Hey everyone 👋 I’m trying to figure out how to build my own voice AI agent, something similar to VAPI Retell..., but custom for my language and accents. We have over 40 accents, and I want to fine-tune a model that can actually understand and speak like people from each region — not just transcribe words, but really adapt to local pronunciation, tone, and vocabulary. I have experience in full-stack development (React, Spring Boot, Go, etc.), but I’ve never done fine-tuning or building voice pipelines before. So my main questions are: - 🧠 What’s the best way to start building a custom voice AI like this? - 🗣️ How can I collect and fine-tune on many accents effectively? - ⚙️ What models, tools, or frameworks do you recommend for speech-to-text, text-to-speech, and conversation handling? - 💾 Do I need to build my own dataset, or can I use / adapt open ones? If anyone here has worked on custom speech AI or VAPI-like systems, I’d really appreciate some pointers 🙏 Thanks in advance!— Khaled Rouissi 🚀

You’re probably still building AI voice agents manually for clients. That’s not a scalable business. It’s time to take your AI agency to the next level by launching your own full-stack SaaS platform — just like I did. The good news: I’ve already built the entire AI voice agent SaaS platform, so you don’t have to. Here’s what’s inside the codebase: 🗣 AI Voice Agent — powered by Vapi & LiveKit⚙ Configurations — prompts, models, knowledge base, and more 📝 Call Logs 📈 Analytics Dashboard 🔐 Sign in / Sign up — Google Authentication ready 💳 Payment Collection — via Stripe 🏢 Multi-Tenant Architecture — one user can be linked to multiple organizations 📞 SIP Connection — integrated with Telnyx ⛔ Rate Limiter — manage usage efficiently And everything else you need to launch a production-ready SaaS platform. 💬 Comment below or DM me if you’re interested in using or acquiring the full codebase for your own AI voice agent business.

This is the space to share your Voice AI wins and fails! Yes, fails too. You don’t often see a “fails” category, but that’s exactly why we need one. Every success story starts with a few stumbles. By sharing your failures, you’re helping others learn faster, stay motivated, and avoid the same pitfalls you faced. Stay tuned — I’ll be sharing my first fail story soon, haha!

I'm experiencing unexpected disconnections in both my **text chat agent** and **voice agent** built with Pipecat. The pipeline is being cancelled due to **"idle timeout detected"** even though: :white_check_mark: Active conversation is ongoing :white_check_mark: LLM is processing/responding :white_check_mark: No actual idle time has occurred :x: Disconnects happen randomly (sometimes within **1 second!**) Error from logs: WARNING | pipecat.pipeline.task:_idle_timeout_detected:831 - Idle timeout detected. WARNING | pipecat.pipeline.task:_idle_timeout_detected:834 - Idle pipeline detected, cancelling pipeline What I've Tried - Checked for idle timeout configurations (not set anywhere in my code) - Verified WebSocket/Daily transport settings - Reviewed pipeline configuration - No obvious long-running API calls causing delays Environment - **Text agent:** Custom WebSocket transport - **Voice agent:** Daily WebRTC transport - **Framework:** Pipecat Flows - **Behavior:** Happens in both test and production environments Question Has anyone encountered this? Is there a hidden **idle timeout setting** in Pipecat that needs to be configured? Any suggestions would be greatly appreciated! :pray: **Screenshots attached:** Chat agent logs + Voice agent logs showing the disconnection.

Feel free to post any job or partnership offers. Here is a short guide that can help you: - What’s the project or goal? - Are you offering a job, collaboration, or partnership? - What kind of skills or expertise are you seeking? - Is it paid, equity-based, or volunteer? - When you need someone to start or how long the project will take - Specify if it's remote, hybrid, or location-specific - Share how to get in touch (in comments, dm, email, link etc.)