Precision Scaling of Neural Networks for Audio Processing

While deep neural networks have shown powerful performance in many audio applications, their large computation and memory demand has been a challenge for real-time processing. In this paper, we study the impact of scaling the precision of neural networks on the performance of two common audio processing tasks, namely, voice-activity detection and single-channel speech enhancement. We determine the optimal pair of weight/neuron bit precision by exploring its impact on both the processing performance and delay. Through experiments conducted with real user data, we demonstrate that deep neural networks that use lower bit precision significantly reduce processing delays (up to 30x). However, their performance impact is low (< 9.54%) only in the case of classification tasks such as those present in voice-activity detection.