Source separation performance has seen tremendous progress in recent years thanks to advances in deep learning technologies. Its real-time setup would give musicians new ways to interact with music, e.g. in djing and music learning. In this talk, we will cover the challenges we had to tackle in order to provide our users with real-time low-latency source separation on iOS mobile devices.
We will talk about:
- state-of-the-art deep learning-based source separation algorithms
- model optimisation for iOS mobile platforms (Core ML and ANE)
- low-level integration for real-time use-cases
IF YOU ARE ATTENDING ONLINE, ALL TALK SESSIONS CAN BE ACCESSED FROM THE MAIN LOBBY:
https://conference.audio.dev