Towards Lifelong Learning of End-to-end ASR

TL;DR

This paper explores lifelong learning approaches for end-to-end automatic speech recognition, aiming to adapt to new environments without forgetting previous knowledge, and reports significant improvements over traditional fine-tuning methods.

Contribution

It introduces novel methods for saving past data to mitigate catastrophic forgetting in lifelong learning for E2E ASR and provides extensive analysis of various approaches.

Findings

Achieved 28.7% relative WER reduction over fine-tuning baseline

First comprehensive analysis of lifelong learning in E2E ASR

Demonstrated effectiveness across diverse benchmark datasets

Abstract

Automatic speech recognition (ASR) technologies today are primarily optimized for given datasets; thus, any changes in the application environment (e.g., acoustic conditions or topic domains) may inevitably degrade the performance. We can collect new data describing the new environment and fine-tune the system, but this naturally leads to higher error rates for the earlier datasets, referred to as catastrophic forgetting. The concept of lifelong learning (LLL) aiming to enable a machine to sequentially learn new tasks from new datasets describing the changing real world without forgetting the previously learned knowledge is thus brought to attention. This paper reports, to our knowledge, the first effort to extensively consider and analyze the use of various approaches of LLL in end-to-end (E2E) ASR, including proposing novel methods in saving data for past domains to mitigate the catastrophic forgetting problem. An overall relative reduction of 28.7% in WER was achieved compared to the fine-tuning baseline when sequentially learning on three very different benchmark corpora. This can be the first step toward the highly desired ASR technologies capable of synchronizing with the continuously changing real world.