Sopro TTS: A 169M model with zero-shot voice cloning that runs on the CPU

Re:Sopro TTS: A 169M model with zero-shot voice cloning that... by sammyyyyyyy on Thursday January 08, 2026 @04:58PM (#46547077)

This is my side “hobby”. And compute is quite expensive. But if the community’s responsive is good, I will definitely think about it! Btw, chatterbox is a great model and inspiration

[ Parent | on HN ]

Re:Sopro TTS: A 169M model with zero-shot voice cloning that... by bicepjai on Thursday January 08, 2026 @06:40PM (#46548114)

Thanks can you share details about compute economics you dealt with ?

[ Parent | on HN ]

Re:Sopro TTS: A 169M model with zero-shot voice cloning that... by sammyyyyyyy on Thursday January 08, 2026 @06:43PM (#46548128)

Yeah sure. The training was about ~250 dollars, which is quite low by today’s standards. And I spent a bit more on ablations and research

[ Parent | on HN ]

Re:Sopro TTS: A 169M model with zero-shot voice cloning that... by bicepjai on Sunday January 11, 2026 @06:29PM (#46581660)

I was on similar path and saw my bills going over 1000 dollars as interests to do research and ablations grew. Then I decided to get one Blackwell Pro 6000 and trying things with that :) If you have suggestions on how to manage metrics let us know. Currenty trying langfuse since its one click install on coolify

[ Parent | on HN ]

Re:Sopro TTS: A 169M model with zero-shot voice cloning that... by btbuildem on Monday January 12, 2026 @10:57AM (#46590158)

Is that something that could be done on a local setup? Eg, 2x RTX3090?

[ Parent | on HN ]

Re:Sopro TTS: A 169M model with zero-shot voice cloning that... by littlestymaar on Friday January 09, 2026 @12:21AM (#46550334)

Very cool work, especially for a hobby project.

Do you have any plans to publish a blog post on how you did that? ?What training data and how much? Your training and ablations methodology, etc.

[ Parent | on HN ]