mlsanityData to 7 Oct 2026

UML: A Universal Monolingual Output Layer for Multilingual ASR

arXiv 2302.11186 ↗First posted 22 Feb 2023eess.AS cs.CL cs.SDcited by 3

Not on the map. Why?

Abstract (arXiv)

Word-piece models (WPMs) are commonly used subword units in state-of-the-art end-to-end automatic speech recognition (ASR) systems. For multilingual ASR, due to the differences in written scripts across languages, multilingual WPMs bring the challenges of having overly large output layers and scaling to more languages. In this work, we propose a universal monolingual output layer (UML) to address such problems. Instead of one output node for only one WPM, UML re-associates each output node with multiple WPMs, one for each language, and results in a smaller monolingual output layer shared across languages. Consequently, the UML enables to switch in the interpretation of each output node depending on the language of the input speech. Experimental results on an 11-language voice search task demonstrated the feasibility of using UML for high-quality and high-efficiency multilingual streaming ASR.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 18 of its 36 references that are in the map. First posted on arXiv 22 Feb 2023.

Builds on

18 of 36 references.

References · 18

  1. 2015Neural Machine Translation of Rare Words with Subword Unitscited by 2,898
  2. 2018Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidatescited by 632
  3. 2020Massively Multilingual ASR: 50 Languages, 1 Model, 1 Billion Parameterscited by 110
  4. 2018Exploring Architectures, Data and Units For Streaming End-to-End Speech Recognition with RNN-Transducercited by 95
  5. 2017Multilingual Speech Recognition With A Single End-To-End Modelcited by 87
  6. 2020Universal Phone Recognition with a Multilingual Allophone Systemcited by 81
  7. 2021Scaling End-to-End Models for Large-Scale Multilingual ASRcited by 61
  8. 2020Cascaded encoders for unifying streaming and non-streaming ASRcited by 51
  9. 2018Improved training of end-to-end attention models for speech recognitioncited by 35
  10. 2018On the End-to-End Solution to Mandarin-English Code-switching Speech Recognitioncited by 30
  11. 2019Massively Multilingual Adversarial Speech Recognitioncited by 28
  12. 2021A Configurable Multilingual Model is All You Need to Recognize All Languagescited by 28
6 more
  1. 2020Transformer-Transducers for Code-Switched Speech Recognitioncited by 27
  2. 2020Less Is More: Improved RNN-T Decoding Using Limited Label Context and Path Mergingcited by 26
  3. 2022Streaming End-to-End Multilingual Speech Recognition with Joint Language Identificationcited by 24
  4. 2017Towards Language-Universal End-to-End Speech Recognitioncited by 14
  5. 2022Improving the fusion of acoustic and text representations in RNN-Tcited by 11
  6. 2022Adaptive multilingual speech recognition with pretrained modelscited by 9

Built on by

Cited by year

Citing papers per year: 2023: 1; 2024: 1; 2025: 1.

3 papers.

  1. 2024Learn and Don't Forget: Adding a New Language to ASR Foundation Modelscited by 16
  2. 2025Low-Rank and Sparse Model Merging for Multi-Lingual Speech Recognition and Translationcited by 6
  3. 2023Building High-accuracy Multilingual ASR with Gated Language Experts and Curriculum Trainingcited by 4