Hello, I´m new in the community :) I´m playing with pipecat and livekit in self-hosted and the problem I have is that I need compatibility with Zadarma SIP (https://zadarma.com/) and I can´t use it :_( I need to use zadarma because Twilio and Tlenyx are so expensive and also I can´t buy phones from a specific part of Spain. Do some one use zadarma SIP with them? Thanks!!😁

Google released the latest version of the Gemini Live model today. This is still the 2.5 series (not based on Gemini 3, yet) Using the AI Studio APIs, the model name is: gemini-2.5-flash-native-audio-preview-12-2025 I've been experimenting with the checkpoints for a couple of weeks, and it's pretty similar to the previous version. (flash-native-audio-preview-09-2025). They focused a lot on tool calling reliability in this release, because everybody told them they needed to make tool calling more reliable! It's definitely better on their benchmarks, but I generally found you could prompt the model pretty well to do tool calling until you get fairly deep in a multi-turn conversation, in which case all bets are off. What we really need from this API is good context engineering capabilities! Interestingly, the model is also GA today (not preview) on Google Cloud Vertex. The Vertex model name is gemini-live-2.5-flash-native-audio I don't really understand the Vertex thinking, here, other than that Vertex can give you contracts for committed TPU capacity, which removes one of the variables that makes it hard to deliver reliable latencies for a model like this. You can try the new model at https://www.pipecat.ai/

Are you using tracing in your voice agent? I thought about this today, because The team at LangChain built voice AI support into their agent debugging and monitoring tool, LangSmith. LangSmith is built around the concept of "tracing." If you've used OpenTelemetery for application logging, you're already familiar with tracing. If you haven't, think about it like this: a trace is a record of an operation that an application performs. Today's production voice agents are complex, multi-model, multi-modal, multi-turn systems! Tracing gives you leverage to understand what your agents are doing. This saves time during development. And it's critical in production. You can dig into what happened during each turn of any session. What did the user say and how was that processed by each model you're using in your voice agent? What was the latency for each inference operation? What audio and text was actually sent back to the user? You can also run analytics using tracing as your observability data. And you can use traces to build evals. Tanushree is an engineer at LangChain. Her video below shows using a local (on-device) model for transcription, then switching to using the OpenAI speech-to-text model running in the cloud. You can see the difference in accuracy. (Using Pipecat, switching between different models is a single-line code change.) Also, the video is fun! It's a French tutor. Which is a voice agent I definitely need. How to debug voice agents with LangSmith (video): https://youtu.be/0FmbIgzKAkQ LangSmith Pipecat integration docs page: https://docs.langchain.com/langsmith/trace-with-pipecat I always like to read the code for nifty Pipecat services like the LangSmith tracing processor. It's here, though I think this nice work will likely make its way into Pipecat core soon: https://github.com/langchain-ai/voice-agents-tracing/blob/main/pipecat/langsmith_processor.py

I mean a system where one can do 10k calls per day. Has anyone built a system like this using livekit and pipecat. did you do it without using our own GPUs ?

Has any one built an voice bot without pipecat and livekit would love to connect with them I have built it and facing some issues in latency part My tech stack is openai llm , Deepgram for ASR and Azure for TTS and one telephony vendor which connects everything