Development of a regional voice dataset and speaker classification based on machine learning

Ismail, Muhammad; Memon, Saima Q.; Dhomeja, Lachhman Das; Shah, Shahid Munir; Hussain, Dostdar; Rahim, Sabit; Ali, İmran

doi:10.1186/s40537-021-00435-9

Cited by 10 publications

(2 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Secondly, the traditional bio-inspired algorithm known as Mel Frequency Cepstral Coefficients (MFCC) is commonly used in the context of speech recognition and speaker verification especially under noisy conditions since they are noise robust and invariant both to additive and convolutional noise [34][35][36]. The idea of MFCC is to capture the most salient features based on a nonlinear frequency scale derived from the human peripheral auditory system, which would reflect how our hearing perceives the VA signal in terms of spectral power density in Mel frequency bands.…”

Section: Feature Extractionmentioning

confidence: 99%

Clustering Methods for Vibro-Acoustic Sensing Features as a Potential Approach to Tissue Characterisation in Robot-Assisted Interventions

Urrutia,

Espejo,

Evens

et al. 2023

Sensors

View full text Add to dashboard Cite

This article provides a comprehensive analysis of the feature extraction methods applied to vibro-acoustic signals (VA signals) in the context of robot-assisted interventions. The primary objective is to extract valuable information from these signals to understand tissue behaviour better and build upon prior research. This study is divided into three key stages: feature extraction using the Cepstrum Transform (CT), Mel-Frequency Cepstral Coefficients (MFCCs), and Fast Chirplet Transform (FCT); dimensionality reduction employing techniques such as Principal Component Analysis (PCA), t-Distributed Stochastic Neighbour Embedding (t-SNE), and Uniform Manifold Approximation and Projection (UMAP); and, finally, classification using a nearest neighbours classifier. The results demonstrate that using feature extraction techniques, especially the combination of CT and MFCC with dimensionality reduction algorithms, yields highly efficient outcomes. The classification metrics (Accuracy, Recall, and F1-score) approach 99%, and the clustering metric is 0.61. The performance of the CT–UMAP combination stands out in the evaluation metrics.

show abstract

Section: Feature Extractionmentioning

confidence: 99%

Clustering Methods for Vibro-Acoustic Sensing Features as a Potential Approach to Tissue Characterisation in Robot-Assisted Interventions

Urrutia,

Espejo,

Evens

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…The accuracy evaluation metric was used to measure the proportion of correctly classified instances in the datasets. It shows how frequently the classifier predicts the correct values and gives the percentage of the samples which were correctly classified from all the samples given [39]. It was calculated as the number of correct predictions divided by the total number of predictions.…”

Section: Accuracymentioning

confidence: 99%

Evaluation of Machine Learning Algorithms using Combined Feature Extraction Techniques for Speaker Identification

Iwok,

Udofia,

Obot

et al. 2023

JERR

View full text Add to dashboard Cite

The aim of this study is to evaluate and compare machine learning algorithms when various feature extraction techniques are employed together and determine the optimal feature combinations for the models studied. The TIMIT online database was used where 5 male and 5 female non-native English speakers from five American locations were selected. Each speaker had ten 3-second utterances, totaling 500. Mel frequency cepstral coefficients (MFCC), linear predictive cepstral coefficients (LPCC), gammatone frequency cepstral coefficients (GFCC), discrete wavelet transforms (DWT) and pitch features were extracted using MATLAB and concatenated. The concatenated features were used to train and evaluate three classifier models—Random Forest (RF), Linear Discriminant Analysis (LDA), and Logistic Regression (LR)—using Python software. The results obtained showed that as the number of features combinations increased, the models’ performances improved as well. This improved performance was observed when all the cepstral features were part of the combinations. This implies that cepstral features are more robust and improve speaker identification systems. The best average score of accurate predictions of ≈ 76% for the LR model was obtained for the MGL (39 features) features combination and dropped to 70% for the highest number of feature combinations MGDLP (53 features). This indicates that more training data improves system performance, however, too much data does not translate to even better performance because the system will eventually achieve its peak performance. This information is useful for applications where limited data can present a problem.

show abstract

Framework for Biometric User Authentication Based on a Dynamic Handwritten Signature

Anikin

Anisimova

2022

Cyber-Physical Systems: Intelligent Models and Algorithms

View full text Add to dashboard Cite

Development of a regional voice dataset and speaker classification based on machine learning

Cited by 10 publications

References 38 publications

Clustering Methods for Vibro-Acoustic Sensing Features as a Potential Approach to Tissue Characterisation in Robot-Assisted Interventions

Clustering Methods for Vibro-Acoustic Sensing Features as a Potential Approach to Tissue Characterisation in Robot-Assisted Interventions

Evaluation of Machine Learning Algorithms using Combined Feature Extraction Techniques for Speaker Identification

Framework for Biometric User Authentication Based on a Dynamic Handwritten Signature

Contact Info

Product

Resources

About