceBot(SingleRoomAgent):
                 room=connection.livekit_room, agent_session=session
             )
 
-            await session.start(agent=agent, room=connection.livekit_room)
+            await session.start(
+                agent=agent,
+                room=connection.livekit_room,
+                room_input_options=RoomInputOptions(
+                    text_enabled=False,
+                    delete_room_on_close=False,
+                ),
+                room_output_options=RoomOutputOptions(
+                    transcription_enabled=True,
+                    audio_enabled=True,
+                ),
+            )
 
             if self.auto_greet_prompt is not None:
                 session.generate_reply(user_input=self.auto_greet_prompt)

meshagent/livekit/version.py

@@ -1 +1 @@
-__version__ = "0.6.1"
+__version__ = "0.6.3"

{meshagent_livekit-0.6.1.dist-info → meshagent_livekit-0.6.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: meshagent-livekit
-Version: 0.6.1
+Version: 0.6.3
 Summary: Livekit support for Meshagent
 License-Expression: Apache-2.0
 Project-URL: Documentation, https://docs.meshagent.com
@@ -17,8 +17,8 @@ Requires-Dist: livekit-agents~=1.2
 Requires-Dist: livekit-plugins-openai~=1.2
 Requires-Dist: livekit-plugins-silero~=1.2
 Requires-Dist: livekit-plugins-turn-detector~=1.2
-Requires-Dist: meshagent-api~=0.6.1
-Requires-Dist: meshagent-tools~=0.6.1
+Requires-Dist: meshagent-api~=0.6.3
+Requires-Dist: meshagent-tools~=0.6.3
 Dynamic: license-file
 
 # [Meshagent](https://www.meshagent.com)

meshagent_livekit-0.6.3.dist-info/RECORD

@@ -0,0 +1,12 @@
+meshagent/livekit/__init__.py,sha256=X78Z4yEg5XfkNKH0HiIdG4k1q5ktB-ampTuXHLNFrAw,58
+meshagent/livekit/livekit_protocol.py,sha256=5Zu4ymLWEGt5SGXLNu94gOeyjnjhaV6uTS2FhSdODqs,1470
+meshagent/livekit/livekit_protocol_test.py,sha256=o7yYxXad4tMazcxFkq44yW-A9tJ0Lk6WdZpG5ifxcU4,2980
+meshagent/livekit/version.py,sha256=zYiFHqR7JwbvdK9dvKrh-RTNfUqjHUwC4CTcFAPVYLc,22
+meshagent/livekit/agents/meeting_transcriber.py,sha256=4B5Qi0vKP0TU00vR4KsNdIiR2uyvfgiOL2hjuzhLotw,10210
+meshagent/livekit/agents/transcriber.py,sha256=S992oVVBt3ShWDQQWprLjyl6Yh0hyNRd8d3qCmg_toU,5795
+meshagent/livekit/agents/voice.py,sha256=9aaQHPxnX86D2SasJn6SMiKIh1OX1dPmfQBow2tCvzE,11857
+meshagent_livekit-0.6.3.dist-info/licenses/LICENSE,sha256=eTt0SPW-sVNdkZe9PS_S8WfCIyLjRXRl7sUBWdlteFg,10254
+meshagent_livekit-0.6.3.dist-info/METADATA,sha256=KLQn05wjmiThdg4OgJKNcRJzO6xSGKct4qAZqf-Xo40,1749
+meshagent_livekit-0.6.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+meshagent_livekit-0.6.3.dist-info/top_level.txt,sha256=GlcXnHtRP6m7zlG3Df04M35OsHtNXy_DY09oFwWrH74,10
+meshagent_livekit-0.6.3.dist-info/RECORD,,

meshagent_livekit-0.6.1.dist-info/RECORD

@@ -1,12 +0,0 @@
-meshagent/livekit/__init__.py,sha256=X78Z4yEg5XfkNKH0HiIdG4k1q5ktB-ampTuXHLNFrAw,58
-meshagent/livekit/livekit_protocol.py,sha256=5Zu4ymLWEGt5SGXLNu94gOeyjnjhaV6uTS2FhSdODqs,1470
-meshagent/livekit/livekit_protocol_test.py,sha256=o7yYxXad4tMazcxFkq44yW-A9tJ0Lk6WdZpG5ifxcU4,2980
-meshagent/livekit/version.py,sha256=baAcEjLSYFIeNZF51tOMmA_zAMhN8HvKael-UU-Ruec,22
-meshagent/livekit/agents/meeting_transcriber.py,sha256=_DHFCuOGCE3LBBBDoF1xGTBMHXaVm2yzqya-OO4g6xg,9778
-meshagent/livekit/agents/transcriber.py,sha256=S992oVVBt3ShWDQQWprLjyl6Yh0hyNRd8d3qCmg_toU,5795
-meshagent/livekit/agents/voice.py,sha256=STgjMSqzUgV9UAmleOy1vkgRXP93MDSYgiOO6Lo0peU,11964
-meshagent_livekit-0.6.1.dist-info/licenses/LICENSE,sha256=eTt0SPW-sVNdkZe9PS_S8WfCIyLjRXRl7sUBWdlteFg,10254
-meshagent_livekit-0.6.1.dist-info/METADATA,sha256=_z2iTjyR9cJEo2Y_6IMJGSD5kKs6-cm7Cf23Qb7b2wo,1749
-meshagent_livekit-0.6.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-meshagent_livekit-0.6.1.dist-info/top_level.txt,sha256=GlcXnHtRP6m7zlG3Df04M35OsHtNXy_DY09oFwWrH74,10
-meshagent_livekit-0.6.1.dist-info/RECORD,,