cat agent.py
/home/ubuntu/landing-denis-call/agent...
# ask-aim
melodic-vegetable-78787
02/27/2025, 2:15 PMcat agent.py
# /home/ubuntu/landing-denis-call/agent.py
from future import annotations
import asyncio
import logging
import os
import redis
from dotenv import load_dotenv
from livekit import rtc, api
from livekit.agents import (
AutoSubscribe,
JobContext,
JobProcess,
WorkerOptions,
cli,
llm,
)
from livekit.agents.pipeline import VoicePipelineAgent
from livekit.plugins import openai, silero, deepgram, elevenlabs
from livekit import protocol
import json
load_dotenv(dotenv_path=".env.local")
logger = logging.getLogger("VOICE-ASSISTANT")
logger.setLevel(logging.DEBUG)
outbound_trunk_id = os.getenv("SIP_OUTBOUND_TRUNK_ID")
redis_client = redis.Redis(
host='redis',
port=6379,
db=0,
username='default',
password='Y3v!n9L#f2@AFp$G1w^K8z*R'
)
_instructions = (
"You are a friendly QFix Connect representative who speaks only in Turkish. "
"Introduce yourself as Denis from QFix Connect in Turkish only, never mentioning AI. "
"Maintain a warm, humorous, and casual approach."
)
async def record_call(ctx: JobContext, phone_number: str):
room_name = ctx.room.name
cleaned_phone_number = phone_number.lstrip('+') # + işaretini kaldır
filepath = f"outbound-{cleaned_phone_number}-{room_name}.m4a"
logger.info(f"Starting recording for room: {room_name}, filepath: {filepath}")
s3_upload = protocol.egress.S3Upload(
access_key="AKIAX3B7KVP7FASNEQOB",
secret="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",
region="eu-central-1",
bucket="asterisk-call-recordings",
force_path_style=True # Endpoint kaldırıldı, eski haline döndü
)
file_output = protocol.egress.EncodedFileOutput(
filepath=filepath,
s3=s3_upload
)
req = protocol.egress.RoomCompositeEgressRequest(
room_name=room_name,
audio_only=True,
file_outputs=[file_output],
)
try:
logger.debug(f"Sending start_room_composite_egress request: {req}")
res = await ctx.api.egress.start_room_composite_egress(req)
logger.info(f"Recording started with egress_id: {res.egress_id}")
return res.egress_id
except Exception as e:
logger.error(f"Failed to start recording: {str(e)}")
raise
async def stop_recording(ctx: JobContext, egress_id: str, caller_number: str, call_to: str):
if not egress_id:
logger.warning("No recording to stop.")
return
try:
logger.debug(f"Stopping recording with egress_id: {egress_id}")
await ctx.api.egress.stop_egress(api.StopEgressRequest(egress_id=egress_id))
logger.info(f"Recording stopped for egress_id: {egress_id}")
egress_info = {
"egress_id": egress_id,
"caller_number": caller_number,
"call_to": call_to
}
redis_client.lpush("egress_ids", json.dumps(egress_info))
logger.info(f"Egress info {egress_info} Redis’e kaydedildi.")
except Exception as e:
logger.error(f"Failed to stop recording: {str(e)}")
raise
async def entrypoint(ctx: JobContext):
logger.info(f"[OUTBOUND] connecting to room {ctx.room.name}")
phone_number = ctx.job.metadata
if not phone_number:
logger.error("[OUTBOUND] No phone number provided in job metadata. Exiting.")
ctx.shutdown()
return
if not outbound_trunk_id or not outbound_trunk_id.startswith("ST_"):
logger.error("[OUTBOUND] SIP_OUTBOUND_TRUNK_ID is missing or invalid.")
ctx.shutdown()
return
await ctx.connect(auto_subscribe=AutoSubscribe.AUDIO_ONLY)
user_identity = "phone_user"
logger.info(f"[OUTBOUND] Dialing {phone_number} with trunk={outbound_trunk_id}, room={ctx.room.name}")
await ctx.api.sip.create_sip_participant(
api.CreateSIPParticipantRequest(
room_name=ctx.room.name,
sip_trunk_id=outbound_trunk_id,
sip_call_to=phone_number,
participant_identity=user_identity,
play_dialtone=True
)
)
logger.info(f"[OUTBOUND] Waiting for SIP participant {user_identity}")
participant = await ctx.wait_for_participant(identity=user_identity)
logger.info(f"[OUTBOUND] SIP participant {user_identity} joined")
egress_id = None
agent_started = [False]
caller_number = participant.attributes.get("sip.trunkPhoneNumber", "unknown")
call_to = phone_number
def on_track_subscribed(track, publication, participant):
nonlocal agent_started, egress_id
logger.debug(f"track_subscribed triggered for participant: {participant.identity}, track kind: {track.kind}")
if participant.identity == user_identity and track.kind == rtc.TrackKind.KIND_AUDIO:
logger.info("SIP audio track subscribed")
if not agent_started[0]:
logger.info("Starting voice pipeline agent from track subscription")
run_voice_pipeline_agent(ctx, participant, _instructions)
agent_started[0] = True
if not egress_id:
logger.debug("Starting recording task from track subscription")
task = asyncio.create_task(record_call(ctx, phone_number))
task.add_done_callback(lambda fut: globals().update(egress_id=fut.result()))
def on_participant_disconnected(participant: rtc.RemoteParticipant):
nonlocal egress_id
if participant.identity == user_identity:
logger.info("SIP participant disconnected; stopping recording")
if egress_id:
asyncio.create_task(stop_recording(ctx, egress_id, caller_number, call_to))
ctx.room.on("track_subscribed", on_track_subscribed)
ctx.room.on("participant_disconnected", on_participant_disconnected)
await asyncio.sleep(2)
if not agent_started[0]:
logger.info("Track subscription not triggered, forcing voice pipeline agent start")
run_voice_pipeline_agent(ctx, participant, _instructions)
agent_started[0] = True
if not egress_id:
logger.debug("Track subscription not triggered, forcing recording start")
egress_id = await record_call(ctx, phone_number)
try:
while True:
await asyncio.sleep(1)
except asyncio.CancelledError:
logger.info("Agent job was cancelled.")
if egress_id:
await stop_recording(ctx, egress_id, caller_number, call_to)
await asyncio.sleep(5)
def run_voice_pipeline_agent(ctx: JobContext, participant: rtc.RemoteParticipant, instructions: str):
logger.info("[OUTBOUND] Starting voice pipeline agent")
initial_ctx = llm.ChatContext().append(role="system", text=instructions)
fn_ctx = OutboundActions(ctx, participant)
agent = VoicePipelineAgent(
vad=ctx.proc.userdata["vad"],
stt=deepgram.STT(model="nova-2-general", language="tr"),
llm=openai.LLM(model="gpt-4o", temperature=0.2),
tts=elevenlabs.TTS(
model="eleven_turbo_v2_5",
voice=elevenlabs.Voice(
id="dOQlT8TikdF47jfPQXpe",
name="Beyda",
category="premade",
settings=elevenlabs.VoiceSettings(
stability=0.71,
similarity_boost=0.5,
style=0.0,
use_speaker_boost=True
)
),
language="tr",
),
chat_ctx=initial_ctx,
fnc_ctx=fn_ctx,
)
agent.start(ctx.room, participant)
logger.info("Agent started, sending initial message")
asyncio.create_task(agent.say("Merhaba! Ben QFix Connect’ten Denis. Nasıl yardımcı olabilirim?", allow_interruptions=True))
class OutboundActions(llm.FunctionContext):
def __init__(self, ctx: JobContext, participant: rtc.RemoteParticipant):
super().__init__()
self.ctx = ctx
self.api = ctx.api
self.participant = participant
self.room = ctx.room
async def end_call(self):
logger.info(f"[OUTBOUND] user ended call => {self.participant.identity}")
await self.api.room.remove_participant(api.RoomParticipantIdentity(room=self.room.name, identity=self.participant.identity))
self.ctx.shutdown()
def prewarm(proc: JobProcess):
proc.userdata["vad"] = silero.VAD.load()
if name == "__main__":
cli.run_app(WorkerOptions(entrypoint_fnc=entrypoint, prewarm_fnc=prewarm, agent_name="landing-denis-call"))
what is best room parameters for outbound calls for my agent ? when ı record routbound call room with egress, agent sound super clear but telephone users sound is so cripsy and bad .
im usıng python agent framework . do the best configiratıon for outbıound calls .
any parameters or code snippet s ?
@millions-hair-24956
g
gray-monkey-83525
02/27/2025, 2:15 PMOutbound çağrılar için en iyi oda parametrelerini ayarlamak, ses kalitesini artırabilir. LiveKit'te, ses kalitesini etkileyen birkaç faktör vardır. Özellikle, ses codec'leri ve örnekleme oranları önemlidir. LiveKit SIP, HD Ses'i destekler ve bu, ses kalitesini artırabilir. G.722 veya Opus gibi geniş bant ses codec'lerini kullanmak, ses kalitesini artırabilir. Ayrıca, HD Ses yapılandırması ile ilgili belgeleri inceleyebilirsiniz.
Ayrıca, sesin netliğini artırmak için ses işleme ayarlarını optimize edebilirsiniz. Örneğin, `silero.VAD`'in
activation_thresholdsttttsdeepgram.STTelevenlabs.TTSm
melodic-vegetable-78787
02/27/2025, 2:16 PMevet @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 2:16 PMAnladım, daha fazla ayrıntı sağlayayım.
1. Ses Codec'leri ve Örnekleme Oranları: LiveKit, HD Ses'i destekler ve bu, ses kalitesini artırabilir. G.722 veya Opus gibi geniş bant ses codec'lerini kullanmak, ses kalitesini artırabilir. Bu codec'ler, daha yüksek örnekleme oranları sunarak daha geniş bir frekans aralığını kapsar ve bu da daha net ses sağlar. HD Ses yapılandırması ile ilgili belgeleri inceleyebilirsiniz.
2. Voice Activity Detection (VAD) Ayarları:
silero.VADactivation_thresholddeepgram.STTelevenlabs.TTSdeepgram.STTmodel="nova-2-general"language="tr" Copy code
room_options = RoomOptions(
defaultAudioCaptureOptions=AudioCaptureOptions(
echoCancellation=False,
noiseSuppression=False
),
defaultAudioPublishOptions=AudioPublishOptions(
encoding=AudioEncoding(maxBitrate=320000), # Yüksek kaliteli ses için
dtx=False
)
)m
melodic-vegetable-78787
02/27/2025, 2:22 PMkoduma nasıl entegreederım onerılerını daha fazla bılgı ver , userın sesı kotu sadece agentınkı mukemmel kayıtlarda @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 2:22 PMKullanıcı ses kalitesini artırmak için aşağıdaki adımları izleyebilirsiniz:
1. Ses Codec'lerini ve Örnekleme Oranlarını Ayarlayın: Kullanıcı ses kalitesini artırmak için geniş bant ses codec'leri kullanabilirsiniz. G.722 veya Opus gibi codec'ler, daha yüksek örnekleme oranları sunarak daha geniş bir frekans aralığını kapsar. Bu codec'leri kullanmak için SIP trunk sağlayıcınızın bu codec'leri desteklediğinden emin olun.
2. Ses İşleme Ayarlarını Optimize Edin: `RoomOptions`'ı kullanarak ses işleme özelliklerini ayarlayabilirsiniz. Yankı iptali ve gürültü bastırma gibi özellikleri devre dışı bırakmak, ses kalitesini artırabilir. Örneğin:
Copy code
room_options = RoomOptions(
defaultAudioCaptureOptions=AudioCaptureOptions(
echoCancellation=False,
noiseSuppression=False
),
defaultAudioPublishOptions=AudioPublishOptions(
encoding=AudioEncoding(maxBitrate=320000), # Yüksek kaliteli ses için
dtx=False
)
)silero.VADactivation_thresholdaudio_only=TrueRoomOptionsAudioCaptureOptionsm
melodic-vegetable-78787
02/27/2025, 2:35 PMcevredekı konsuanları engellemek ıcın sıvero vad degerı ne olmalı ? kalabalıkta sadece konusanı algılasın ıstıyorum . @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 2:35 PMSilero VADactivation_thresholdactivation_thresholdactivation_threshold Copy code
vad = silero.VAD.load(activation_threshold=0.7)m
melodic-vegetable-78787
02/27/2025, 2:37 PMkalabalık bır oramda sesı kesıyor bunu onlemek ıstıyoruz @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 2:37 PMKalabalık bir ortamda ses kesilmesini önlemek için
Silero VADactivation_thresholdactivation_threshold Copy code
vad = silero.VAD.load(activation_threshold=0.5)min_silence_durationmin_speech_durationm
melodic-vegetable-78787
02/27/2025, 2:43 PMi dont want extra sensıtıve , i want noise supressıon , sometımes we are ın crowd agent stop speakıng because of thıs .
melodic-vegetable-78787
02/27/2025, 2:43 PMi dont want extra sensıtıve , i want noise supressıon , sometımes we are ın crowd agent stop speakıng because of thıs . @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 2:43 PMKalabalık ortamlarda gürültü bastırma sağlamak ve ajanın konuşmasını kesmesini önlemek için, LiveKit'in Krisp entegrasyonunu kullanabilirsiniz. Krisp, AI tabanlı gelişmiş gürültü bastırma teknolojisi sunar ve bu, özellikle kalabalık ortamlarda arka plan gürültüsünü azaltmada etkilidir. Krisp entegrasyonu, LiveKit Cloud projelerinde Scale veya Enterprise planlarında kullanılabilir.
Krisp entegrasyonunu etkinleştirmek için, LiveKit projenizde Krisp gürültü filtresini kurmanız ve etkinleştirmeniz gerekmektedir. İşte temel adımlar:
1. Krisp Gürültü Filtresini Kurun: Projenize Krisp gürültü filtresini ekleyin.
Copy code
import { KrispNoiseFilter } from '@livekit/krisp-noise-filter'; Copy code
room.on(RoomEvent.LocalTrackPublished, async (trackPublication) => {
if (trackPublication.source === Track.Source.Microphone && trackPublication.track instanceof LocalAudioTrack) {
const krispProcessor = KrispNoiseFilter();
await trackPublication.track.setProcessor(krispProcessor);
}
});m
melodic-vegetable-78787
02/27/2025, 2:45 PMim using python , and sip phone . @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 2:45 PMPython ve SIP telefon kullanıyorsanız, gürültü bastırma için Krisp entegrasyonunu doğrudan kullanamayabilirsiniz, çünkü Krisp genellikle tarayıcı tabanlı uygulamalar için uygundur. Ancak, Python tabanlı bir çözüm için aşağıdaki adımları deneyebilirsiniz:
1. Ses İşleme Kütüphaneleri: Python'da ses işleme için
pydubscipySilero VADactivation_thresholdnoisereducenoisereduce Copy code
import noisereduce as nr
import soundfile as sf
# Ses dosyasını yükle
data, rate = sf.read('input.wav')
# Gürültü bastırma
reduced_noise = nr.reduce_noise(y=data, sr=rate)
# Sonucu kaydet
sf.write('output.wav', reduced_noise, rate)m
melodic-vegetable-78787
02/27/2025, 7:12 PMSTANT", "pid": 30, "timestamp": "2025-02-27T190617.938206+00:00"}
landing-denis-call | {"message": "job process initialized", "level": "INFO", "name": "livekit.agents", "pid": 30, "timestamp": "2025-02-27T190617.938447+00:00"}
landing-denis-call | {"message": "initializing job process", "level": "INFO", "name": "livekit.agents", "pid": 41, "timestamp": "2025-02-27T190619.077826+00:00"}
landing-denis-call | {"message": "Prewarm completed: VAD loaded.", "level": "INFO", "name": "VOICE-ASSISTANT", "pid": 41, "timestamp": "2025-02-27T190619.125209+00:00"}
landing-denis-call | {"message": "job process initialized", "level": "INFO", "name": "livekit.agents", "pid": 41, "timestamp": "2025-02-27T190619.125484+00:00"}
landing-denis-call | {"message": "received job request", "level": "INFO", "name": "livekit.agents", "job_id": "AJ_skfCe3BjAnUi", "dispatch_id": "AD_h8ApkUvU7eF9", "room_name": "room-efhNH6RyGtio", "agent_name": "landing-denis-call", "resuming": false, "timestamp": "2025-02-27T190629.358466+00:00"}
landing-denis-call | {"message": "[OUTBOUND] connecting to room room-efhNH6RyGtio", "level": "INFO", "name": "VOICE-ASSISTANT", "pid": 18, "job_id": "AJ_skfCe3BjAnUi", "timestamp": "2025-02-27T190629.359745+00:00"}
landing-denis-call | {"message": "unhandled exception while running the job task\nTraceback (most recent call last):\n File \"/app/agent.py\", line 123, in entrypoint\n await ctx.api.room.update_room(\n ^^^^^^^^^^^^^^^^^^^^^^^^\nAttributeError: 'RoomService' object has no attribute 'update_room'", "level": "ERROR", "name": "livekit.agents", "pid": 18, "job_id": "AJ_skfCe3BjAnUi", "timestamp": "2025-02-27T190629.360087+00:00"}
landing-denis-call | {"message": "initializing job process", "level": "INFO", "name": "livekit.agents", "pid": 70, "timestamp": "2025-02-27T190630.492432+00:00"}
landing-denis-call | {"message": "Prewarm completed: VAD loaded.", "level": "INFO", "name": "VOICE-ASSISTANT", "pid": 70, "timestamp": "2025-02-27T190630.538732+00:00"}
landing-denis-call | {"message": "job process initialized", "level": "INFO", "name": "livekit.agents", "pid": 70, "timestamp": "2025-02-27T190630.538965+00:00"}
how to update room with parameters ?
melodic-vegetable-78787
02/27/2025, 7:13 PMSTANT", "pid": 30, "timestamp": "2025-02-27T190617.938206+00:00"}
landing-denis-call | {"message": "job process initialized", "level": "INFO", "name": "livekit.agents", "pid": 30, "timestamp": "2025-02-27T190617.938447+00:00"}
landing-denis-call | {"message": "initializing job process", "level": "INFO", "name": "livekit.agents", "pid": 41, "timestamp": "2025-02-27T190619.077826+00:00"}
landing-denis-call | {"message": "Prewarm completed: VAD loaded.", "level": "INFO", "name": "VOICE-ASSISTANT", "pid": 41, "timestamp": "2025-02-27T190619.125209+00:00"}
landing-denis-call | {"message": "job process initialized", "level": "INFO", "name": "livekit.agents", "pid": 41, "timestamp": "2025-02-27T190619.125484+00:00"}
landing-denis-call | {"message": "received job request", "level": "INFO", "name": "livekit.agents", "job_id": "AJ_skfCe3BjAnUi", "dispatch_id": "AD_h8ApkUvU7eF9", "room_name": "room-efhNH6RyGtio", "agent_name": "landing-denis-call", "resuming": false, "timestamp": "2025-02-27T190629.358466+00:00"}
landing-denis-call | {"message": "[OUTBOUND] connecting to room room-efhNH6RyGtio", "level": "INFO", "name": "VOICE-ASSISTANT", "pid": 18, "job_id": "AJ_skfCe3BjAnUi", "timestamp": "2025-02-27T190629.359745+00:00"}
landing-denis-call | {"message": "unhandled exception while running the job task\nTraceback (most recent call last):\n File \"/app/agent.py\", line 123, in entrypoint\n await ctx.api.room.update_room(\n ^^^^^^^^^^^^^^^^^^^^^^^^\nAttributeError: 'RoomService' object has no attribute 'update_room'", "level": "ERROR", "name": "livekit.agents", "pid": 18, "job_id": "AJ_skfCe3BjAnUi", "timestamp": "2025-02-27T190629.360087+00:00"}
landing-denis-call | {"message": "initializing job process", "level": "INFO", "name": "livekit.agents", "pid": 70, "timestamp": "2025-02-27T190630.492432+00:00"}
landing-denis-call | {"message": "Prewarm completed: VAD loaded.", "level": "INFO", "name": "VOICE-ASSISTANT", "pid": 70, "timestamp": "2025-02-27T190630.538732+00:00"}
landing-denis-call | {"message": "job process initialized", "level": "INFO", "name": "livekit.agents", "pid": 70, "timestamp": "2025-02-27T190630.538965+00:00"}
how to update room with parameters ? @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 7:13 PMIt seems that you are trying to use a method
update_roomRoomServiceUpdateRoomMetadataupdate_roomUpdateRoomMetadata Copy code
from livekit import api
async def update_room_metadata(ctx, room_name, new_metadata):
try:
await ctx.api.room.update_room_metadata(
api.UpdateRoomMetadataRequest(
room=room_name,
metadata=new_metadata
)
)
print(f"Room metadata updated for {room_name}")
except Exception as e:
print(f"Failed to update room metadata: {str(e)}")roomAdminm
melodic-vegetable-78787
02/27/2025, 7:15 PMhow to add sound specs to room ? @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 7:15 PMTo add sound specifications to a room in LiveKit, you can configure audio settings using
RoomOptionsRoomOptions Copy code
from livekit import rtc
# Define room options with specific audio settings
room_options = rtc.RoomOptions(
default_audio_capture_options=rtc.AudioCaptureOptions(
echo_cancellation=True,
noise_suppression=True,
auto_gain_control=True
),
default_audio_publish_options=rtc.AudioPublishOptions(
audio_bitrate=64000, # Set the desired audio bitrate
dtx=True # Enable discontinuous transmission
)
)
# Use these options when creating or updating a roomm
melodic-vegetable-78787
02/27/2025, 7:33 PMim using pyhton so how to add sound specs to room ? @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 7:33 PMTo configure sound specifications for a room in LiveKit using Python, you can set audio options when creating or managing a room. This involves specifying parameters for audio capture and publish options to control features like echo cancellation, noise suppression, and bitrate.
Here's how you can set these audio specifications using the LiveKit Python SDK:
Copy code
from livekit import rtc
# Define audio capture options
audio_capture_options = rtc.AudioCaptureOptions(
echo_cancellation=True, # Enable echo cancellation
noise_suppression=True, # Enable noise suppression
auto_gain_control=True # Enable automatic gain control
)
# Define audio publish options
audio_publish_options = rtc.AudioPublishOptions(
audio_bitrate=64000, # Set the desired audio bitrate
dtx=True # Enable discontinuous transmission
)
# Create room options with the specified audio settings
room_options = rtc.RoomOptions(
default_audio_capture_options=audio_capture_options,
default_audio_publish_options=audio_publish_options
)
# Use these options when creating or updating a roomm
melodic-vegetable-78787
02/27/2025, 7:35 PMis this params good for low latency outbound callls i need to use opus
encoding = protocol.egress.EncodingOptions( audio_bitrate=64000, # Agentın önerdiği 64 kbps bitrate audio_sample_rate=48000, # 48 kHz örnekleme oranı ) @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 7:35 PMOpus codec kullanarak düşük gecikmeli outbound çağrılar için belirttiğiniz parametreler genellikle uygundur. Opus, düşük gecikme ve yüksek ses kalitesi sağlamak için tasarlanmıştır. İşte parametrelerinizin değerlendirmesi:
1. audio_bitrate=64000: 64 kbps, Opus için iyi bir denge sağlar. Bu bitrate, konuşma için yeterli kaliteyi sunarken, bant genişliği kullanımını da optimize eder. Düşük gecikme için uygun bir değerdir.
2. audio_sample_rate=48000: 48 kHz örnekleme oranı, yüksek kaliteli ses sağlar ve Opus codec ile uyumludur. Bu örnekleme oranı, geniş bir frekans aralığını kapsar ve sesin netliğini artırır.
Bu ayarlar, düşük gecikmeli ve kaliteli ses iletimi için uygundur. Ancak, ağ koşullarınıza ve uygulama gereksinimlerinize göre ayarları test etmek önemlidir. Daha fazla ayrıntı veya başka bir konuda yardım ister misiniz?
Sources:
- https://livekit-users.slack.com/archives/C07FVFARRK7/p1730731323.248749
- https://livekit-users.slack.com/archives/C07FYCB7FRQ/p1728674567.746139 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.
m
melodic-vegetable-78787
02/27/2025, 7:36 PMHere's how you can set these audio specifications using the LiveKit Python SDK:
Copy code
from livekit import rtc
# Define audio capture options
audio_capture_options = rtc.AudioCaptureOptions(
echo_cancellation=True, # Enable echo cancellation
noise_suppression=True, # Enable noise suppression
auto_gain_control=True # Enable automatic gain control
)
# Define audio publish options
audio_publish_options = rtc.AudioPublishOptions(
audio_bitrate=64000, # Set the desired audio bitrate
dtx=True # Enable discontinuous transmission
)
# Create room options with the specified audio settings
room_options = rtc.RoomOptions(
default_audio_capture_options=audio_capture_options,
default_audio_publish_options=audio_publish_options
)
# Use these options when creating or updating a roomg
gray-monkey-83525
02/27/2025, 7:36 PMLiveKit Python SDK kullanarak ses özelliklerini ayarlamak için aşağıdaki kod örneğini kullanabilirsiniz. Bu örnek, ses yakalama ve yayınlama seçeneklerini ayarlayarak bir oda oluşturmayı veya güncellemeyi gösterir.
Copy code
from livekit import rtc
# Define audio capture options
audio_capture_options = rtc.AudioCaptureOptions(
echo_cancellation=True, # Yankı iptali etkinleştir
noise_suppression=True, # Gürültü bastırma etkinleştir
auto_gain_control=True # Otomatik kazanç kontrolü etkinleştir
)
# Define audio publish options
audio_publish_options = rtc.AudioPublishOptions(
audio_bitrate=64000, # İstenen ses bitrate'ini ayarla
dtx=True # Kesintili iletimi etkinleştir
)
# Create room options with the specified audio settings
room_options = rtc.RoomOptions(
default_audio_capture_options=audio_capture_options,
default_audio_publish_options=audio_publish_options
)
# Function to create or update a room with the specified options
async def create_or_update_room(api_client, room_name):
try:
# Create or update the room with the specified options
await api_client.room.create_room(
name=room_name,
options=room_options
)
print(f"Room '{room_name}' created or updated successfully with specified audio settings.")
except Exception as e:
print(f"Failed to create or update room: {str(e)}")
# Example usage
# Assuming you have an API client instance
# api_client = YourApiClientInstance()
# asyncio.run(create_or_update_room(api_client, "your-room-name"))m
melodic-vegetable-78787
02/27/2025, 7:41 PMno i want to use this inside my agent.py code ! @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 7:41 PMTo integrate the audio specifications directly into your
agent.pyagent.py Copy code
from livekit import rtc, api
from livekit.agents import JobContext, WorkerOptions, cli
# Define audio capture options
audio_capture_options = rtc.AudioCaptureOptions(
echo_cancellation=True, # Enable echo cancellation
noise_suppression=True, # Enable noise suppression
auto_gain_control=True # Enable automatic gain control
)
# Define audio publish options
audio_publish_options = rtc.AudioPublishOptions(
audio_bitrate=64000, # Set the desired audio bitrate
dtx=True # Enable discontinuous transmission
)
# Create room options with the specified audio settings
room_options = rtc.RoomOptions(
default_audio_capture_options=audio_capture_options,
default_audio_publish_options=audio_publish_options
)
async def entrypoint(ctx: JobContext):
# Connect to the room with the specified audio settings
await ctx.connect(
auto_subscribe=rtc.AutoSubscribe.AUDIO_ONLY,
room_options=room_options
)
# Your existing logic for handling participants and tracks
# ...
if __name__ == "__main__":
cli.run_app(WorkerOptions(entrypoint_fnc=entrypoint))room_optionsm
melodic-vegetable-78787
02/27/2025, 7:47 PMng-denis-call | ^^^^^^^^^^^^^^^^^^^^^^^
landing-denis-call | AttributeError: module 'livekit.rtc' has no attribute 'AudioCaptureOptions'
landing-denis-call | Traceback (most recent call last):
landing-denis-call | File "/app/agent.py", line 48, in <module>
landing-denis-call | audio_capture_options = rtc.AudioCaptureOptions(
landing-denis-call | ^^^^^^^^^^^^^^^^^^^^^^^
landing-denis-call | AttributeError: module 'livekit.rtc' has no attribute 'AudioCaptureOptions'
landing-denis-call | Traceback (most recent call last):
@millions-hair-24956
g
gray-monkey-83525
02/27/2025, 7:47 PMIt seems that the
AudioCaptureOptionslivekit.rtcAttributeError Copy code
pip install --upgrade livekitlivekit.rtc Copy code
import livekit.rtc
print(dir(livekit.rtc))AudioCaptureOptionsAudioCaptureOptionsm
melodic-vegetable-78787
02/27/2025, 7:50 PMlivekit>=0.18.2
livekit-agents>=0.12.1
livekit-plugins-openai>=0.10.9
livekit-plugins-deepgram>=0.6.13
livekit-plugins-silero>=0.7.4
livekit-plugins-google==0.10.1
livekit-plugins-elevenlabs==0.7.11
python-dotenv~=1.0
redis==4.5.1
grpcio
requests==2.32.3
pydub
wave
livekit-api
boto3
aiohttp
is my livekit latest ? @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 7:50 PMSorunuz için teşekkürler. Verdiğiniz sürüm bilgilerine göre, LiveKit ve ilgili eklentilerinizin sürümleri oldukça güncel görünüyor. Ancak, en son sürümleri kullandığınızdan emin olmak için aşağıdaki adımları izleyebilirsiniz:
1. Paketlerin Güncellenmesi: Aşağıdaki komutu kullanarak tüm paketlerinizi güncelleyebilirsiniz:
Copy code
pip install --upgrade livekit livekit-agents livekit-plugins-openai livekit-plugins-deepgram livekit-plugins-silero livekit-plugins-google livekit-plugins-elevenlabs Copy code
pip list | grep livekitm
melodic-vegetable-78787
02/27/2025, 7:51 PMim in docker @millions-hair-24956 how to check which livekit im usıng
g
gray-monkey-83525
02/27/2025, 7:51 PMDocker ortamında hangi LiveKit sürümünü kullandığınızı kontrol etmek için aşağıdaki adımları izleyebilirsiniz:
1. Docker Container İçine Giriş Yapın: İlk olarak, çalışmakta olan Docker konteynerinize giriş yapmanız gerekiyor. Bunu aşağıdaki komutla yapabilirsiniz:
Copy code
docker exec -it <container_id_or_name> /bin/bash<container_id_or_name> Copy code
pip list | grep livekitm
melodic-vegetable-78787
02/27/2025, 7:55 PMdocker exec -it livekit_server /bin/bash
OCI runtime exec failed: exec failed: unable to start container process: exec: "/bin/bash": stat /bin/bash: no such file or directory: unknown @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 7:55 PMBu hata, Docker konteynerinizde
/bin/bashbashshbashsh Copy code
docker exec -it <container_id_or_name> /bin/sh<container_id_or_name>shm
melodic-vegetable-78787
02/27/2025, 8:00 PM, "pid": 18, "job_id": "AJ_4xecx65wFytp", "timestamp": "2025-02-27T195838.442398+00:00"} landing-denis-call | {"message": "Starting recording for room: room-Ba6GxXTThvvc, filepath: outbound-905421950620-room-Ba6GxXTThvvc.m4a", "level": "INFO", "name": "VOICE-ASSISTANT", "pid": 18, "job_id": "AJ_4xecx65wFytp", "timestamp": "2025-02-27T195838.442505+00:00"} landing-denis-call | {"message": "unhandled exception while running the job task\nTraceback (most recent call last):\n File \"/app/agent.py\", line 178, in entrypoint\n egress_id = await record_call(ctx, phone_number)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/app/agent.py\", line 68, in record_call\n encoding = protocol.egress.EncodingOptions(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nValueError: Protocol message EncodingOptions has no \"audio_sample_rate\" field.", "level": "ERROR", "name": "livekit.agents", "pid": 18, "job_id": "AJ_4xecx65wFytp", "timestamp": "2025-02-27T195838.442622+00:00"}
you gıve me wrong parameters . @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 8:01 PMGörünüşe göre
EncodingOptionsaudio_sample_rateValueErrorEncodingOptions Copy code
from livekit import protocol
# Define encoding options without audio_sample_rate
encoding_options = protocol.egress.EncodingOptions(
audio_bitrate=64000 # Set the desired audio bitrate
# Remove audio_sample_rate as it is not supported
)
# Use encoding options in your egress request
file_output = protocol.egress.EncodedFileOutput(
filepath="outbound-905421950620-room-Ba6GxXTThvvc.m4a",
encoding_options=encoding_options
)
# Example egress request
egress_request = protocol.egress.RoomCompositeEgressRequest(
room_name="room-Ba6GxXTThvvc",
audio_only=True,
file_outputs=[file_output]
)
# Function to start recording
async def record_call(ctx, phone_number):
try:
res = await ctx.api.egress.start_room_composite_egress(egress_request)
print(f"Recording started with egress_id: {res.egress_id}")
return res.egress_id
except Exception as e:
print(f"Failed to start recording: {str(e)}")
raiseaudio_sample_rateEncodingOptionsValueErrorm
melodic-vegetable-78787
02/27/2025, 8:23 PM| 2025-02-27T201934.798Z INFO sip sip/inbound.go:173 processing invite {"nodeID": "NE_JqUsGrfwQTjf", "callID": "SCL_XubV68egB4ys", "fromIP": "162.243.10.169", "toIP": "18.184.217.4:5060", "fromHost": "1.1.1.1", "fromUser": "100", "toHost": "1.1.1.1", "toUser": "100", "sipTag": "31326238643930343133633401383835363736313535", "sipCallID": "1060962311795574933117669"}
sip_server | 2025-02-27T201934.802Z DEBUG sip sip/inbound.go:413 Rejecting inbound flood {"nodeID": "NE_JqUsGrfwQTjf", "callID": "SCL_XubV68egB4ys", "fromIP": "162.243.10.169", "toIP": "18.184.217.4:5060", "fromHost": "1.1.1.1", "fromUser": "100", "toHost": "1.1.1.1", "toUser": "100", "sipTag": "31326238643930343133633401383835363736313535", "sipCallID": "1060962311795574933117669"}
sip_server | 2025-02-27T201934.802Z INFO sip sip/inbound.go:714 Closing inbound call {"nodeID": "NE_JqUsGrfwQTjf", "callID": "SCL_XubV68egB4ys", "fromIP": "162.243.10.169", "toIP": "18.184.217.4:5060", "fromHost": "1.1.1.1", "fromUser": "100", "toHost": "1.1.1.1", "toUser": "100", "sipTag": "31326238643930343133633401383835363736313535", "sipCallID": "1060962311795574933117669", "reason": "flood"}
how to prevent this ? @millions-hair-24956
g
gray-monkey-83525
02/27/2025, 8:23 PMGörünüşe göre SIP sunucunuz, gelen çağrıları "flood" (aşırı yük) olarak algılıyor ve bu nedenle çağrılar reddediliyor. Bu tür sorunları önlemek için aşağıdaki adımları izleyebilirsiniz:
1. Rate Limiting (Hız Sınırlandırma): Gelen çağrılar için hız sınırlandırma uygulayarak, belirli bir süre içinde kabul edilebilecek maksimum çağrı sayısını sınırlayabilirsiniz. Bu, sunucunun aşırı yüklenmesini önlemeye yardımcı olabilir.
2. IP Whitelisting (IP Beyaz Listeleme): Sadece belirli IP adreslerinden gelen çağrılara izin vererek, istenmeyen veya kötü niyetli çağrıları engelleyebilirsiniz. Bu, güvenilir IP adreslerini beyaz listeye ekleyerek yapılabilir.
3. Firewall Ayarları: Güvenlik duvarı kurallarınızı gözden geçirerek, sadece belirli IP adreslerinden veya IP aralıklarından gelen SIP trafiğine izin verebilirsiniz.
4. SIP Trunk Ayarları: SIP trunk sağlayıcınızın ayarlarını kontrol edin ve sadece belirli numaralardan veya IP adreslerinden gelen çağrılara izin verildiğinden emin olun.
5. Log Analizi: Logları düzenli olarak analiz ederek, hangi IP adreslerinden veya numaralardan aşırı çağrı geldiğini tespit edebilir ve bu adresleri engelleyebilirsiniz.
Bu adımlar, SIP sunucunuzun aşırı yüklenmesini önlemeye yardımcı olabilir. Daha fazla ayrıntı veya başka bir konuda yardım ister misiniz?
Sources:
- https://livekit-users.slack.com/archives/C07FRP66XGE/p1731980365.334299 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.