Learning ASR pathways: A sparse multilingual ASR model
This addresses the challenge of efficient and effective multilingual ASR for applications requiring support for multiple languages, especially low-resource ones, though it is incremental as it builds on existing pruning techniques.
The paper tackles the problem of multilingual automatic speech recognition (ASR) by introducing ASR pathways, a sparse model that activates language-specific sub-networks to avoid performance drops from language-agnostic pruning, resulting in better performance on low-resource languages compared to dense and pruned models.
Neural network pruning compresses automatic speech recognition (ASR) models effectively. However, in multilingual ASR, language-agnostic pruning may lead to severe performance drops on some languages because language-agnostic pruning masks may not fit all languages and discard important language-specific parameters. In this work, we present ASR pathways, a sparse multilingual ASR model that activates language-specific sub-networks ("pathways"), such that the parameters for each language are learned explicitly. With the overlapping sub-networks, the shared parameters can also enable knowledge transfer for lower-resource languages via joint multilingual training. We propose a novel algorithm to learn ASR pathways, and evaluate the proposed method on 4 languages with a streaming RNN-T model. Our proposed ASR pathways outperform both dense models and a language-agnostically pruned model, and provide better performance on low-resource languages compared to the monolingual sparse models.