Noisy Training Improves E2E ASR for the Edge

Abstract

Automatic speech recognition (ASR) has become increasingly ubiquitous on modern edge devices. Past work developed streaming End-to-End (E2E) all-neural speech recognizers that can run compactly on edge devices. However, E2E ASR models are prone to overfitting and have difficulties in generalizing to unseen testing data. Various techniques have been proposed to regularize the training of ASR models, including layer normalization, dropout, spectrum data augmentation and speed distortions in the inputs. In this work, we present a simple yet effective noisy training strategy to further improve the E2E ASR model training. By introducing random noise to the parameter space during training, our method can produce smoother models at convergence that generalize better. We apply noisy training to improve both dense and sparse state-of-the-art Emformer models and observe consistent WER reduction. Specifically, when training Emformers with 90% sparsity, we achieve 12% and 14% WER improvements on the LibriSpeech Test-other and Test-clean data set, respectively.

Low-Rank+Sparse Tensor Compression for Neural Networks

“…Using machine learning to enable multiple applications on edge devices requires multiple task-specific persistent models (Yang et al, 2020). These models are used for tasks ranging from computer vision (Howard et al, 2019) to automatic speech recognition (Wang et al, 2021). The trend towards multiple applications and multiple models is constrained by the fact that off-chip memory reads incur high latency and power costs (Sze et al, 2017).…”