I am using tts from OpenAI "gpt-4o-mini-tts" TTSMetrics streamed is No, how to make it stream

i have deployed livekit frontend using https://github.com/livekit-examples/agent-deployment.git. and stt-llm-tts code has been deployed on ec2 server. but both things are not getting connected

This message was deleted.

Where do I get the SYS_ADMIN from?

docker run --rm \
  --cap-add SYS_ADMIN \
  -e EGRESS_CONFIG_FILE=/out/config.yaml \
  -v ~/livekit-egress:/out \
  livekit/egress

Traceback (most recent call last): File "/Users/sandeep/projects/agntv1/lib/python3.10/site-packages/livekit/agents/stt/stt.py", line 246, in _main_task return await self._run() File "/Users/sandeep/projects/agntv1/lib/python3.10/site-packages/livekit/plugins/deepgram/stt.py", line 562, in _run ws = await self._connect_ws() File "/Users/sandeep/projects/agntv1/lib/python3.10/site-packages/livekit/plugins/deepgram/stt.py", line 622, in _connect_ws ws = await asyncio.wait_for( File "/opt/homebrew/Cellar/python@3.10/3.10.17_1/Frameworks/Python.framework/Versions/3.10/lib/python3.10/asyncio/tasks.py", line 445, in wait_for return fut.result() File "/Users/sandeep/projects/agntv1/lib/python3.10/site-packages/aiohttp/client.py", line 1409, in send return self._coro.send(arg) File "/Users/sandeep/projects/agntv1/lib/python3.10/site-packages/aiohttp/client.py", line 1021, in _ws_connect raise WSServerHandshakeError( aiohttp.client_exceptions.WSServerHandshakeError: 400, message='Invalid response status', url='wss://api.deepgram.com/v1/listen?model=nova-3&amp;punctuate=true&amp;smart_format=true&amp;no_delay=true&amp;interim_results=true&amp;encoding=linear16&amp;vad_events=true&amp;sample_rate=16000&amp;channels=1&amp;endpointing=25&amp;filler_words=true&amp;profanity_filter=false&amp;numerals=false&amp;mip_opt_out=false&amp;language=hi' {"pid": 51196, "job_id": "AJ_DPoiADppLTSW"} close event: type='close' error=STTError(type='stt_error', timestamp=1748600166.066468, label='livekit.plugins.deepgram.stt.STT', error=WSServerHandshakeError(RequestInfo(url=URL('wss://api.deepgram.com/v1/listen?model=nova-3&amp;punctuate=true&amp;smart_format=true&amp;no_delay=true&amp;interim_results=true&amp;encoding=linear16&amp;vad_events=true&amp;sample_rate=16000&amp;channels=1&amp;endpointing=25&amp;filler_words=true&amp;profanity_filter=false&amp;numerals=false&amp;mip_opt_out=false&amp;language=hi'), method='GET', headers=<CIMultiDictProxy('Host': 'api.deepgram.com', 'Authorization': 'Token cf89eed70e555301638f2f263d1c78a6518099d6', 'Upgrade': 'websocket', 'Connection': 'Upgrade', 'Sec-WebSocket-Version': '13', 'Sec-WebSocket-Key': 'HSgA5yKBhcQpD8BV7tTqbg==', 'Accept': '*/*', 'Accept-Encoding': 'gzip, deflate', 'User-Agent': 'Python/3.10 aiohttp/3.11.18')>, real_url=URL('wss://api.deepgram.com/v1/listen?model=nova-3&amp;punctuate=true&amp;smart_format=true&amp;no_delay=true&amp;interim_results=true&amp;encoding=linear16&amp;vad_events=true&amp;sample_rate=16000&amp;channels=1&amp;endpointing=25&amp;filler_words=true&amp;profanity_filter=false&amp;numerals=false&amp;mip_opt_out=false&amp;language=hi')), (), status=400, message='Invalid response status', headers=<CIMultiDictProxy('Content-Type': 'application/json', 'dg-error': 'Bad Request', 'Vary': 'origin, access-control-request-method, access-control-request-headers', 'Vary': 'accept-encoding', 'Access-Control-Allow-Credentials': 'true', 'Access-Control-Expose-Headers': 'dg-model-name,dg-model-uuid,dg-char-count,dg-request-id,dg-error', 'Content-Encoding': 'gzip', 'dg-request-id': '9817948c-84cd-43c7-a6fe-41022d6bee6e', 'Transfer-Encoding': 'chunked', 'Date': 'Fri, 30 May 2025 101605 GMT')>), recoverable=False)

I have launched sandbox livekit application and it's running fine. How do I integrate this setup on my server?

When communicating using SIP protocol via TCP, my telephony provider is asking for media server IP to whitelist. Where can I find the media server IP?

This message was deleted.

in python3 sdk room token i want to reduce the time when room considered disconnected when the all participant. left

Explain the VAD config options to leyman

i mean RoomConfiguration(empty_timeout) (edited)

i mean RoomConfiguration(empty_timeout) (edited) in python3 sdk

can someone suggest me the best tts and sst for product?

I'm using json file where I need to receive all the SIP headers coming in my inbound request... how can I read 'em in my AI agent where I 'm getting SIP participant ??

Subject: Persistent "failed to retrieve region info" error with Python SDK Hello LiveKit peeps, I am encountering a persistent connection error when using the Python SDK and I'm hoping you can help, as I have exhausted all troubleshooting steps. The Problem: When I run a Python script to connect to my LiveKit Cloud project, it consistently fails with the error:

livekit.rtc.room.ConnectError: engine: signal failure: failed to retrieve region info: error decoding response body: expected value at line 1 column 1

. Project URL:

<wss://twindersagarstrip-xyhjwf0v.livekit.cloud>

Here is what I've discovered: * Browser Connection Works: I can connect to my project room (

avatar_room_a

) successfully from the same machine using the web app at

<http://meet.livekit.io|meet.livekit.io>

. This proves my network path and credentials are fundamentally correct. * CLI Token Generation Works: I can use the

lk

CLI tool to generate valid tokens for my project. * Python Connection Fails Consistently: The Python script fails with the

region info

error under all of the following conditions: * Running on my local Windows machine. * Running inside a brand new, clean Gitpod cloud container (Linux). * Using a token generated by the Python

AccessToken

class (this sometimes gives a

401 Unauthorized

instead, suggesting the SDK's token signing is also incompatible with my setup). * Using a known-good token generated by the

lk

CLI. * Connecting to the main project URL. * Connecting directly to the regional URL (

<wss://eu-central-1.livekit.cloud>

). * Things I've Ruled Out: * Local firewalls (the error persists even with Windows Firewall completely disabled). * Local proxies (no proxy environment variables are set). * Corrupted Python environment (the error persists in a fresh

venv

inside Gitpod). Since the browser (using the JS SDK) can connect but the Python SDK cannot, even from a pristine cloud environment, this strongly suggests a specific bug or incompatibility between the Python SDK's connection process and the cloud backend for my project. Here is the traceback from the last attempt in Gitpod, using a valid token generated by the CLI:

((venv) ) gitpod /workspace/sagar (main) $ python avatar_a_controller.py
--- Full Avatar Control Script (using CLI token) ---
Attempting to connect to <wss://twindersagarstrip-xyhjwf0v.livekit.cloud>...
livekit::rtc_engine:392:livekit::rtc_engine - failed to connect: Signal(RegionError("error decoding response body: expected value at line 1 column 1")), retrying... (1/3)
An error occurred: engine: signal failure: failed to retrieve region info: error decoding response body: expected value at line 1 column 1
Traceback (most recent call last):
  File "/workspace/sagar/avatar_a_controller.py", line 39, in main
    await room.connect(LIVEKIT_URL, token)
  File "/workspace/sagar/venv/lib/python3.12/site-packages/livekit/rtc/room.py", line 394, in connect
    raise ConnectError(cb.connect.error)
livekit.rtc.room.ConnectError: engine: signal failure: failed to retrieve region info: error decoding response body: expected value at line 1 column 1

Could you please provide any insight into what might be causing this? Thank you for your help. 🙏🙏🙏

This message was deleted.

Can i change TTS for Gemini live api?

I am using unable to hear what the remote participant is saying even after being subscribed to remote audio track. I am using livekit_client flutter package only. import 'dart:async'; import 'dart:convert'; import 'dart:io'; import 'package:flutter/material.dart'; import 'package:fluttertoast/fluttertoast.dart'; import 'package:livekit_client/livekit_client.dart' as livekit; import 'package:permission_handler/permission_handler.dart'; import 'package:uuid/uuid.dart'; import '../../models/avatar_model.dart'; import '../../services/cheercast_api_service.dart'; import '../../theme/theme.dart'; class VideoCallScreen extends StatefulWidget { final AvatarModel avatar; final String mode; final String? outfit; final String? topic; final String? subtopic; final String? level; const VideoCallScreen({ super.key, required this.avatar, required this.mode, this.outfit, this.subtopic, this.topic, this.level, }); @override State<VideoCallScreen> createState() => _VideoCallScreenState(); } class _VideoCallScreenState extends State<VideoCallScreen> { final CheercastApiService _apiService = CheercastApiService(); livekit.Room? _room; livekit.RemoteVideoTrack? _remoteVideoTrack; livekit.RemoteAudioTrack? _remoteAudioTrack; livekit.LocalVideoTrack? _localVideoTrack; bool _isLoading = true; String? _sessionId; Timer? _timer; int _callDuration = 0; dynamic _socket; final bool _canSubscribe = false; String? agentSid; @override void initState() { super.initState(); _startVideoCall(); } void _startTimer() { timer = Timer.periodic(const Duration(seconds: 1), () { setState(() { _callDuration++; }); }); } String _formatDuration(int seconds) { final minutes = (seconds / 60).floor().toString().padLeft(2, '0'); final secs = (seconds % 60).toString().padLeft(2, '0'); return '$minutes:$secs'; } Future<void> _requestPermission() async { var status = await Permission.bluetooth.request(); if (status.isPermanentlyDenied) { print('Bluetooth Permission disabled'); } status = await Permission.bluetoothConnect.request(); if (status.isPermanentlyDenied) { print('Bluetooth Connect Permission disabled'); } status = await Permission.microphone.request(); if (status.isPermanentlyDenied) { print('Microphone Permission disabled'); } status = await Permission.camera.request(); if (status.isPermanentlyDenied) { print('Camera Permission disabled'); } return; } Future<void> _startVideoCall() async { try { final sessionId = const Uuid().v4(); /// TODO: Replace this with real backend userId when available const int userId = 21511; print("Getting to response"); final group = await _apiService.getChatGroup( userId: userId, avatarCode: widget.avatar.avatarCode, ); final channel = group['channel']; final socket = await _apiService.startSocket( chatGroupChannel: channel, sessionId: sessionId, userId: userId, ); await _requestPermission(); livekit.LiveKitClient.initialize(); socket.stream.listen((data) async { print("Received data: $data"); print("Received data type: ${data.runtimeType}"); final received = jsonDecode(data); if (received['code'] != 200) { Fluttertoast.showToast( msg: received['message'] ?? 'An error occurred', toastLength: Toast.LENGTH_SHORT, gravity: ToastGravity.BOTTOM, ); Navigator.pop(context); } else { if (received['data']['event'] == 'on_start_streaming') { final wssUrl = received['data']['metadata']['wss']; final token = received['data']['metadata']['token']; final room = livekit.Room( roomOptions: const livekit.RoomOptions(adaptiveStream: false), ); room.events.listen((event) async { print("This is the event in room: $event"); if (event is livekit.TrackSubscribedEvent) { if (event.track is livekit.RemoteVideoTrack) { setState(() { _remoteVideoTrack = event.track as livekit.RemoteVideoTrack; }); print("Video track is muted: ${_remoteVideoTrack?.muted}"); } else if (event.track is livekit.RemoteAudioTrack) { setState(() { _remoteAudioTrack = event.track as livekit.RemoteAudioTrack; }); _remoteAudioTrack?.enable(); _remoteAudioTrack?.mediaStreamTrack.enabled = true; _remoteAudioTrack?.start(); await room.setSpeakerOn(true, forceSpeakerOutput: true); await setSpeakerphoneOn(true); print( "Audio track found : ${_remoteAudioTrack?.isActive} && ${_remoteAudioTrack?.muted} ", ); } } }); await room.prepareConnection(wssUrl, token); try { // _localVideoTrack = // await livekit.LocalVideoTrack.createCameraTrack(); // final localAudioTrack = await livekit.LocalAudioTrack.create(); // await room.localParticipant?.publishVideoTrack(_localVideoTrack!); // await room.localParticipant?.publishAudioTrack(localAudioTrack); await room.localParticipant?.setCameraEnabled(true); await room.localParticipant?.setMicrophoneEnabled(true); await room.connect( wssUrl, token, fastConnectOptions: livekit.FastConnectOptions( camera: livekit.TrackOption( enabled: true, // track: _localVideoTrack, ), microphone: livekit.TrackOption( enabled: true, // track: localAudioTrack, ), ), ); } catch (error) { print('Could not publish video/audio, error: $error'); } final audioOuts = await livekit.Hardware.instance.audioOutputs(); final audioIns = await livekit.Hardware.instance.audioInputs(); print("Audio track outs: $audioOuts"); print( "Audio track selected out: ${room.selectedAudioOutputDeviceId} && ${_remoteAudioTrack?.isActive} && ${_remoteAudioTrack?.prevStats} && ${await _remoteAudioTrack?.getReceiverStats()}", ); print("Audio track ins: $audioIns"); setState(() { _room = room; _isLoading = false; _sessionId = sessionId; }); } } }); await Future.delayed(const Duration(milliseconds: 350), () { final payload = { 'event': 'start_video_call', 'id': sessionId, 'data': { 'mode': widget.mode, if (widget.outfit != null) 'outfit': widget.outfit, if (widget.topic != null) 'topic': widget.topic, if (widget.subtopic != null) 'subtopic': widget.subtopic, if (widget.level != null) 'level': widget.level, }, }; socket.sink.add(jsonEncode(payload)); print("Sent video call event"); _startTimer(); }); setState(() { _socket = socket; }); } catch (e, stacktrace) { debugPrint('Video Call Error: $e'); debugPrintStack(stackTrace: stacktrace, label: "Video Call Stacktrace"); if (mounted) { ScaffoldMessenger.of(context).showSnackBar( SnackBar(content: Text('Failed to start video call: $e')), ); Navigator.pop(context); } } } setSpeakerphoneOn(bool speakerOn) async { if (Platform.isIOS) { if (livekit.Hardware.instance.speakerOn == false && livekit.Hardware.instance.preferSpeakerOutput) { await livekit.Hardware.instance.setSpeakerphoneOn(false); } } await livekit.Hardware.instance.setSpeakerphoneOn(speakerOn); print("Speaker set to $speakerOn"); } Future<void> _endCall() async { try { _timer?.cancel(); _socket?.sink.add( jsonEncode({'event': 'stop_video_call', 'id': _sessionId}), ); _room?.localParticipant?.setCameraEnabled(false); _room?.localParticipant?.setMicrophoneEnabled(false); await _room?.disconnect(); _socket?.sink.close(); _socket = null; } catch (e) { debugPrint('Error ending video call: $e'); } finally { if (mounted) Navigator.pop(context); } } @override void dispose() { _timer?.cancel(); _socket?.sink.add( jsonEncode({'event': 'stop_video_call', 'id': _sessionId}), ); _room?.localParticipant?.setCameraEnabled(false); _room?.localParticipant?.setMicrophoneEnabled(false); _room?.disconnect(); _socket?.sink.close(); _socket = null; super.dispose(); } Widget _buildVideo() { return Stack( children: [ Positioned.fill( child: _remoteVideoTrack != null ? livekit.VideoTrackRenderer(_remoteVideoTrack!) : Center( child: Text( 'Waiting for ${widget.avatar.name} to join...', style: const TextStyle( color: Colors.white70, fontSize: 18, ), ), ), ), // if (_localVideoTrack != null) // Positioned( // right: 10, // top: 10, // width: 120, // height: 160, // child: livekit.VideoTrackRenderer(_localVideoTrack!), // ), Positioned( top: 16, left: 16, child: Container( padding: const EdgeInsets.symmetric(horizontal: 12, vertical: 6), decoration: BoxDecoration( color: Colors.black45, borderRadius: BorderRadius.circular(8), ), child: Text( _formatDuration(_callDuration), style: const TextStyle(color: Colors.white, fontSize: 16), ), ), ), ], ); } @override Widget build(BuildContext context) { return Scaffold( backgroundColor: Colors.black, appBar: AppBar( title: Text('Video Call - ${widget.avatar.name}'), backgroundColor: AppTheme.primaryColor, actions: [ IconButton(icon: const Icon(Icons.call_end), onPressed: _endCall), ], ), body: _isLoading ? const Center(child: CircularProgressIndicator()) : _buildVideo(), ); } }

This message was deleted.

This message was deleted.

can i use same number for inbound and outbound

This message was deleted.

does DeleteRoom emit an event ?

If I pay for LiveKit enterprise, will I be able to deploy the mesh network on my own infra?

This message was deleted.

@icy-policeman-64898 can I give metadata to agent when connect using livekit-frontend example

This message was deleted.

can i connect to twilio only using account sid and auth token not not username and password?

do i need to create a livekit sip trunk per phone number, ? one trunk for inbound and one trunk for outbound per phone number? one inbound trunk per phone number and only one outbound trunk for all phone numbers?

Is it possible to inject context from outside into a room during a session, through API or directly to the agent? In python