Arabic speech recognition by end-to-end, modular systems and human

Amir Hussein*, Shinji Watanabe, Ahmed Ali

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Recent advances in automatic speech recognition (ASR) have achieved accuracy levels comparable to human transcribers, which led researchers to debate if the machine has reached human performance. Previous work focused on the English language and modular hidden Markov model-deep neural network (HMM–DNN) systems. In this paper, we perform a comprehensive benchmarking for end-to-end transformer ASR, modular HMM–DNN ASR, and human speech recognition (HSR) on the Arabic language and its dialects. For the HSR, we evaluate linguist performance and lay-native speaker performance on a new dataset collected as a part of this study. For ASR the end-to-end work led to 12.5%, 27.5%, 33.8% WER; a new performance milestone for the MGB2, MGB3, and MGB5 challenges respectively. Our results suggest that human performance in the Arabic language is still considerably better than the machine with an absolute WER gap of 3.5% on average.

Original languageEnglish
Article number101272
JournalComputer Speech and Language
Volume71
DOIs
Publication statusPublished - 2022 Jan
Externally publishedYes

Keywords

  • Dialectal arabic
  • End-to-end speech recognition
  • Human speech recognition
  • Modern standard arabic
  • Transformer

ASJC Scopus subject areas

  • Software
  • Theoretical Computer Science
  • Human-Computer Interaction

Fingerprint

Dive into the research topics of 'Arabic speech recognition by end-to-end, modular systems and human'. Together they form a unique fingerprint.

Cite this