Hosting Sesame 1B CSM in a voice chat app
Hi! I’m building a voice chat app similar to ChatGPT’s advanced voice or Sesame AI’s own demo — real-time back-and-forth conversation with an LLM, where every model reply is converted to speech using the 1B Sesame TTS model. Each user chats ~20 min/day. Latency needs to stay under 300ms. What would be the estimated monthly cost using Inference Endpoints (e.g., L4 or A100) at 1k–20k daily active users? Overall if anyone with expereience could guide me a bit, I would be soo thankful! It would really help to know approximately, how much does it cost for Sesame AI to host their own demo per user? Thank you!
hi,
I am curious to know the answer, as I am also wanting to build a similar or same chat bot,
let me know if you get a answer.
thank you
hey, sure, mind letting me know as well when you have the answer?
let me know too!
Please let me know as well