Children's Speech Recognition through Discrete Token Enhancement

Children's speech recognition is considered a low-resource task mainly due to the lack of publicly available data. There are several reasons for such data scarcity, including expensive data collection and annotation processes, and data privacy, among others. Transforming speech signals into discrete tokens that do not carry sensitive information but capture both linguistic and acoustic information could be a solution for privacy concerns. In this study, we investigate the integration of discrete speech tokens into children's speech recognition systems as input without significantly degrading the ASR performance. Additionally, we explored single-view and multi-view strategies for creating these discrete labels. Furthermore, we tested the models for generalization capabilities with unseen domain and nativity dataset. Results reveal that the discrete token ASR for children achieves nearly equivalent performance with an approximate 83% reduction in parameters.

• Publication year

  2024

• Venue

  Interspeech

• Publication date

  2024-06-19

• Fields of study

  Computer Science, Engineering, Education

• Identifiers
  DOI 10.48550/arXiv.2406.13431arXiv 2406.13431
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Beyond Orthography: Automatic Recovery of Short Vowels and Dialectal Sounds in Arabic

  2024cited by this paper
• Exploring Speech Recognition, Translation, and Understanding with Discrete Speech Units: A Comparative Study

  2023cited by this paper
• My Science Tutor (MyST)–a Large Corpus of Children’s Conversational Speech

  2023cited by this paper
• Kid-Whisper: Towards Bridging the Performance Gap in Automatic Speech Recognition for Children VS. Adults

  2023cited by this paper
• Robust Self Supervised Speech Embeddings for Child-Adult Classification in Interactions involving Children with Autism

  2023cited by this paper
• Developmental Articulatory and Acoustic Features for Six to Ten Year Old Children

  2023cited by this paper
• BabySLM: language-acquisition-friendly benchmark of self-supervised spoken language models

  2023cited by this paper
• Exploration of Efficient End-to-End ASR using Discretized Input from Self-Supervised Learning

  2023cited by this paper
• DRAFT: A Novel Framework to Reduce Domain Shifting in Self-supervised Learning and Its Application to Children's ASR

  2022cited by this paper
• A WAV2VEC2-Based Experimental Study on Self-Supervised Learning Methods to Improve Child Speech Recognition

  2022cited by this paper
• Audio Augmentation for Non-Native Children’s Speech Recognition through Discriminative Learning

  2022cited by this paper
• Robust Speech Recognition via Large-Scale Weak Supervision

  2022cited by this paper
• E-Branchformer: Branchformer with Enhanced Merging for Speech Recognition

  2022cited by this paper
• Towards Better Domain Adaptation for Self-Supervised Models: A Case Study of Child ASR

  2022cited by this paper
• Branchformer: Parallel MLP-Attention Architectures to Capture Local and Global Context for Speech Recognition and Understanding

  2022cited by this paper
• Low Resource German ASR with Untranscribed Data Spoken by Non-native Children - INTERSPEECH 2021 Shared Task SPAPL System

  2021cited by this paper
• WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing

  2021cited by this paper
• Analysis of Disfluency in Children's Speech

  2020cited by this paper
• wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations

  2020cited by this paper
• ESPnet: End-to-End Speech Processing Toolkit

  2018cited by this paper
• On the Difficulties of Automatic Speech Recognition for Kindergarten-Aged Children

  2018cited by this paper
• Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates

  2018cited by this paper
• Attention is All you Need

  2017cited by this paper
• A survey about databases of children's speech

  2013cited by this paper
• Automated speech scoring for non-native middle school students with multiple task types

  2013cited by this paper
• Why and How Our Automated Reading Tutor Listens

  2012cited by this paper
• Speech production variability in fricatives of children and adults: results of functional data analysis.

  2008cited by this paper
• Stop-consonant voicing and intraoral pressure contours in women and children.

  2008cited by this paper
• Vowel acoustic space development in children: a synthesis of acoustic and anatomic data.

  2007cited by this paper
• Multi-view clustering

  2004cited by this paper
• LANGUAGE AND DISFLUENCY IN NONSTUTTERING CHILDREN'S CONVERSATIONAL SPEECH

  1999cited by this paper
• Acoustics of children's speech: developmental changes of temporal and spectral parameters.

  1999cited by this paper
• Analysis of children's speech: duration, pitch and formants

  1997cited by this paper
• Relationships between duration and temporal variability in children's speech.

  1992cited by this paper
• Vector quantization in speech coding

  1985cited by this paper

• Phone-purity Guided Discrete Tokens for Dysarthric Speech Recognition

  2025cites this paper
• Can Layer-Wise SSL Features Improve Zero-Shot ASR Performance for Children’s Speech?

  2025cites this paper
• From CHAT towards ASR: A Hybrid Pipeline for Constructing the HUKILC-CO Hungarian Child Speech Dataset

  2025cites this paper
• Comparing Unsupervised and Supervised Semantic Speech Tokens: A Case Study of Child ASR

  2025cites this paper
• Beyond Orthography: Automatic Recovery of Short Vowels and Dialectal Sounds in Arabic

  2024cites this paper