Extremely Low Footprint End-to-End ASR System for Smart Device

Abstract

Recently, end-to-end (E2E) speech recognition has become popular, since it can integrate the acoustic, pronunciation and language models into a single neural network, which outperforms conventional models. Among E2E approaches, attention-based models, e.g. Transformer, have emerged as being superior. Such models have opened the door to deployment of ASR on smart devices, however they still suffer from requiring a large number of model parameters. We propose an extremely low footprint E2E ASR system for smart devices, to achieve the goal of satisfying resource constraints without sacrificing recognition accuracy. We design cross-layer weight sharing to improve parameter efficiency and further exploit model compression methods including sparsification and quantization, to reduce memory storage and boost decoding efficiency. We evaluate our approaches on the public AISHELL-1 and AISHELL-2 benchmarks. On the AISHELL-2 task, the proposed method achieves more than 10x compression (model size reduces from 248 to 24MB), at the cost of only minor performance loss (CER reduces from 6.49% to 6.92%).

arXiv comment5 pages, 2 figures, accepted by INTERSPEECH 2021

References in corpus (10)

• SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
• Attention-Based Models for Speech Recognition
• Sequence Transduction with Recurrent Neural Networks
• SNIP: Single-shot Network Pruning based on Connection Sensitivity
• Online and Linear-Time Attention by Enforcing Monotonic Alignments
• AISHELL-2: Transforming Mandarin ASR Research Into Industrial Scale
• The Evolved Transformer
• Streaming Chunk-Aware Multihead Attention for Online End-to-End Speech Recognition
• SAN-M: Memory Equipped Self-Attention for End-to-End Speech Recognition
• Evaluating the Performance of NVIDIA's A100 Ampere GPU for Sparse Linear Algebra Computations