APSIPA Transactions on Signal and Information Processing > Vol 11 > Issue 1

Recent Advances in End-to-End Automatic Speech Recognition

Industrial Technology Advances

Jinyu Li, Microsoft, USA, jinyli@microsoft.com
 
Suggested Citation
Jinyu Li (2022), "Recent Advances in End-to-End Automatic Speech Recognition", APSIPA Transactions on Signal and Information Processing: Vol. 11: No. 1, e8. http://dx.doi.org/10.1561/116.00000050

Publication Date: 20 Apr 2022
© 2022 J. Li
 
Subjects
 
Keywords
End-to-endautomatic speech recognitionstreamingattentiontransducertransformeradaptation
 

Share

Open Access

This is published under the terms of CC BY-NC.

Downloaded: 18321 times

In this article:
Introduction 
End-to-End Models 
Encoder 
Other Training Criterion 
Multilingual Modeling 
Adaptation 
Advanced Models 
Miscellaneous Topics 
Conclusions and Future Directions 
References 

Abstract

Recently, the speech community is seeing a significant trend of moving from deep neural network based hybrid modeling to end-to-end (E2E) modeling for automatic speech recognition (ASR). While E2E models achieve the state-of-the-art results in most benchmarks in terms of ASR accuracy, hybrid models are still used in a large proportion of commercial ASR systems at the current time. There are lots of practical factors that affect the production model deployment decision. Traditional hybrid models, being optimized for production for decades, are usually good at these factors. Without providing excellent solutions to all these factors, it is hard for E2E models to be widely commercialized. In this paper, we will overview the recent advances in E2E models, focusing on technologies addressing those challenges from the industry’s perspective.

DOI:10.1561/116.00000050

Replication Data | 116.00000050_supp.zip (ZIP).

This file contains the data that is required to replicate the data on your own system.

DOI: 10.1561/116.00000050_supp