Back to Search Start Over

Collaborative Training of Acoustic Encoders for Speech Recognition

Authors :
Nagaraja, Varun
Shi, Yangyang
Venkatesh, Ganesh
Kalinli, Ozlem
Seltzer, Michael L.
Chandra, Vikas
Publication Year :
2021

Abstract

On-device speech recognition requires training models of different sizes for deploying on devices with various computational budgets. When building such different models, we can benefit from training them jointly to take advantage of the knowledge shared between them. Joint training is also efficient since it reduces the redundancy in the training procedure's data handling operations. We propose a method for collaboratively training acoustic encoders of different sizes for speech recognition. We use a sequence transducer setup where different acoustic encoders share a common predictor and joiner modules. The acoustic encoders are also trained using co-distillation through an auxiliary task for frame level chenone prediction, along with the transducer loss. We perform experiments using the LibriSpeech corpus and demonstrate that the collaboratively trained acoustic encoders can provide up to a 11% relative improvement in the word error rate on both the test partitions.<br />Comment: INTERSPEECH 2021

Details

Database :
arXiv
Publication Type :
Report
Accession number :
edsarx.2106.08960
Document Type :
Working Paper