Lune

EMNLP Findings2025

MetaMixSpeech: Meta Task Augmentation for Low-Resource Speech Recognition

Yaqi Chen, Hao Zhang, Wenlin Zhang, XuKui Yang, Dan Qu, Yunpeng Liu

Abstract

Meta-learning has proven to be a powerful paradigm for effectively improving the performance of low-resource speech recognition by learning generalizable knowledge across multiple tasks. However, multilingual meta learning also faces challenges such as task overfitting and learner overfitting, thereby reducing its ability to generalize to new tasks. To address these issues, we augment the metatraining task with "more data" during both training and evaluation phases. Concretely, we propose an interpolation-based task augmentation method called MetaMixSpeech, which includes both support augmentation and query augmentation. MetaMixSpeech enhances task diversity by linearly combining perturbed features from the support and query sets and performing the same linear interpolation on their corresponding losses. Experimental results on the FLEURS and Common Voice datasets demonstrate that MetaMixSpeech achieves a 6.35 % improvement in Word Error Rate (WER) compared to meta-learning approaches, effectively mitigating the overfitting problem and showcasing superior generalization across diverse datasets and language families.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

Questions to start from

Lune can draw on

the full paper · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Improving Generalization in Meta-learning via Task Augmentation cites Meta-Learning Requires Meta-Augmentation
  • Meta-Learning Requires Meta-Augmentation cites Meta-Learning without Memorization
  • Robust Speech Recognition via Large-Scale Weak Supervision cites wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
Builds onCited byLeft to right by year; the more cited, the higher.

Builds on7

  1. wav2vec 2.0: A Framework for Self-Supervised Learning of Speech RepresentationsAlexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, Michael AuliNeurIPS 2020 · 9,451 citations
  2. Robust Speech Recognition via Large-Scale Weak SupervisionAlec Radford, Jong Wook Kim, Tao Xu, Greg Brockman et al.ICML 2023 · 6,966 citations
  3. Meta-Learning without MemorizationMingzhang Yin, George Tucker, Mingyuan Zhou, Sergey Levine et al.ICLR 2020 · 201 citations
  4. Meta-Learning Requires Meta-AugmentationJanarthanan Rajendran, Alexander Irpan, Eric JangNeurIPS 2020 · 111 citations
  5. Improving Generalization in Meta-learning via Task AugmentationHuaxiu Yao, Long-Kai Huang, Linjun Zhang, Ying Wei et al.ICML 2021 · 98 citations

Related papers

  1. Adversarial Meta Sampling for Multilingual Low-Resource Speech RecognitionYubei Xiao, Ke Gong, Pan Zhou, Guolin Zheng et al.AAAI 2021 · 37 citations
  2. Set-based Meta-Interpolation for Few-Task Meta-LearningSeanie Lee, Bruno Andreis, Kenji Kawaguchi, Juho Lee et al.NeurIPS 2022 · 13 citations
  3. Learning from Multiple Noisy Augmented Data Sets for Better Cross-Lingual Spoken Language UnderstandingYingmei Guo, Linjun Shou, Jian Pei, Ming Gong et al.EMNLP 2021 · 2 citations
  4. Entropy-guided Vocabulary Augmentation of Multilingual Language Models for Low-resource TasksArijit Nag, Bidisha Samanta, Animesh Mukherjee, Niloy Ganguly et al.ACL Findings 2023 · 3 citations
  5. Improving Meta-learning for Low-resource Text Classification and Generation via Memory ImitationYingxiu Zhao, Zhiliang Tian, Huaxiu Yao, Yinhe Zheng et al.ACL 2022 · 21 citations
Dusk over the sea between two cliffs drawn in fine vertical lines