GPU-Powered Real-Time Audio Source Separation

Date/Time: Monday, November 10, 2025 09:00 - 12:30 GMT

Room: SS Great Britain

Hosted by Chris D, Head of Pro Audio with technical presentation by Alexander Prokopchuk, CTO

In this hands-on workshop, we provide an overview of the GPU AUDIO SDK—its architecture, core features, and how it enables the development of accelerated audio modules. We'll explore how to integrate state-of-the-art, open-source neural network-based source separation models that utilize the power of your GPU.

 Key Elements

• Build a custom software stack and integrate with virtually any GPU-powered environment
• Achieve sub-millisecond latencies while running audio compute alongside other GPU workloads
• Replicate the setup on your own machine using the provided codebase and development environment

By the end of the session, you'll have a solid foundation in GPU-accelerated real-time audio source separation.