Parameterization of Sequence of MFCCs for DNN-based voice disorder detection

Grzywalski, Tomasz; Maciaszek, Adam; Biniakowski, Adam; Orwat, Jan; Drgas, Szymon; Piecuch, M.; Belluzzo, Riccardo; Joachimiak, Krzysztof; Niemiec, Dawid; Ptaszynski, Jakub; Szarzyński, Krzysztof

doi:10.1109/bigdata.2018.8622012

Cited by 17 publications

(10 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Currently, any research team can use it for their study purposes. Moreover, our method is ready to be released to the market and could be used, e.g., with StethoMe [41] device for remote diagnosis and educational purposes.…”

Section: Discussionmentioning

confidence: 99%

Early and Remote Detection of Possible Heartbeat Problems With Convolutional Neural Networks and Multipart Interactive Training

Wołk

2019

IEEE Access

View full text Add to dashboard Cite

In this study, the convolutional neural network (CNN) and multipart interactive training were used to create a state-of-the-art classifier for the early detection of cardiac pathologies. The classification was performed using three sets of data samples; the first was recorded using a digital stethoscope and the others were recorded with mobile smart devices. These data were part of the competition on the Kaggle platform and used on the CNN in the form of audio samples as well as in spectrogram format, and experiments were conducted using both methods. Moreover, this method was used not only to distinguish healthy from unhealthy heart rhythm but also to attempt reaching some initial diagnosis. It was possible to identify a set of problems with the dataset, make corrections, and share them with the scientific community. The experiments were conducted using the so-called multi-part interactive training and an additional ResNet pre-trained network, resulting in more than 93% precision. This allows anyone to undertake prophylactical diagnosis using a smartphone alone. It also has a great educational potential for young doctors and students. Greater improvement is possible by supplying more data to the described method.

show abstract

Section: Discussionmentioning

confidence: 99%

Early and Remote Detection of Possible Heartbeat Problems With Convolutional Neural Networks and Multipart Interactive Training

Wołk

2019

IEEE Access

View full text Add to dashboard Cite

show abstract

“…where k is the number of frames and n is the number of samples by which the window is shifted in order to yield the ith frame then taking the DFT of the resulting signal [21].…”

Section: ) Framing and Windowingmentioning

confidence: 99%

“…These energies are also known as the Mel spectrum and can be used for calculating the first 13 coefficients using DCT. A popular formula to convert f in hertz into ݂ is given in (2) [20,21]:…”

Section: ) Mel-spectrummentioning

confidence: 99%

Classification of Chest X-Ray Images using Wavelet and MFCC Features and Support Vector Machine Classifier

Owida

Al-Ghraibah

Altayeb

2021

Eng. Technol. Appl. Sci. Res.

View full text Add to dashboard Cite

The shortage and availability limitation of RT-PCR test kits and is a major concern regarding the COVID-19 pandemic. The authorities' intention is to establish steps to control the propagation of the pandemic. However, COVID-19 is radiologically diagnosable using x-ray lung images. Deep learning methods have achieved cutting-edge performance in medical diagnosis software assistance. In this work, a new diagnostic method for detecting COVID-19 disease is implemented using advanced deep learning. Effective features were extracted using wavelet analysis and Mel Frequency Cepstral Coefficients (MFCC) method, and they used in the classification process using the Support Vector Machine (SVM) classifier. A total of 2400 X-ray images, 1200 of them classified as Normal (healthy) and 1200 as COVID-19, have been derived from a combination of public data sets to verify the validity of the proposed model. The experimental results obtained an overall accuracy of 98.8% by using five wavelet features, where the classification using MFCC features, MFCC-delta, and MFCC-delta-delta features reached accuracy around 97% on average. The results show that the proposed model has reached the required level of success to be applicable in COVID 19 diagnosis.

show abstract

“…The recent works applied unsupervised domain adaptation to address the hardware variation [19] . Various acoustic features, including cepstral features [20] , [21] , vocal jitter [22] , and entropy [23] were also investigated in the literature. The IEEE Big Data conference held an international competition in Seattle 2018, called FEMH-Challenge, in which voice pathology detection systems from different research groups worldwide are evaluated empirically on the same dataset, which was published by Far Eastern Memorial Hospital (FEMH), Taiwan [24] .…”

Section: Introductionmentioning

confidence: 99%

Continuous Speech for Improved Learning Pathological Voice Disorders

Wang

Lai

et al. 2022

IEEE Open J. Eng. Med. Biol.

View full text Add to dashboard Cite

Numerous studies had successfully differentiated normal and abnormal voice samples. Nevertheless, further classification had rarely been attempted. This study proposes a novel approach, using continuous Mandarin speech instead of a single vowel, to classify four common voice disorders (i.e. functional dysphonia, neoplasm, phonotrauma, and vocal palsy). Methods: In the proposed framework, acoustic signals are transformed into mel-frequency cepstral coefficients, and a bi-directional long-short term memory network (BiLSTM) is adopted to model the sequential features. The experiments were conducted on a large-scale database, wherein 1,045 continuous speech were collected by the speech clinic of a hospital from 2012 to 2019. Results: Experimental results demonstrated that the proposed framework yields significant accuracy and unweighted average recall improvements of 78.12-89.27% and 50.92-80.68%, respectively, compared with systems that use a single vowel. Conclusions: The results are consistent with other machine learning algorithms, including gated recurrent units, random forest, deep neural networks, and LSTM. The sensitivities for each disorder were also analyzed, and the model capabilities were visualized via principal component analysis. An alternative experiment based on a balanced dataset again confirms the advantages of using continuous speech for learning voice disorders.

show abstract

Parameterization of Sequence of MFCCs for DNN-based voice disorder detection

Cited by 17 publications

References 10 publications

Early and Remote Detection of Possible Heartbeat Problems With Convolutional Neural Networks and Multipart Interactive Training

Early and Remote Detection of Possible Heartbeat Problems With Convolutional Neural Networks and Multipart Interactive Training

Classification of Chest X-Ray Images using Wavelet and MFCC Features and Support Vector Machine Classifier

Continuous Speech for Improved Learning Pathological Voice Disorders

Contact Info

Product

Resources

About