Skip to main navigation Skip to search Skip to main content

Empowering Māori Automatic Speech Recognition through EMD-Based Augmentation

  • Chengxi Lei*
  • , Sheng Li
  • , Satwinder Singh
  • , Feng Hou
  • , Huia Jahnke
  • , Ruili Wang
  • *Corresponding author for this work

Research output: Chapter in Book/Conference proceedingConference contributionpeer-review

Abstract

Low-resource languages like Māori face significant challenges in developing robust Automatic Speech Recognition (ASR) systems due to limited annotated data and linguistic resources. This paper proposes a novel data augmentation framework that enriches training data for ASR models through Empirical Mode Decomposition (EMD) based frequency band perturbation. EMD is employed to decompose speech signals into intrinsic mode functions (IMFs), enabling selective removal of specific frequency components to simulate variations in speaker traits and acoustic environments. Experiments on a self-collected 17-hour Māori speech corpus demonstrate consistent improvements across three ASR architectures, including DeepSpeech, Wav2Vec 2.0 XLS-R, and HuBERT. The proposed method significantly reduces Word Error Rates (WER), especially when combined with SpecAugment, underscoring its complementary benefits and effectiveness in enhancing generalization for Māori ASR.

Original languageEnglish
Title of host publicationPRICAI 2025
Subtitle of host publicationTrends in Artificial Intelligence - 22nd Pacific Rim International Conference on Artificial Intelligence, PRICAI 2025, Proceedings
EditorsYi Mei, Bing Xue, Chao Qian, Quan Bai, Sankalp Khanna
PublisherSpringer Science and Business Media Deutschland GmbH
Pages676-683
Number of pages8
ISBN (Print)9789819570713
DOIs
Publication statusPublished - 2026
Externally publishedYes
Event22nd Pacific Rim International Conference on Artificial Intelligence, PRICAI 2025 - Wellington, New Zealand
Duration: 17 Nov 202521 Nov 2025

Publication series

NameLecture Notes in Computer Science
Volume16452 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference22nd Pacific Rim International Conference on Artificial Intelligence, PRICAI 2025
Country/TerritoryNew Zealand
CityWellington
Period17/11/2521/11/25

Free Keywords

  • Data Augmentation
  • Empirical Mode Decomposition (EMD)
  • Low-resource Māori ASR

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Empowering Māori Automatic Speech Recognition through EMD-Based Augmentation'. Together they form a unique fingerprint.

Cite this