Gen-A: Nokia's Universal Solution for Spatial Audio Encoding

Опубликовано 11 ноября 2025, 10:46

Our research solves a critical limitation in spatial audio processing. Although deep neural networks surpass traditional methods, they must be trained separately for each microphone array, making the process expensive and slow. Our Gen-A method introduces a universal solution that works with any microphone configuration without retraining. Using an innovative multi-level encoder, it processes both microphone geometry and audio signals simultaneously, delivering superior performance across all frequencies.

Discover the full research on by Mikko Heikkinen, Archontis Politis, Konstantinos Drossos and Tuomas Virtanen.

nokia.com/standardization/tech...