Hi team, How can I find the outbound SIP IP addresses for my LiveKit Cloud project? I'm integrating with a SIP provider (Exacall) that requires whitelisting a fixed outbound IP for authorizing outbound calls.

Is anyone here building a telephony agent using Gemini 3.1 Flash Live ? I built one but the latency is high nearly 2 seconds. I am looking for a fix and don't really know what to do. I used Google-Livekit Plugin hosted it on Livekit cloud ( used Livekit's native telephony provider ) , VAD = 0.3 seconds only. I cannot think of any other thing that I can tweak. I am from India so the data has to travel a long distance but VAPI has a latency of only about 1.2 seconds with similar setup. Any ideas on how I can fix this ?

Hello everyone, can you help me to find in documentation/examples to check session is inbound or outbound in a agent entrypoint

Don't use Gemini 3.1 Flash Live for Telephony

@

When I first saw the model release, I thought it would make for a fantastic telephony agent—but that is not the case. I’m not entirely sure why, but the response generation time is consistently between 1.6 to 2.4 seconds. This is only the model-side latency. On top of that, additional delays come from: • Transport time • Telephony layer • LiveKit Cloud transmission • Other pipeline components These further increase the total end-to-end latency. To be clear, the 1.6–2.4 seconds reflects: Time for LiveKit to send data to Google’s service • Time for Gemini 3.1 Flash Live to generate a response • Time to transmit the response back to LiveKit Cloud The transmission time between Google’s service and LiveKit Cloud is approximately ~30 ms, which is negligible compared to the model latency. I’ve attached PCAP files from test calls. These can be analyzed (e.g., via Wireshark or even an LLM) to verify the timings. At this point, there is essentially nothing to optimize on our end. The bottleneck is the model’s response time, which is outside our control. Also, to rule out obvious issues: • “Thinking” is turned OFF • No misconfiguration affecting response time • Multiple checks were performed to ensure consistency So this latency is inherent to the current behavior of Gemini 3.1 Flash Live. PCAP File of test call: ( u can ask an LLM to analyse it )

How can I get assistance from someone to help on a SIP trunk after the bot assist here (ask-ai) just keeps running in circles?

This message was deleted.

Hi! Hitting SIP/2.0 503 Try again later on every Twilio inbound call to my LiveKit Cloud project — would love a sanity check from anyone who's done Twilio → LiveKit recently. Setup: • LiveKit Cloud inbound URI (from Project Settings): gwmt8ehuoo6.sip.livekit.cloud • Twilio number → TwiML returns <Dial><Sip secure="true">sip:+1XXX...@...sip.livekit.cloud</Sip></Dial> • Inbound trunk has E.164 number in

numbers

, allowed_addresses=['0.0.0.0/0'], dispatch rule → room • Agent worker running, [Router] log confirms tenant mapping • Also tried Twilio Elastic SIP Trunk with Secure Trunking → same 503 What the PCAP shows: • INVITE arrives on UDP:5060 (despite secure="true" in TwiML) • LiveKit → 100 Processing • Then → 503 Try again later Before fixing the URI (was using a wrong subdomain earlier), I got 404 No trunk found. Fixing URI resolved 404 → now 503. Two questions: 1. Does LiveKit Cloud inbound require TLS (port 5061) and reject UDP:5060 with 503? 2. Or is 503 pointing at a downstream dispatch/worker issue I should check? Happy to DM the PCAP + trunk dump. Thanks in advance!

Hola tengo un troncal sip que se conecta desde issabel 4 a livekit, el levanta pero a las 24 horas se cae y no puede conectar nuevamente, existe alguna lista de direcciones IP permitidas o lista negra, o falta algun parmetro adicional a lo normales

Currently I am using telnyx and Twilio. What are other cheapest and the options that have good latency ?

Hi any Indian Voice Builders , I have question best ways to do call Australian number to test ur agents .

I see jio and airtel have got ISD plans but setup is tol painful

Hi, SIP inbound returns 503 but no record in Calls dashboard Project: alertivo (id: p_23wunvd1nk2), region Germany 2. Inbound SIP calls from Twilio Elastic SIP Trunk fail with SIP 503, but nothing appears in the Calls dashboard (Telephony → Calls, past 24h shows “No results”). I’m at my wit’s end on diagnosis. Setup: - Twilio origination URL: sip:23wunvd1nk2.sip.livekit.cloud;transport=tcp - Inbound trunk ST_tjNxEUu9gGtf, numbers +420910923523,420910923523, no auth, encryption DISABLE - Dispatch rule SDR_g3rVynwCybMM, Individual (Caller), roomPrefix: “call-“, references the trunk, room_config.agents: [{agent_name: “alertivo-support”}] - Agent worker registered as agent_name: “alertivo-support” (livekit-agents 1.5.4, AgentServer + @breezy-insurance-66772.rtc_session(agent_name=...) pattern) - Twilio call SID example: CA097cd1e8a4f4a2f391f7534c7a033f62, returns: 503 service error with sip:+420910923523@23wunvd1nk2.sip.livekit.cloud;transport=tcp What works: - Manual dispatch via lk dispatch create --agent-name alertivo-support --room ... works end-to-end — agent receives job, connects Deepgram STT, etc. - SIP OPTIONS over TCP to 23wunvd1nk2.sip.livekit.cloud:5060 returns SIP/2.0 200 OK - SIP URI on Trunks page shows sip:23wunvd1nk2.sip.livekit.cloud (no custom port) - UDP 5060 times out — confirming TCP-only behavior What doesn’t: - TCP INVITE returns 503 every time (5x retries visible in Twilio debugger) - Calls dashboard shows nothing for these attempts — so the INVITE never becomes a registered “call” in LiveKit Any idea why the SIP edge returns 503 without registering the call? Is there a known issue with room_config.agents auto-dispatch for SIP inbound, or a specific flag I’m missing?

Is there a way to do SIP transfers with extension? Example:

VDN : 9993119982
Skill : 1018108

We’re integrating LiveKit voice agents with our telephony platform (Invoca) and running into an issue where tel: URI parameters in the Refer-To header are being stripped by the LiveKit SIP bridge before they reach our downstream FreeSWITCH server. What we’re doing: Our voice agent calls transfer_participant with a Refer-To value of: tel:+15559876543;x-ai-agent-outcome=transfer%3ATech%20Support We can confirm via our agent logs that the SDK is sending this correctly (transfer_sip_refer_sending event shows the full URI with params). What we see on the other side: When the SIP REFER arrives at our FreeSWITCH, sip_refer_to contains only: tel:+15559876543 The ;x-ai-agent-outcome=transfer%3ATech%20Support parameter is dropped somewhere in the LiveKit SIP infrastructure between the agent SDK and the outbound REFER message. Why we need this: We’re using tel: URI parameters (per RFC 3966) to pass metadata like agent outcome back to our call platform during blind transfers. FreeSWITCH reliably propagates sip_refer_to — the params just need to survive the LiveKit SIP bridge. Environment: LiveKit Cloud, using livekit-agents Python SDK, SIP trunk via UDP to FreeSWITCH. Is this a known limitation? Is there a way to preserve tel: URI parameters through the SIP REFER, or an alternative mechanism we should use? Already posted here, but still no response.

How can we purchase or add phone lines? Seems limited at 10 lines even on the Scale plan?

I am trying to connect asterisk with livekit, from asterisk (PBX) the call status is ringing but on livekit cloud, no logs are showing under telephony

Hi everyone, I’m testing a French voice agent over phone calls and I’m having an audio capture issue: the LLM/STT often doesn’t hear the speaker well unless the caller speaks quite loudly. What we tested so far: • We removednoise_cancellationfromRoomInputOptions, and that improved things noticeably on desktop / local microphone input. • However, on smartphone calls through telephony, the problem is still there: the caller is sometimes only partially heard, or not heard well enough unless speaking loudly. • We are using: ◦ deepgram.STT(model="nova-3", language="fr") ◦ silero.VAD.load() ◦ turn_detection="vad" • We also checked Deepgram model options, but for French there doesn’t seem to be a dedicated phone-call model like the Englishnova-2-phonecall. • On the telephony side, we are also looking at codec ordering on the SIP trunk (G711U,G711A,G722,G729) to see whether that helps. My question is: What is the best way in LiveKit to improve how well the agent hears a caller on phone calls, especially when the incoming voice level is low? More specifically: • Are there recommended LiveKit settings for telephony audio input sensitivity? • Should we tune Silero VAD for this kind of issue, or is that mostly turn detection rather than capture? • Any known best practices for Telnyx + LiveKit + Deepgram when callers are not picked up reliably unless they speak loudly? Any guidance would be really appreciated.

Hello everyone, I’m currently building an outbound voice agent using LiveKit Agents with a Telnyx SIP trunk, and I’m experiencing a consistent audio issue on specific PSTN numbers. Issue Summary: The agent’s voice becomes choppy and gets cut intermittently during calls. The user reports difficulty hearing clearly. What’s interesting: • Calls to my Zoom Phone number work perfectly (clear audio, no cutting). • Calls to my personal verizon phone numbers consistently experience audio cutting on every call. Environment: • LiveKit Agents version: 1.5.6 • Region: India West • SIP Provider: Telnyx • TTS: Cartesia (also tested with slower speeds) • STT: Deepgram • No background audio enabled • Noise cancellation tested both ON and OFF • Conservative turn handling (no aggressive interruptions) What I’ve already tested: • Disabled background audio completely • Reduced TTS speed • Adjusted turn handling for stability • Disabled noise cancellation • Added delay before agent speaks • Issue persists only on certain PSTN numbers Request: Could anyone help investigate if this might be related to: • RTP packetization or jitter buffering • SIP trunk routing differences • Codec negotiation (e.g., PCMU vs Opus) • Any known issues with Telnyx → US carrier routing I am attaching a video recording of the issue. Thanks in advance for your help. @clean-businessperson-32244 @refined-appointment-81829 @lively-winter-88668 @gray-napkin-98176 @freezing-france-64950 @alert-architect-29784 @loud-dawn-97520 @dry-elephant-14928 @bored-accountant-30277 @polite-rain-22350 @elegant-businessperson-51313

I'm trying to get inbound calls to work, but am getting an error - "Room subscription timed out" The Agent status is in pending status. Is that expected?

Hey guys I have been using gpt 4o mini as the model; sarvam v3 as TTS and Deepgram STT Nova 3 and am getting abnormal latency. Is it because of livekit or something with my code. And when I am using tool calling thats taking like 10-10 sec. The voice quality also isnt very good , it keeps on breaking. What should I do/ optimise??

Hi there! Looking for a sanity check / DM partner on a fairly urgent inbound SIP issue. Pattern I am seeing: - Inbound INVITE via Telnyx FQDN connection → LK Cloud - LK edge returns 404 within ~60–240ms before any session is created - Dashboard shows 0 SIP sessions over 7 days despite multiple test calls - Telnyx debugger confirms LK is the one 404'ing (not a Telnyx-side issue) Already verified via Twirp API: - Trunk: numbers match the inbound DID, allowedAddresses: 0.0.0.0/0, krispEnabled _- Dispatch rule: room_config.agents properly nested with the agent_name_ _- Pipecat worker online and registered with that agent_name_ Emailed Support yesterday and got auto-reply, no ticket # yet, follow-up sent. I have a demo Thursday so a bit time-sensitive. 😬 _*Common causes I might be missing?* Project tier gate? INVITE Request-URI normalization (Telnyx sending 8175551234 vs trunk holding +1817_5551234_)? Regional SIP edge routing?_ Happy to share trunk ID / dispatch rule ID / project ID in DM with anyone who has bandwidth to dig in — would rather not paste them publicly. Thanks 🙏

This message was deleted.

Hey 👋 — running into a Cloud Agents + SIP dispatch issue I can't crack. Inbound calls hit my LiveKit edge fine (100 Processing) but always get 503 Try again later even though my Cloud Agent is Running with 1/1/1 replicas. Manual lk dispatch create works perfectly; only the SIP-driven dispatch path returns 503. lk CLI: 2.16.3 Trunk config: media_encryption: DISABLE Dispatch rule config: trunkIds: [ST_zMLJfuZoMHmE] inbound_numbers: Direct SIP probe via openssl s_client to sip:+18313152253@6zow2x1cz42.sip.livekit.cloud:5061 (TLS) reproduces the 503 every time: ← SIP/2.0 100 Processing ← SIP/2.0 503 Try again later What works: function tools firing within our allowlist) What doesn't: results for past 24h despite multiple SIP attempts (so the 503 happens before the call hits the project's call history) Question: what does 503 Try again later mean in the LiveKit Cloud SIP dispatch path, and is there an additional configuration step required to bind the deployed Cloud Agent (CA_uqGP6MwtWSeV) to the SIP dispatch rule (SDR_kahRSNTMBowp)? Manual API dispatches succeed, only SIP fails — so the binding/match seems fine, but something in the SIP→agent handoff isn't. Thanks 🙏

Hello there! I'm facing

Unexpected status from INVITE response: sip status: 503: CPS Limit reached P05

error, this happened suddenly, is there any place to increase this limit?

"I have an FQDN SIP connection pointing to LiveKit Cloud (

3cetozcjms6.sip.livekit.cloud

, port 5060). Inbound calls to my Greek number +302318510764 are immediately rejected with Q.850 cause code 1 (0 seconds duration). The FQDN resolves correctly and ports 5060/5061 are reachable. Can you check the SIP Call Flow for Telnyx ID

234e4d8a-51e5-11f1-a2f6-da4d813c7a6f

and tell me what response LiveKit is returning to the INVITE?"

We successfully receive WhatsApp inbound call webhooks from Meta and can hit the call accept API, but we’re blocked on generating valid SDP answers. Python SDK only exposes SIP APIs and Node SDK lacks WhatsApp call wrappers. Is there a supported WhatsApp media bridge or beta API for this?

Hiya team! Did any changes get made to SIP REFERs on the LK SIP gateway today?

Hi any downtime in SIP inbound transfers today on LK ?

archived the channel