INRC Forum March 21th: Intel Neuromorphic DNS Challenge

A critical enabler for progress in neuromorphic computing research is challenge problems with real-world relevance, full-system evaluations, and state-of-the-art reference solutions. Thus, to help foster neuromorphic innovation, we introduce the Intel Neuromorphic Deep Noise Suppression Challenge (Intel N-DNS Challenge). The Intel N-DNS Challenge is inspired by the Microsoft DNS Challenge and tackles a ubiquitous and commercially relevant task: real-time audio denoising. Audio denoising is likely to reap the benefits of neuromorphic computing due to its low-bandwidth, temporal nature and its relevance for low-power devices.

The Intel N-DNS Challenge has two tracks: a simulation-based track to encourage algorithmic innovation, and a neuromorphic hardware track to rigorously evaluate solutions on Loihi 2. Both tracks are evaluated based on energy, latency, and resource consumption in addition to output audio quality. We provide dataset scripts and evaluation code, and we release a neuromorphic baseline solution with promising audio quality, high power efficiency, and low resource consumption compared to Microsoft NsNet2 and a production Intel denoising model. We also present $55k in total prizes to the winners. We hope the winners of the N-DNS challenge will demonstrate that for problems like audio denoising, significant gains in power and resource efficiency can be realized on neuromorphic systems.