11.30 AM - 12.00 PM
Aula Magna

Infernos: cost efficient AI inference for real-time applications

Maksym Sobolev @ Sippy Software

In this presentation I will give a brief overview of the state of the art landscape of ML frameworks in general and models for text-to-speech and speech-to-text in particular. I would also describe some of the challenges deploying those models and having them integrated into a server-side real-time application today. Then I would go to more in-depth details of how we addressed those changes in Infernos.

Long term OpenSIPS / Kamailio / FreeBSD contributor, open source aficionado. Author and maintainer of few open source projects.