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

TL;DR

This paper presents a highly efficient end-to-end speech recognition system optimized for smart devices, combining model compression and weight sharing to drastically reduce size with minimal accuracy loss.

Contribution

It introduces a novel low-footprint E2E ASR model using cross-layer weight sharing and compression techniques, enabling deployment on resource-constrained devices.

Findings

Achieves over 10x model size reduction on AISHELL-2

Maintains near-original accuracy with only 0.43% CER increase

Demonstrates effective deployment feasibility on smart devices

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%).