Show HN: Real time AI voice chat with Elon

[arielcai]

Hey everyone! We've just rolled out a fun Valentine's Day project: an Elon Musk chatbot. Our goal was to capture Elon's unique voice, adding a bit more depth and charm, and to ensure chatting with AI in real-time is a smooth experience.

We're using GPT-3.5 (faster than 4), Deepgram for ASR, and lmnt-tts-0131 for TTS. Our average latency is 300ms. We plan to open-source the project after February 14th. Check it out at https://www.sexyelon.ai/ and share your thoughts on latency, voice quality or anything else.

Are you building a chatbot with voices? We're happy to share our experiences and discuss any challenges if that’s helpful. Drop me a line!

[jansenderr]

There seems to be many ways to vary this tech stack. What is the logic behind each of the decisions you made?

[arielcai]

That's a long story. Any project you're currently working on that our decision-making process could help with?

[jansenderr]

Trying to build a real-time telephone agent. So the same requirements as your project except going through twilio I would think

[dxcqcv]

Do you have plan to open source?

[arielcai]

Definitely! We'll open source on Feb 15 and share the GitHub link here. If you're using Replit, our deployed demo is live: https://sexy-elon-lmnt.replit.app/