Optimized Hybrid Model for Enhanced Parkinson’s Disease Classification Using Feature Fused Voice Signal DOI Open Access

S. Sharanyaa,

M Sambath

International Journal of Electronics and Communication Engineering, Journal Year: 2023, Volume and Issue: 10(11), P. 11 - 26

Published: Nov. 30, 2023

Parkinson’s Disease (PD) is a common neuro disorder that leads to reduced nerve function in the brain as result of decreased dopamine generation. The disease progressive, and patients may have difficulty speaking, resulting speech variations. Hence, it essential detect at an early stage, through proper diagnosis, effect can be controlled. This work aims classify PD based on vocal feature set using hybrid CNN-ALSTM model. model trained with Spectral, Acoustic, Mel-Spectrogram features obtained from de-noised voice signals. proposed involves four phases. In first phase, signals are extracted input data, de-noising done Improved Optimized Variational Mode Decomposition (IO-VMD). second Mel-Spectrograms generated pre-processed where deep Custom CNN, EfficientNetB0, Inceptionv3 models. third metaheuristic Squirrel Search Water Cycle Algorithm (SSWA) applied vectors, SSWA used for selection hyper parameter tuning. Finally, spectral acoustic concatenated mel spectrogram trained, classified Attention Long Short Term Memory (ALSTM) A comparative analysis models like CNN-ALSTM, Inceptionv3- ALSTM, EfficientNetB0-ALSTM carried out PD. From analysis, algorithm achieves accuracy 96.8% performs better than other

Language: Английский

Diagnosis of pathological speech with streamlined features for long short-term memory learning DOI Creative Commons
Tuan D. Pham, Simon Holmes,

Lifong Zou

et al.

Computers in Biology and Medicine, Journal Year: 2024, Volume and Issue: 170, P. 107976 - 107976

Published: Jan. 8, 2024

Pathological speech diagnosis is crucial for identifying and treating various disorders. Accurate aids in developing targeted intervention strategies, improving patients' communication abilities, enhancing their overall quality of life. With the rising incidence speech-related conditions globally, including oral health, need efficient reliable diagnostic tools has become paramount, emphasizing significance advanced research this field.

Language: Английский

Citations

9
Bio-inspired optimization of feature selection and SVM tuning for voice disorders detection DOI
Maria Habib, Víctor Vicente-Palacios, Pablo García‐Sánchez

et al.

Knowledge-Based Systems, Journal Year: 2025, Volume and Issue: unknown, P. 112950 - 112950

Published: Jan. 1, 2025

Language: Английский

Citations

1
Pathological Voice Classification Using MEEL Features and SVM-Tabnet Model DOI
Mohammed Zakariah, Muna Al‐Razgan, Taha Alfakih

et al.

Speech Communication, Journal Year: 2024, Volume and Issue: 162, P. 103100 - 103100

Published: July 1, 2024

Language: Английский

Citations

4
Smartphone-derived multidomain features including voice, finger-tapping movement and gait aid early identification of Parkinson’s disease DOI Creative Commons
Wei Ling Florence Lim, Sung‐Pin Fan,

Shu-I Chiu

et al.

npj Parkinson s Disease, Journal Year: 2025, Volume and Issue: 11(1)

Published: May 5, 2025

Language: Английский

Citations

0
Automatic cross‐ and multi‐lingual recognition of dysphonia by ensemble classification using deep speaker embedding models DOI Creative Commons
Dosti Aziz, Dávid Sztahó

Expert Systems, Journal Year: 2024, Volume and Issue: 41(10)

Published: June 12, 2024

Abstract Machine Learning (ML) algorithms have demonstrated remarkable performance in dysphonia detection using speech samples. However, their efficacy often diminishes when tested on languages different from the training data, raising questions about suitability clinical settings. This study aims to develop a robust method for cross‐ and multi‐lingual that overcomes limitation of language dependency existing ML methods. We propose an innovative approach leverages embeddings speaker verification models, especially ECAPA x‐vector employs majority voting ensemble classifier. utilize features extracted train three distinct classifiers. The significant advantage these embedding models lies capability capture characteristics language‐independent manner, forming fixed‐dimensional feature spaces. Additionally, we investigate impact generating synthetic data within space Synthetic Minority Oversampling Technique (SMOTE). Our experimental results unveil effectiveness proposed detection. Compared obtained embeddings, consistently demonstrates superior distinguishing between healthy dysphonic speech, achieving accuracy values 93.33% 96.55% both cross‐lingual scenarios, respectively. highlights capabilities ECAPA, capturing enhance overall performance. effectively addresses challenges combined with classifiers, show potential improving reliability scenarios.

Language: Английский

Citations

1
Beyond breathalyzers: AI-powered speech analysis for alcohol intoxication detection DOI
Federica Amato, Valerio Cesarini, Gabriella Olmo

et al.

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 262, P. 125656 - 125656

Published: Nov. 6, 2024

Language: Английский

Citations

1
Reverb and Noise as Real-World Effects in Speech Recognition Models: A Study and a Proposal of a Feature Set DOI Creative Commons
Valerio Cesarini, Giovanni Costantini

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(23), P. 11446 - 11446

Published: Dec. 9, 2024

Reverberation and background noise are common unavoidable real-world phenomena that hinder automatic speaker recognition systems, particularly because these systems typically trained on noise-free data. Most models rely fixed audio feature sets. To evaluate the dependency of features reverberation noise, this study proposes augmenting commonly used mel-frequency cepstral coefficients (MFCCs) with relative spectral (RASTA) features. The performance was assessed using noisy data generated by applying pink to DEMoS dataset, which includes 56 speakers. Verification were clean MFCCs, RASTA features, or their combination as inputs. They validated augmented progressively increasing levels. results indicate MFCCs struggle identify main speaker, while method has difficulty opposite class. hybrid set, derived from combination, demonstrates best overall a compromise between two. Although MFCC is standard performs well training data, it shows significant tendency misclassify in scenarios, critical limitation for modern user-centric verification applications. therefore, proves effective balanced solution, optimizing both sensitivity specificity.

Language: Английский

Citations

1
Optimized Hybrid Model for Enhanced Parkinson’s Disease Classification Using Feature Fused Voice Signal DOI Open Access

S. Sharanyaa,

M Sambath

International Journal of Electronics and Communication Engineering, Journal Year: 2023, Volume and Issue: 10(11), P. 11 - 26

Published: Nov. 30, 2023

Parkinson’s Disease (PD) is a common neuro disorder that leads to reduced nerve function in the brain as result of decreased dopamine generation. The disease progressive, and patients may have difficulty speaking, resulting speech variations. Hence, it essential detect at an early stage, through proper diagnosis, effect can be controlled. This work aims classify PD based on vocal feature set using hybrid CNN-ALSTM model. model trained with Spectral, Acoustic, Mel-Spectrogram features obtained from de-noised voice signals. proposed involves four phases. In first phase, signals are extracted input data, de-noising done Improved Optimized Variational Mode Decomposition (IO-VMD). second Mel-Spectrograms generated pre-processed where deep Custom CNN, EfficientNetB0, Inceptionv3 models. third metaheuristic Squirrel Search Water Cycle Algorithm (SSWA) applied vectors, SSWA used for selection hyper parameter tuning. Finally, spectral acoustic concatenated mel spectrogram trained, classified Attention Long Short Term Memory (ALSTM) A comparative analysis models like CNN-ALSTM, Inceptionv3- ALSTM, EfficientNetB0-ALSTM carried out PD. From analysis, algorithm achieves accuracy 96.8% performs better than other

Language: Английский

Citations

0