Cross-Dataset, Age, and Gender Generalization: A Comprehensive Analysis of Fine-Tuning Strategies for Low-Resource Children's ASR

2026-06-18

AI Summary

This work addresses the challenge of poor generalization across datasets, ages, and genders in child speech recognition under low-resource conditions, primarily caused by high acoustic variability in disordered articulation. To enhance robustness, the study proposes a novel approach based on Factorized Time-Delay Neural Networks (F-TDNN), which uniquely integrates pitch as an explicit acoustic feature, employs a chunked overlapping-frame training strategy, and incorporates hybrid DNN/HMM sequence-discriminative training. Evaluated on the TORGO database, the proposed method achieves significant improvements, reducing the relative word error rate by 4.65% for isolated words and by 4.63% for sentence-level recognition, thereby effectively mitigating a core bottleneck in dysarthric speech recognition.

Abstract

The challenge associated with recognizing dysarthric speech primarily arises from pronounced acoustic variability attributed to impaired articulatory precision. Past research has demonstrated improved recognition through the use of hybrid DNN/HMM sequence discriminative training. This paper presents a comprehensive investigation of various combinations of acoustic features tailored to different Acoustic Models, offering suitable feature selections for each. The incorporation of Pitch features notably improved recognition performance, especially for sentence recognition tasks involving dysarthric speech. Through a systematic examination of the TORGO database, we have demonstrated the potential to enhance the performance of the state-of-the-art Factorized Time Delay Neural Network (F-TDNN) model for recognizing dysarthric speech. Our methods, implemented with the F-TDNN model, resulted in a 4.65\% relative improvement in isolated word recognition and a 4.63\% relative improvement in sentence recognition for dysarthric speech, compared to previous research. This improvement effectively compensates for speech variability, attributable to our deliberate selection of the number of overlapping frames between consecutive training example chunks.

Problem

• dysarthric speech
• acoustic variability
• automatic speech recognition
• articulatory impairment
• speech recognition performance

Innovation

• dysarthric speech recognition
• Pitch features
• F-TDNN
• acoustic feature selection
• frame overlap optimization