meshagent-livekit 0.0.37__py3-none-any.whl → 0.0.38__py3-none-any.whl

meshagent/livekit/__init__.py

@@ -1 +1,3 @@
-from .version import __version__
+from .version import __version__
+
+__all__ = [__version__]

meshagent/livekit/agents/transcriber.py

@@ -1,15 +1,8 @@
 import logging
 import asyncio
 
-from meshagent.api.schema_document import Element,Text
-from meshagent.api.room_server_client import RoomClient
-from meshagent.api.websocket_protocol import WebSocketClientProtocol
 
-
-import asyncio
 import os
-import logging
-import json
 
 from livekit import api
 
@@ -17,7 +10,6 @@ from livekit.agents import stt, transcription, utils
 from livekit.plugins import openai, silero
 from livekit import rtc
 from livekit.rtc import TranscriptionSegment
-from livekit.agents import utils
 from livekit.agents import stt as speech_to_text
 
 from meshagent.api.runtime import RuntimeDocument
@@ -26,7 +18,7 @@ from typing import Optional
 
 from meshagent.api.schema import MeshSchema
 
-from meshagent.api.schema import MeshSchema, ElementType, ChildProperty, ValueProperty
+from meshagent.api.schema import ElementType, ChildProperty, ValueProperty
 
 from meshagent.agents.agent import AgentCallContext
 from meshagent.agents import TaskRunner
@@ -37,151 +29,221 @@ logger = logging.getLogger("transcriber")
 transcription_schema = MeshSchema(
     root_tag_name="transcript",
     elements=[
-        ElementType(tag_name="transcript", description="a transcript", properties=[
-            ChildProperty(name="transcriptions", description="the transcript entries", child_tag_names=[ "speech" ])
-        ]),
-        ElementType(tag_name="speech", description="transcribed speech", properties=[
-            ValueProperty(name="text", description="the transcribed text", type="string"),
-            ValueProperty(name="startTime", description="the time of the start of this speech", type="number"),
-            ValueProperty(name="endTime", description="the time of th end of this speech", type="number"),
-            ValueProperty(name="participantId", description="the identity of the participant", type="string"),
-            ValueProperty(name="participantName", description="the name of the participant", type="string")
-        ])
-    ]
+        ElementType(
+            tag_name="transcript",
+            description="a transcript",
+            properties=[
+                ChildProperty(
+                    name="transcriptions",
+                    description="the transcript entries",
+                    child_tag_names=["speech"],
+                )
+            ],
+        ),
+        ElementType(
+            tag_name="speech",
+            description="transcribed speech",
+            properties=[
+                ValueProperty(
+                    name="text", description="the transcribed text", type="string"
+                ),
+                ValueProperty(
+                    name="startTime",
+                    description="the time of the start of this speech",
+                    type="number",
+                ),
+                ValueProperty(
+                    name="endTime",
+                    description="the time of th end of this speech",
+                    type="number",
+                ),
+                ValueProperty(
+                    name="participantId",
+                    description="the identity of the participant",
+                    type="string",
+                ),
+                ValueProperty(
+                    name="participantName",
+                    description="the name of the participant",
+                    type="string",
+                ),
+            ],
+        ),
+    ],
 )
 
-class Transcriber(TaskRunner):
 
-    def __init__(self, *, livekit_url: Optional[str] = None, livekit_api_key: Optional[str] = None, livekit_api_secret: Optional[str] = None, livekit_identity: Optional[str] = None):
+class Transcriber(TaskRunner):
+    def __init__(
+        self,
+        *,
+        livekit_url: Optional[str] = None,
+        livekit_api_key: Optional[str] = None,
+        livekit_api_secret: Optional[str] = None,
+        livekit_identity: Optional[str] = None,
+    ):
         super().__init__(
             name="livekit.transcriber",
             title="transcriber",
             description="connects to a livekit room and transcribes the conversation",
             input_schema={
-                "type" : "object",
-                "additionalProperties" : False,
-                "required" : [ "room_name", "path" ],
-                "properties" : {
-                    "room_name" : {
-                        "type": "string"
-                    },
-                    "path" : {
-                        "type" : "string"
-                    }
-                }
+                "type": "object",
+                "additionalProperties": False,
+                "required": ["room_name", "path"],
+                "properties": {
+                    "room_name": {"type": "string"},
+                    "path": {"type": "string"},
+                },
             },
             output_schema={
-                "type" : "object",
-                "additionalProperties" : False,
-                "required" : [],
-                "properties" : {
-                }
-            }
+                "type": "object",
+                "additionalProperties": False,
+                "required": [],
+                "properties": {},
+            },
         )
         self._livekit_url = livekit_url
         self._livekit_api_key = livekit_api_key
         self._livekit_api_secret = livekit_api_secret
         self._livekit_identity = livekit_identity
 
-    async def _transcribe_participant(self, doc: RuntimeDocument, room: rtc.Room, participant: rtc.RemoteParticipant,
-            stt_stream: stt.SpeechStream, stt_forwarder: transcription.STTSegmentsForwarder
-        ):
-            logger.info("transcribing participant %s", participant.sid)
-            """Forward the transcription to the client and log the transcript in the console"""
-            async for ev in stt_stream:
-                logger.info("event from participant %s %s", participant.sid, ev)
-                             
-                if ev.type == stt.SpeechEventType.FINAL_TRANSCRIPT:
-                    logger.info("transcript: %s", ev.alternatives[0].text)
-                    if len(ev.alternatives) > 0:
-                        alt = ev.alternatives[0]                                                
-                        doc.root.append_child(tag_name="speech", attributes={ "text":  alt.text, "startTime": alt.start_time, "endTime" : alt.end_time, "participantId" : participant.identity, "participantName" : participant.name })
-
-            logger.info("done forwarding %s", participant.sid)
+    async def _transcribe_participant(
+        self,
+        doc: RuntimeDocument,
+        room: rtc.Room,
+        participant: rtc.RemoteParticipant,
+        stt_stream: stt.SpeechStream,
+        stt_forwarder: transcription.STTSegmentsForwarder,
+    ):
+        logger.info("transcribing participant %s", participant.sid)
+        """Forward the transcription to the client and log the transcript in the console"""
+        async for ev in stt_stream:
+            logger.info("event from participant %s %s", participant.sid, ev)
+
+            if ev.type == stt.SpeechEventType.FINAL_TRANSCRIPT:
+                logger.info("transcript: %s", ev.alternatives[0].text)
+                if len(ev.alternatives) > 0:
+                    alt = ev.alternatives[0]
+                    doc.root.append_child(
+                        tag_name="speech",
+                        attributes={
+                            "text": alt.text,
+                            "startTime": alt.start_time,
+                            "endTime": alt.end_time,
+                            "participantId": participant.identity,
+                            "participantName": participant.name,
+                        },
+                    )
+
+        logger.info("done forwarding %s", participant.sid)
 
     def should_transcribe(self, p: rtc.Participant) -> bool:
         # don't transcribe other agents
         # todo: maybe have a better way to detect
         return ".agent" not in p.identity
-    
+
     async def _wait_for_disconnect(self, room: rtc.Room):
         disconnected = asyncio.Future()
+
         def on_disconnected(_):
             disconnected.set_result(True)
+
         room.on("disconnected", on_disconnected)
 
         logger.info("waiting for disconnection")
-        await disconnected                 
-
+        await disconnected
 
     async def ask(self, *, context: AgentCallContext, arguments: dict):
         logger.info("Transcriber connecting to %s", arguments)
         output_path = arguments["path"]
         room_name = arguments["room_name"]
 
-        client = context.room        
+        client = context.room
         doc = await client.sync.open(path=output_path)
         try:
+            vad = silero.VAD.load()
+            utils.http_context._new_session_ctx()
 
-            vad = silero.VAD.load()                
-            utils.http_context._new_session_ctx()              
-            
             pending_tasks = list()
-            participantNames = dict[str,str]()
+            participantNames = dict[str, str]()
 
             sst_provider = openai.STT()
-            #sst_provider = fal.WizperSTT()
+            # sst_provider = fal.WizperSTT()
 
             room_options = rtc.RoomOptions(auto_subscribe=False)
 
             room = rtc.Room()
 
-            url = self._livekit_url if self._livekit_url is not None else os.getenv('LIVEKIT_URL')
-            api_key = self._livekit_api_key if self._livekit_api_key is not None else os.getenv('LIVEKIT_API_KEY')
-            api_secret = self._livekit_api_secret if self._livekit_api_secret is not None else os.getenv('LIVEKIT_API_SECRET')
-            identity = self._livekit_identity if self._livekit_identity is not None else os.getenv('AGENT_IDENTITY')
-
-            token = api.AccessToken(api_key=api_key, api_secret=api_secret) \
-                .with_identity(identity) \
-                .with_name("Agent") \
-                .with_kind("agent") \
-                .with_grants(api.VideoGrants(
-                    can_update_own_metadata=True,
-                    room_join=True,
-                    room=room_name,
-                    agent=True
-                ))
-            
+            url = (
+                self._livekit_url
+                if self._livekit_url is not None
+                else os.getenv("LIVEKIT_URL")
+            )
+            api_key = (
+                self._livekit_api_key
+                if self._livekit_api_key is not None
+                else os.getenv("LIVEKIT_API_KEY")
+            )
+            api_secret = (
+                self._livekit_api_secret
+                if self._livekit_api_secret is not None
+                else os.getenv("LIVEKIT_API_SECRET")
+            )
+            identity = (
+                self._livekit_identity
+                if self._livekit_identity is not None
+                else os.getenv("AGENT_IDENTITY")
+            )
+
+            token = (
+                api.AccessToken(api_key=api_key, api_secret=api_secret)
+                .with_identity(identity)
+                .with_name("Agent")
+                .with_kind("agent")
+                .with_grants(
+                    api.VideoGrants(
+                        can_update_own_metadata=True,
+                        room_join=True,
+                        room=room_name,
+                        agent=True,
+                    )
+                )
+            )
+
             jwt = token.to_jwt()
-            
+
             await room.connect(url=url, token=jwt, options=room_options)
-            
 
-            logger.info(
-                "connected to room: %s",
-                room_name
-            )
-        
+            logger.info("connected to room: %s", room_name)
+
             audio_streams = list[rtc.AudioStream]()
 
-            async def transcribe_track(participant: rtc.RemoteParticipant, track: rtc.Track):
+            async def transcribe_track(
+                participant: rtc.RemoteParticipant, track: rtc.Track
+            ):
                 audio_stream = rtc.AudioStream(track)
                 stt_forwarder = transcription.STTSegmentsForwarder(
                     room=room, participant=participant, track=track
                 )
 
                 audio_streams.append(audio_stream)
-                
+
                 stt = sst_provider
-                if not sst_provider.capabilities.streaming:                
+                if not sst_provider.capabilities.streaming:
                     stt = speech_to_text.StreamAdapter(
                         stt=stt,
                         vad=vad,
                     )
-                
+
                 stt_stream = stt.stream()
 
-                pending_tasks.append(asyncio.create_task(self._transcribe_participant(doc, room, participant, stt_stream, stt_forwarder)))
+                pending_tasks.append(
+                    asyncio.create_task(
+                        self._transcribe_participant(
+                            doc, room, participant, stt_stream, stt_forwarder
+                        )
+                    )
+                )
 
                 async for ev in audio_stream:
                     stt_stream.push_frame(ev.frame)
@@ -189,49 +251,67 @@ class Transcriber(TaskRunner):
             def subscribe_if_needed(pub: rtc.RemoteTrackPublication):
                 if pub.kind == rtc.TrackKind.KIND_AUDIO:
                     pub.set_subscribed(True)
-                
+
             for p in room.remote_participants.values():
                 participantNames[p.identity] = p.name
                 if self.should_transcribe(p):
                     for pub in p.track_publications.values():
                         subscribe_if_needed(pub)
-            
+
             first_parts = dict[str, rtc.Participant]()
-        
-            def on_transcript_event(segments: list[TranscriptionSegment], part: rtc.Participant | None, pub: rtc.TrackPublication | None = None) -> None:
+
+            def on_transcript_event(
+                segments: list[TranscriptionSegment],
+                part: rtc.Participant | None,
+                pub: rtc.TrackPublication | None = None,
+            ) -> None:
                 nonlocal room
                 logger.info("Got transcription segment %s %s %s", segments, part, pub)
                 for segment in segments:
-                    if segment.id not in first_parts and part != None:
+                    if segment.id not in first_parts and part is not None:
                         first_parts[segment.id] = part
 
                     if segment.final:
-                        if part == None and segment.id in first_parts:
+                        if part is None and segment.id in first_parts:
                             part = first_parts[segment.id]
                             first_parts.pop(segment.id)
-                        
-                        if part != None:
-                            doc.root.append_child(tag_name="speech", attributes={ "text":  segment.text, "startTime": segment.start_time, "endTime" : segment.end_time, "participantId" : part.identity, "participantName" : part.name })
+
+                        if part is not None:
+                            doc.root.append_child(
+                                tag_name="speech",
+                                attributes={
+                                    "text": segment.text,
+                                    "startTime": segment.start_time,
+                                    "endTime": segment.end_time,
+                                    "participantId": part.identity,
+                                    "participantName": part.name,
+                                },
+                            )
                         else:
-                            logger.warning("transcription was missing participant information")
+                            logger.warning(
+                                "transcription was missing participant information"
+                            )
 
             def on_participant_connected(p: rtc.RemoteParticipant):
                 participantNames[p.identity] = p.name
 
-            def on_track_published(pub: rtc.RemoteTrackPublication, p: rtc.RemoteParticipant):
+            def on_track_published(
+                pub: rtc.RemoteTrackPublication, p: rtc.RemoteParticipant
+            ):
                 if self.should_transcribe(p):
                     subscribe_if_needed(pub)
 
             subscriptions = dict()
-            
-            def on_track_unpublished(pub: rtc.RemoteTrackPublication, p: rtc.RemoteParticipant):
+
+            def on_track_unpublished(
+                pub: rtc.RemoteTrackPublication, p: rtc.RemoteParticipant
+            ):
                 if pub in subscriptions:
                     logger.info("track unpublished, stopping transcription")
                     # todo: maybe could be more graceful
                     subscriptions[pub].cancel()
                     subscriptions.pop(pub)
-                
-            
+
             def on_track_subscribed(
                 track: rtc.Track,
                 publication: rtc.TrackPublication,
@@ -239,22 +319,23 @@ class Transcriber(TaskRunner):
             ):
                 if track.kind == rtc.TrackKind.KIND_AUDIO:
                     logger.info("transcribing track %s", track.sid)
-                    track_task = asyncio.create_task(transcribe_track(participant, track))
+                    track_task = asyncio.create_task(
+                        transcribe_track(participant, track)
+                    )
+
                     def on_transcription_done(t):
                         try:
                             t.result()
                         except Exception as e:
                             logger.error("Transcription failed", exc_info=e)
 
-
                     track_task.add_done_callback(on_transcription_done)
                     pending_tasks.append(track_task)
                     subscriptions[publication] = track_task
-            
 
             for p in room.remote_participants.values():
                 on_participant_connected(p)
-                            
+
             room.on("participant_connected", on_participant_connected)
 
             room.on("track_published", on_track_published)
@@ -263,16 +344,15 @@ class Transcriber(TaskRunner):
             room.on("transcription_received", on_transcript_event)
 
             await self._wait_for_disconnect(room)
-                        
+
             logger.info("waited for termination")
             await room.disconnect()
-            
+
             logger.info("closing audio streams")
-            
+
             for stream in audio_streams:
                 await stream.aclose()
 
-    
             logger.info("waiting for pending tasks")
             gather_future = asyncio.gather(*pending_tasks)
 
@@ -280,7 +360,7 @@ class Transcriber(TaskRunner):
             try:
                 await gather_future
             except Exception as e:
-                if isinstance(e, asyncio.CancelledError) == False:
+                if not isinstance(e, asyncio.CancelledError):
                     logger.warning("Did not shut down cleanly", exc_info=e)
                 pass
 
@@ -288,11 +368,10 @@ class Transcriber(TaskRunner):
         except Exception as e:
             logger.info("Transcription failed", exc_info=e)
         finally:
-
             await utils.http_context._close_http_ctx()
             logger.info("Transcription done")
 
             await asyncio.sleep(5)
             await client.sync.close(path=output_path)
-            
+
             return {}

meshagent/livekit/agents/voice.py

@@ -2,7 +2,15 @@ import logging
 import asyncio
 from asyncio import CancelledError
 
-from meshagent.api import RoomMessage, ErrorResponse, JsonResponse, FileResponse, Requirement, Participant, JsonResponse, EmptyResponse, TextResponse
+from meshagent.api import (
+    RoomMessage,
+    ErrorResponse,
+    Requirement,
+    Participant,
+    JsonResponse,
+    EmptyResponse,
+    TextResponse,
+)
 from meshagent.api.room_server_client import RoomClient
 
 from meshagent.agents import ToolResponseAdapter
@@ -14,34 +22,23 @@ from openai import AsyncOpenAI
 from meshagent.agents import AgentChatContext
 from livekit.agents import BackgroundAudioPlayer, AudioConfig, BuiltinAudioClip
 
-from typing import Annotated
 from livekit.plugins import openai, silero
-#from livekit.plugins.turn_detector.multilingual import MultilingualModel
-import uuid
-import asyncio
-import logging
+# from livekit.plugins.turn_detector.multilingual import MultilingualModel
 
-import os
 
-import json 
+import json
 
 from typing import Any
 
-from livekit.plugins import openai
 
-from livekit.plugins import openai, silero
 from livekit import rtc
-from livekit.agents import Agent, AgentSession, RunContext
+from livekit.agents import RunContext
 
 from typing import Optional
 
-from copy import deepcopy
-
-from meshagent.api.schema_util import merge, prompt_schema
 
 from meshagent.agents import SingleRoomAgent
 
-from livekit.plugins.turn_detector.multilingual import MultilingualModel
 
 import re
 
@@ -52,7 +49,7 @@ def _replace_non_matching(text: str, allowed_chars: str, replacement: str) -> st
     """
     Replaces every character in `text` that does not match the given
     `allowed_chars` regex set with `replacement`.
-    
+
     Parameters:
     -----------
     text : str
@@ -62,7 +59,7 @@ def _replace_non_matching(text: str, allowed_chars: str, replacement: str) -> st
         For example, "a-zA-Z0-9" will keep only letters and digits.
     replacement : str
         The string to replace non-matching characters with.
-        
+
     Returns:
     --------
     str
@@ -72,35 +69,40 @@ def _replace_non_matching(text: str, allowed_chars: str, replacement: str) -> st
     pattern = rf"[^{allowed_chars}]"
     return re.sub(pattern, replacement, text)
 
+
 def safe_tool_name(name: str):
     return _replace_non_matching(name, "a-zA-Z0-9_-", "_")
 
+
 class VoiceConnection:
     def __init__(self, *, room: RoomClient, breakout_room: str):
         self.room = room
         self.breakout_room = breakout_room
-     
-    async def __aenter__(self):
 
+    async def __aenter__(self):
         client = self.room
-    
+
         room_options = rtc.RoomOptions(auto_subscribe=True)
 
         room = rtc.Room()
 
         self.livekit_room = room
 
-        connection_info = await client.livekit.get_connection_info(breakout_room=self.breakout_room)
-        
-        await room.connect(url=connection_info.url, token=connection_info.token, options=room_options)
+        connection_info = await client.livekit.get_connection_info(
+            breakout_room=self.breakout_room
+        )
+
+        await room.connect(
+            url=connection_info.url, token=connection_info.token, options=room_options
+        )
 
         return self
-    
+
     async def __aexit__(self, exc_type, exc, tb):
         await self.livekit_room.disconnect()
 
-class VoiceBot(SingleRoomAgent):
 
+class VoiceBot(SingleRoomAgent):
     def __init__(
         self,
         name: str,
@@ -112,20 +114,20 @@ class VoiceBot(SingleRoomAgent):
         auto_greet_prompt: Optional[str] = None,
         tool_adapter: ToolResponseAdapter = None,
         toolkits: list[Toolkit] = None,
-        requires: list[Requirement] = None
+        requires: list[Requirement] = None,
     ):
-        if toolkits == None:
+        if toolkits is None:
             toolkits = []
 
         self.toolkits = toolkits
-        
-        if rules == None:
-            rules = [ "You are a helpful assistant communicating through voice." ]
-        
+
+        if rules is None:
+            rules = ["You are a helpful assistant communicating through voice."]
+
         self.tool_adapter = tool_adapter
         self.auto_greet_message = auto_greet_message
         self.auto_greet_prompt = auto_greet_prompt
-       
+
         self.rules = rules
 
         super().__init__(
@@ -133,7 +135,7 @@ class VoiceBot(SingleRoomAgent):
             description=description,
             title=title,
             labels=labels,
-            requires=requires
+            requires=requires,
         )
 
     async def start(self, *, room):
@@ -141,66 +143,75 @@ class VoiceBot(SingleRoomAgent):
         await room.local_participant.set_attribute("supports_voice", True)
         await room.messaging.enable()
         room.messaging.on("message", self.on_message)
-    
+
     def on_message(self, message: RoomMessage):
         if message.type == "voice_call":
             breakout_room = message.message["breakout_room"]
-            
+
             logger.info(f"joining breakout room {breakout_room}")
-            
+
             def on_done(task: asyncio.Task):
                 try:
                     task.result()
-                except CancelledError as e:
-                        pass
+                except CancelledError:
+                    pass
                 except Exception as e:
                     logger.error(f"{e}", exc_info=e)
 
             for participant in self.room.messaging.remote_participants:
-
                 if participant.id == message.from_participant_id:
-
-                    task = asyncio.create_task(self.run_voice_agent(participant=participant, breakout_room=breakout_room))
+                    task = asyncio.create_task(
+                        self.run_voice_agent(
+                            participant=participant, breakout_room=breakout_room
+                        )
+                    )
                     task.add_done_callback(on_done)
                     return
-                
+
             logger.error(f"unable to find participant {message.from_participant_id}")
-                
-    
+
     async def _wait_for_disconnect(self, room: rtc.Room):
         disconnected = asyncio.Future()
+
         def on_disconnected(_):
             disconnected.set_result(True)
+
         room.on("disconnected", on_disconnected)
 
         logger.info("waiting for disconnection")
         await disconnected
 
     async def make_function_tools(self, *, context: ToolContext):
-
-        toolkits = [
-            *await self.get_required_toolkits(context=context),
-            *self.toolkits
-        ]
+        toolkits = [*await self.get_required_toolkits(context=context), *self.toolkits]
 
         tools = []
 
         for toolkit in toolkits:
-
             for tool in toolkit.tools:
-                
-                tools.append(self._make_function_tool(toolkits, context, tool.name, tool.description, tool.input_schema))
+                tools.append(
+                    self._make_function_tool(
+                        toolkits,
+                        context,
+                        tool.name,
+                        tool.description,
+                        tool.input_schema,
+                    )
+                )
 
         return tools
 
     def _make_function_tool(
-        self, toolkits: list[Toolkit], context: ToolContext, name: str, description: str | None, input_schema: dict
+        self,
+        toolkits: list[Toolkit],
+        context: ToolContext,
+        name: str,
+        description: str | None,
+        input_schema: dict,
     ) -> RawFunctionTool:
-        
         name = safe_tool_name(name)
+
         async def _tool_called(raw_arguments: dict) -> Any:
             try:
-
                 tool = None
                 for toolkit in toolkits:
                     for t in toolkit.tools:
@@ -208,9 +219,7 @@ class VoiceBot(SingleRoomAgent):
                             tool = t
 
                 if tool is None:
-                    raise ToolError(
-                    f"Could not find tool {name}"
-                    )
+                    raise ToolError(f"Could not find tool {name}")
 
                 try:
                     logger.info(f"executing tool {name}: {raw_arguments}")
@@ -218,32 +227,31 @@ class VoiceBot(SingleRoomAgent):
                 except Exception as e:
                     logger.error(f"failed to call tool {tool.name}: {e}")
                     return ToolError("f{e}")
-                if self.tool_adapter == None:
-
+                if self.tool_adapter is None:
                     if isinstance(tool_result, ErrorResponse):
                         raise ToolError(tool_result.text)
 
                     if isinstance(tool_result, JsonResponse):
                         return json.dumps(tool_result.json)
-                    
+
                     if isinstance(tool_result, TextResponse):
                         return tool_result.text
-        
+
                     if isinstance(tool_result, EmptyResponse):
                         return "success"
-                    
-                    if tool_result == None:
+
+                    if tool_result is None:
                         return "success"
-                    
 
                     raise ToolError(
                         f"Tool '{name}' returned an unexpected result {type(tool_result)}, attach a tool response adapter"
                     )
 
                 else:
-                    
-                    text = await self.tool_adapter.to_plain_text(room=context.room, response=tool_result)
-                    if text == None:
+                    text = await self.tool_adapter.to_plain_text(
+                        room=context.room, response=tool_result
+                    )
+                    if text is None:
                         text = "success"
                     return text
 
@@ -251,14 +259,17 @@ class VoiceBot(SingleRoomAgent):
                 logger.error("unable to call tool", exc_info=e)
                 raise
 
-
         return function_tool(
             _tool_called,
-            raw_schema={"name": name, "description": description, "strict" : True, "parameters": input_schema},
+            raw_schema={
+                "name": name,
+                "description": description,
+                "strict": True,
+                "parameters": input_schema,
+            },
         )
 
     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"
@@ -266,8 +277,8 @@ class VoiceBot(SingleRoomAgent):
             session.say(text)
             return "success"
 
-        ctx=ChatContext()
-    
+        ctx = ChatContext()
+
         initial_context = await self.init_chat_context()
         for message in initial_context.messages:
             ctx.add_message(role=message["role"], content=message["content"])
@@ -276,10 +287,7 @@ 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), say],
         )
 
         # agent = Agent(
@@ -291,88 +299,77 @@ class VoiceBot(SingleRoomAgent):
         #    tts=openai.TTS(),
         #    vad=silero.VAD.load(),
         #   allow_interruptions=True
-        #)      
+        # )
 
     async def init_chat_context(self) -> AgentChatContext:
-        return AgentChatContext() 
-    
+        return AgentChatContext()
+
     def create_session(self, *, context: ToolContext) -> AgentSession:
+        token: str = context.room.protocol.token
+        url: str = context.room.room_url
 
-        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
-            }
+            default_headers={"Meshagent-Session": context.room.session_id},
         )
 
         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"
-            ),
-            llm=openai.LLM(
-                client=oaiclient
-            ),
-            #turn_detection=MultilingualModel(),
-
+            stt=openai.STT(client=oaiclient),
+            tts=openai.TTS(client=oaiclient, voice="echo"),
+            llm=openai.LLM(client=oaiclient),
+            # turn_detection=MultilingualModel(),
         )
         return session
-    
-    
+
     async def run_voice_agent(self, *, participant: Participant, breakout_room: str):
-        
-        async with VoiceConnection(room=self.room, breakout_room=breakout_room) as connection:
-            
+        async with VoiceConnection(
+            room=self.room, breakout_room=breakout_room
+        ) as connection:
             logger.info("starting voice agent")
 
             context = ToolContext(
                 room=self.room,
                 caller=self.room.local_participant,
-                on_behalf_of=participant
+                on_behalf_of=participant,
             )
 
             session = self.create_session(context=context)
 
             agent = await self.create_agent(context=context, session=session)
-           
+
             background_audio = BackgroundAudioPlayer(
                 thinking_sound=[
-                    #AudioConfig(
+                    # AudioConfig(
                     #    os.path.dirname(os.path.abspath(__file__)) +"/sfx/thinking.mp3", volume=0.2),
                     AudioConfig(BuiltinAudioClip.KEYBOARD_TYPING, volume=0.3),
                     AudioConfig(BuiltinAudioClip.KEYBOARD_TYPING2, volume=0.4),
                 ],
             )
-            await background_audio.start(room=connection.livekit_room, agent_session=session)
+            await background_audio.start(
+                room=connection.livekit_room, agent_session=session
+            )
 
             await session.start(agent=agent, room=connection.livekit_room)
 
-            if self.auto_greet_prompt != None:
+            if self.auto_greet_prompt is not None:
                 session.generate_reply(user_input=self.auto_greet_prompt)
 
-            if self.auto_greet_message != None:
+            if self.auto_greet_message is not None:
                 session.say(self.auto_greet_message)
-            
+
             logger.info("started voice agent")
             await self._wait_for_disconnect(room=connection.livekit_room)
-            
-
 
-    
 
 class Voicebot(VoiceBot):
     def __init__(self, **kwargs):
-        logger.warning("Voicebot is deprecated, use VoiceBot instead. This class will be removed in a future release.")
+        logger.warning(
+            "Voicebot is deprecated, use VoiceBot instead. This class will be removed in a future release."
+        )
         super().__init__(**kwargs)

meshagent/livekit/livekit_protocol.py

@@ -14,35 +14,37 @@ class LivekitProtocol(Protocol):
         self.remote = remote
         self.topic = topic
 
-
     async def __aenter__(self):
         self.room.on("data_received", self._on_data_packet)
         return await super().__aenter__()
-    
+
     async def __aexit__(self, exc_type, exc, tb):
         self.room.off("data_received", self._on_data_packet)
-    
-        return await super().__aexit__(exc_type, exc, tb)
 
-    async def send_packet(self, data:bytes) -> None:
+        return await super().__aexit__(exc_type, exc, tb)
 
-        logger.info("sending data packet %s  %s to %s", self.topic, self.remote.identity, self.room.remote_participants[self.remote.identity].sid)
+    async def send_packet(self, data: bytes) -> None:
+        logger.info(
+            "sending data packet %s  %s to %s",
+            self.topic,
+            self.remote.identity,
+            self.room.remote_participants[self.remote.identity].sid,
+        )
 
         await self.local.publish_data(
             payload=data,
-            topic = self.topic,
-            reliable = True,
-            destination_identities = [ self.remote.identity ],
+            topic=self.topic,
+            reliable=True,
+            destination_identities=[self.remote.identity],
         )
 
     def _on_data_packet(self, evt: rtc.DataPacket):
-        
         if self.remote != evt.participant:
             return
 
-        logger.info("received data packet %s from %s", evt.topic, evt.participant.identity)
+        logger.info(
+            "received data packet %s from %s", evt.topic, evt.participant.identity
+        )
 
         if evt.topic == self.topic:
             self.receive_packet(evt.data)
-
-     

meshagent/livekit/livekit_protocol_test.py

@@ -15,60 +15,76 @@ import asyncio
 
 logger = logging.getLogger(__name__)
 
+
 @pytest.mark.asyncio
 async def test_protocol():
+    url = os.getenv("LIVEKIT_URL")
+    api_key = os.getenv("LIVEKIT_API_KEY")
+    api_secret = os.getenv("LIVEKIT_API_SECRET")
+
+    token1 = (
+        api.AccessToken(api_key=api_key, api_secret=api_secret)
+        .with_identity("core:user.test.agent-send")
+        .with_name("Agent")
+        .with_kind("agent")
+        .with_grants(
+            api.VideoGrants(
+                can_update_own_metadata=True,
+                room_join=True,
+                room="test-process",
+                agent=True,
+            )
+        )
+    )
 
-    url = os.getenv('LIVEKIT_URL')
-    api_key = os.getenv('LIVEKIT_API_KEY')
-    api_secret = os.getenv('LIVEKIT_API_SECRET')
-
-
-    token1 = api.AccessToken(api_key=api_key, api_secret=api_secret) \
-        .with_identity('core:user.test.agent-send') \
-        .with_name("Agent") \
-        .with_kind("agent") \
-        .with_grants(api.VideoGrants(
-            can_update_own_metadata=True,
-            room_join=True,
-            room="test-process",
-            agent=True
-        ))
-    
     jwt1 = token1.to_jwt()
 
-
-    token2 = api.AccessToken(api_key=api_key, api_secret=api_secret) \
-        .with_identity('core:user.test.agent-recv') \
-        .with_name("Agent") \
-        .with_kind("agent") \
-        .with_grants(api.VideoGrants(
-            can_update_own_metadata=True,
-            room_join=True,
-            room="test-process",
-            agent=True
-        ))
-    
-    jwt2 = token2.to_jwt()       
-
-    room1 = rtc.Room()  
+    token2 = (
+        api.AccessToken(api_key=api_key, api_secret=api_secret)
+        .with_identity("core:user.test.agent-recv")
+        .with_name("Agent")
+        .with_kind("agent")
+        .with_grants(
+            api.VideoGrants(
+                can_update_own_metadata=True,
+                room_join=True,
+                room="test-process",
+                agent=True,
+            )
+        )
+    )
+
+    jwt2 = token2.to_jwt()
+
+    room1 = rtc.Room()
     await room1.connect(url=url, token=jwt1)
-    
-    room2 = rtc.Room()  
+
+    room2 = rtc.Room()
     await room2.connect(url=url, token=jwt2)
-                  
+
     topic = "test_topic"
 
     while True:
         await asyncio.sleep(0.1)
 
-        if room2.local_participant.identity in room1.remote_participants and room1.local_participant.identity in room2.remote_participants:
+        if (
+            room2.local_participant.identity in room1.remote_participants
+            and room1.local_participant.identity in room2.remote_participants
+        ):
             break
 
-    async with livekit_protocol.LivekitProtocol(room = room1, remote = room1.remote_participants[room2.local_participant.identity], topic=topic) as proto1:
-        async  with livekit_protocol.LivekitProtocol(room = room2, remote = room2.remote_participants[room1.local_participant.identity], topic=topic) as proto2:
-
+    async with livekit_protocol.LivekitProtocol(
+        room=room1,
+        remote=room1.remote_participants[room2.local_participant.identity],
+        topic=topic,
+    ) as proto1:
+        async with livekit_protocol.LivekitProtocol(
+            room=room2,
+            remote=room2.remote_participants[room1.local_participant.identity],
+            topic=topic,
+        ) as proto2:
             test_data_builder = bytearray()
-            for i in range(1024*1024):
+            for i in range(1024 * 1024):
                 test_data_builder.append(i % 255)
 
             test_data = bytes(test_data_builder)
@@ -76,16 +92,17 @@ async def test_protocol():
             done = asyncio.Future[bool]()
 
             matches = 0
-            async def test_fn(protocol, id: int, type:str, data:bytes):
+
+            async def test_fn(protocol, id: int, type: str, data: bytes):
                 nonlocal matches
                 logger.info("Message received")
                 if test_data != data:
                     raise "data isn't equal"
-                matches+=1
+                matches += 1
 
                 if matches == 2:
                     done.set_result(True)
-            
+
             proto2.register_handler("test", test_fn)
 
             await asyncio.sleep(1)
@@ -95,12 +112,5 @@ async def test_protocol():
 
             await done
 
-
     await room2.disconnect()
     await room1.disconnect()
-
-
-    
-
-
-    

meshagent/livekit/version.py

@@ -1 +1 @@
-__version__ = "0.0.37"
+__version__ = "0.0.38"

meshagent_livekit-0.0.38.dist-info/METADATA

@@ -0,0 +1,37 @@
+Metadata-Version: 2.4
+Name: meshagent-livekit
+Version: 0.0.38
+Summary: Livekit support for Meshagent
+License-Expression: Apache-2.0
+Project-URL: Documentation, https://docs.meshagent.com
+Project-URL: Website, https://www.meshagent.com
+Project-URL: Source, https://www.meshagent.com
+Requires-Python: >=3.12
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: pytest~=8.4
+Requires-Dist: pytest-asyncio~=0.26
+Requires-Dist: strip-markdown~=1.3
+Requires-Dist: livekit-api>=1.0
+Requires-Dist: livekit-agents~=1.1
+Requires-Dist: livekit-plugins-openai~=1.1
+Requires-Dist: livekit-plugins-silero~=1.1
+Requires-Dist: livekit-plugins-turn-detector~=1.1
+Requires-Dist: meshagent-api~=0.0.38
+Requires-Dist: meshagent-tools~=0.0.38
+Dynamic: license-file
+
+## MeshAgent LiveKit
+The ``meshagent.livekit`` package equips agents with real-time audio and voice capabilities via the LiveKit SDK. 
+
+### VoiceBot
+The ``VoiceBot`` agent handles two-way voice conversations allowing users to interact with the agent verbally. Agents based on the ``VoiceBot`` class can be given the same tools as ``ChatBot`` based agents. This means you only need to write a tool once and the same tool can be used across both text and voice based agents. Check out the [Build and Deploy a Voice Agent](https://docs.meshagent.com/agents/standard/buildanddeployvoicebot) example to learn how to create a simple Voice Agent without tools then add built-in MeshAgent tools and custom tools to the agent.
+
+---
+### Learn more about MeshAgent on our website or check out the docs for additional examples!
+
+**Website**: [www.meshagent.com](https://www.meshagent.com/)
+
+**Documentation**: [docs.meshagent.com](https://docs.meshagent.com/)
+
+---

meshagent_livekit-0.0.38.dist-info/RECORD

@@ -0,0 +1,11 @@
+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=R5QxTjVaID7odO0eBWpOnyCjNQxBZ7cpyruM_NMOoDc,23
+meshagent/livekit/agents/transcriber.py,sha256=oqfHBhBSwU62LbsO8WFiJg3Xoi4vkWlTFzgTxBP0erg,13297
+meshagent/livekit/agents/voice.py,sha256=STgjMSqzUgV9UAmleOy1vkgRXP93MDSYgiOO6Lo0peU,11964
+meshagent_livekit-0.0.38.dist-info/licenses/LICENSE,sha256=eTt0SPW-sVNdkZe9PS_S8WfCIyLjRXRl7sUBWdlteFg,10254
+meshagent_livekit-0.0.38.dist-info/METADATA,sha256=pVJlT13Hr3NnFwMSuA60PG_U_uZoPgWcwlepxZl9k_w,1721
+meshagent_livekit-0.0.38.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+meshagent_livekit-0.0.38.dist-info/top_level.txt,sha256=GlcXnHtRP6m7zlG3Df04M35OsHtNXy_DY09oFwWrH74,10
+meshagent_livekit-0.0.38.dist-info/RECORD,,

meshagent_livekit-0.0.37.dist-info/METADATA

@@ -1,24 +0,0 @@
-Metadata-Version: 2.4
-Name: meshagent-livekit
-Version: 0.0.37
-Summary: Livekit support for Meshagent
-License-Expression: Apache-2.0
-Project-URL: Documentation, https://docs.meshagent.com
-Project-URL: Website, https://www.meshagent.com
-Project-URL: Source, https://www.meshagent.com
-Requires-Python: >=3.12
-Description-Content-Type: text/markdown
-License-File: LICENSE
-Requires-Dist: pytest~=8.3
-Requires-Dist: pytest-asyncio~=0.26
-Requires-Dist: strip-markdown~=1.3
-Requires-Dist: livekit-api>=1.0
-Requires-Dist: livekit-agents~=1.1
-Requires-Dist: livekit-plugins-openai~=1.1
-Requires-Dist: livekit-plugins-silero~=1.1
-Requires-Dist: livekit-plugins-turn-detector~=1.1
-Requires-Dist: meshagent-api~=0.0.37
-Requires-Dist: meshagent-tools~=0.0.37
-Dynamic: license-file
-
-### Meshagent LiveKit

meshagent_livekit-0.0.37.dist-info/RECORD

@@ -1,11 +0,0 @@
-meshagent/livekit/__init__.py,sha256=8zLGg-DfQhnDl2Ky0n-zXpN-8e-g7iR0AcaI4l4Vvpk,32
-meshagent/livekit/livekit_protocol.py,sha256=K9yP-qpxag5_7TXlKjFEx3cOJJJpYI_z6zGzFHoN1Hs,1421
-meshagent/livekit/livekit_protocol_test.py,sha256=n_ZQjt7n4u7TM7eENzH8L0tw8LvypS_JHF_PuJ2o6h4,2836
-meshagent/livekit/version.py,sha256=JaGEpJ5xP3R4j7pGgCziGajlIRjy1_NJdv_OaXPQius,22
-meshagent/livekit/agents/transcriber.py,sha256=Dq1Ijx4gmA-0jQGM-f3w7X-JIZpkRCFDxWae9AOwz-k,12290
-meshagent/livekit/agents/voice.py,sha256=CHbzuLdkxEcuHfC1skjl_0KHV-hmf8i6NWxV7AYTWyc,12348
-meshagent_livekit-0.0.37.dist-info/licenses/LICENSE,sha256=eTt0SPW-sVNdkZe9PS_S8WfCIyLjRXRl7sUBWdlteFg,10254
-meshagent_livekit-0.0.37.dist-info/METADATA,sha256=tdVvbqxsQp0faK5xMyHqFUdLUd2nXywubcydiE2B9kM,790
-meshagent_livekit-0.0.37.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-meshagent_livekit-0.0.37.dist-info/top_level.txt,sha256=GlcXnHtRP6m7zlG3Df04M35OsHtNXy_DY09oFwWrH74,10
-meshagent_livekit-0.0.37.dist-info/RECORD,,