arXiv subjects

Paban Sapkota

Publications and source records attributed to Paban Sapkota.

5 recordsLinked to original sources

arXiv2026

Autoencoder based optimized SSL representations: Complexity Minimization and improved Dysarthric ASR

Self-supervised learning (SSL) models extract rich speech representations but often come with high-dimensional features, increasing computational complexity. This work explores an SSL-AutoEncoder (SSL-AE) bottlenecking approach to efficiently reduce feature dimensions while maintaining dysarthric Automatic Speech Recognition (ASR) performance. By leveraging an autoencoder, we transform high-dimensional SSL features into a compact space, reducing model complexity and training time. Our method preserves essential speech information, achieving reduced Word Error Rates (WER) while significantly lowering computational costs. Experiments show SSL-AE bottlenecking reduces training time by 8x compared to the SSL baseline, demonstrating efficiency without sacrificing recognition performance. These results highlight AE as an effective solution for SSL feature compression in resource-constrained environments.

eess.AS↗

arXiv2026

DSSCNet: A Transfer Learning Framework for Cross-Corpus Dysarthric Speech Severity Classification

Dysarthric speech severity classification is challenging due to speaker variability, class imbalance, and limited datasets. This study introduces DSSCNet, a deep learning model that employs transfer learning and multi-corpus learning to enhance speaker-independent classification. By pre-training on one dysarthric speech corpus and fine-tuning on another, DSSCNet achieves improved feature extraction and cross-corpus generalization. Experimental results demonstrate that DSSCNet outperforms state-of-the-art models for speaker-independent severity classification, achieving 75.80\% accuracy on TORGO and 68.25\% on UA-Speech, significantly reducing misclassification errors. The findings confirm that leveraging knowledge transfer between datasets improves model robustness, making DSSCNet well-suited for automated dysarthria assessment. This research contributes to the development of more effective assistive speech technologies for individuals with speech impairments.

eess.AS↗

arXiv2026

Systematic Study of Dysarthric Speech Recognition: Spectral Features and Acoustic Models

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.

eess.AS↗

arXiv2026

Improving End-to-End Speech Recognition for Dysarthric Speech through In-Domain Data Augmentation

Dysarthric speech recognition is crucial for facilitating effective communication among individuals with dysarthria. However, accurately recognizing dysarthric speech poses significant challenges due to varying severity levels and limited data availability. In this paper, we explore data augmentation techniques for dysarthric automatic speech recognition (ASR) systems by fine-tuning the End-to-End pre-trained Wav2Vec2 model, with a specific focus on severity levels. To address the challenges of data scarcity and the need for extensive data in fine-tuning pre-trained ASR systems for dysarthric speech, we investigate four prominent data augmentation methods: Speaking-Rate Modification (SRM), Pitch Modification (PM), Formant Modification (FM), and vocal tract Length Perturbation (VTLP), tailored to different aspects of dysarthria. The study uses individually fine-tuned Wav2Vec2 models for each severity class as baseline systems. Additionally, we conducted severity-specific fine-tuning of the ASR model using augmented data. Results demonstrate distinct efficacy patterns for each augmentation technique across severity levels. The best WERs were achieved with SRM ($s$=0.8) for \textit{low} (9.02\%) and \textit{medium} (38.11\%) severities, and with PM ($\tau$=0.8) for \textit{high} severity (55.15\%), reflecting relative improvements of 30.02\%, 16.64\%, and 15.47\%, respectively. These results confirm the effectiveness of the augmentation methods in improving dysarthric ASR performance.

eess.AS↗

arXiv2025

Enhancing Speaker-Independent Dysarthric Speech Severity Classification with DSSCNet and Cross-Corpus Adaptation

Dysarthric speech severity classification is crucial for objective clinical assessment and progress monitoring in individuals with motor speech disorders. Although prior methods have addressed this task, achieving robust generalization in speaker-independent (SID) scenarios remains challenging. This work introduces DSSCNet, a novel deep neural architecture that combines Convolutional, Squeeze-Excitation (SE), and Residual network, helping it extract discriminative representations of dysarthric speech from mel spectrograms. The addition of SE block selectively focuses on the important features of the dysarthric speech, thereby minimizing loss and enhancing overall model performance. We also propose a cross-corpus fine-tuning framework for severity classification, adapted from detection-based transfer learning approaches. DSSCNet is evaluated on two benchmark dysarthric speech corpora: TORGO and UA-Speech under speaker-independent evaluation protocols: One-Speaker-Per-Severity (OSPS) and Leave-One-Speaker-Out (LOSO) protocols. DSSCNet achieves accuracies of 56.84% and 62.62% under OSPS and 63.47% and 64.18% under LOSO setting on TORGO and UA-Speech respectively outperforming existing state-of-the-art methods. Upon fine-tuning, the performance improves substantially, with DSSCNet achieving up to 75.80% accuracy on TORGO and 68.25% on UA-Speech in OSPS, and up to 77.76% and 79.44%, respectively, in LOSO. These results demonstrate the effectiveness and generalizability of DSSCNet for fine-grained severity classification across diverse dysarthric speech datasets.

eess.AS↗