Mel-frequency Cepstral Coefficients Coefficients Research Articles

MFCC feature extraction for COPD detection

Mel Frequency Cepstral Coefficients (MFCC) are acoustic features, which are based on human auditory perception, focus on spectral properties and capture relevant features of the audio signal related to vocal tract shape and energy in different frequency bands. The methodology for feature extraction involves several stages. First, the audio signals go through a preprocessing stage in which they are initially normalized, then background noise is reduced, and finally, they pass through a median filter. In the next block of MFCC feature extraction, the first task was to split the audio signal into small frames. Then, a hamming window is applied to smooth the edges of each frame and the short time Fourier transform of each frame is calculated. Next, Mel filters are applied, which adjusts the representation of the frequency spectrum to human auditory perception. Finally, cepstral coefficients are calculated from the frequency spectrum. The MFCC coefficients are then used as input features for classifiers and machine learning algorithms.

Towards interpretable speech biomarkers: exploring MFCCs

While speech biomarkers of disease have attracted increased interest in recent years, a challenge is that features derived from signal processing or machine learning approaches may lack clinical interpretability. As an example, Mel frequency cepstral coefficients (MFCCs) have been identified in several studies as a useful marker of disease, but are regarded as uninterpretable. Here we explore correlations between MFCC coefficients and more interpretable speech biomarkers. In particular we quantify the MFCC2 endpoint, which can be interpreted as a weighted ratio of low- to high-frequency energy, a concept which has been previously linked to disease-induced voice changes. By exploring MFCC2 in several datasets, we show how its sensitivity to disease can be increased by adjusting computation parameters.

Classification of normal and abnormal heart by classifying PCG signal using MFCC coefficients and CGP-ANN classifier

Globally, A leading cause of death is heart disease and a serious public health concern. The anomalies in heart sound appears before the heart disease symptoms. The sounds are type of auscultation, which is a process dealing with sounds in a body that generates due to mechanical vibrations of organs, Auscultation is a potential method in medical science to detect abnormalities in heart sounds and in case of suspicion The patient follows up with a referral for other evaluations, such as an electrocardiogram. In medical sciences early detection of symptoms are of major importance, this research work is a good step toward the detection of abnormalities in heart before symptom appearing by processing the phonocardiogram (PCG) signal. In this paper PCG signals are classified by utilizing the features of Mel frequency cepstral coefficients (MFCC) through Cartesian Genetic Programming - Artificial Network (CGP-ANN) Classifier. The diagnostic accuracy of proposed methodology is found 99.50% which is more than other classifiers like Support Vector Machine (SVM) and Convolutional Neural Network (CNN). The accuracy of model as compared to other models can prove the performance superiority of the proposed system.

A Novel Driving Noise Analysis Method for On-Road Traffic Detection.

Effective noise reduction and abnormal feature extraction are important for abnormal sound detection occurring in urban traffic operations. However, to improve the detection accuracy of continuous traffic flow and even overlapping vehicle bodies, effective methods capable to achieve accurate signal-to-noise ratio and appropriate characteristic parameters should be explored. In view of the disadvantages of traditional traffic detection methods, such as Short-Time Energy (STE) and Mel Frequency Cepstral Coefficients (MFCC), this study adopts an improved spectral subtraction method to analyze traffic noise. Through the feature fusion of STE and MFCC coefficients, an innovative feature parameter, E-MFCC, is obtained, assisting to propose a traffic noise detection solution based on Triangular Wave Analysis (TWA). APP Designer in MATLAB was used to establish a traffic detection simulation platform. The experimental results showed that compared with the accuracies of traffic detection using the traditional STE and MFCC methods as 67.77% and 76.01%, respectively, the detection accuracy of the proposed TWA is significantly improved, attaining 91%. The results demonstrated the effectiveness of the traffic detection method proposed in solving the overlapping problem, thus achieving accurate detection of road traffic volume and improving the efficiency of road operation.

Recognition Using DNN with Bacterial Foraging Optimization Using MFCC Coefficients

Speech is an important mode of communication for people. For a long time, researchers have been working hard to develop conversational machines which will communicate with speech technology. Voice recognition is a part of a science called signal processing. Speech recognition is becoming more successful for providing user authentication. The process of user recognition is becoming more popular now a days for providing security by authenticating the users. With the rising importance of automated information processing and telecommunications, the usefulness of recognizing an individual from the features of user voice is increasing. In this paper, the three stages of speech recognition processing are defined as pre-processing, feature extraction and decoding. Speech comprehension has been significantly enhanced by using foreign languages. Automatic Speech Recognition (ASR) aims to translate text to speech. Speaker recognition is the method of recognizing an individual through his/her voice signals. The new speaker initially privileges identity for speaker authentication, and then the stated model is used for identification. The identity argument is approved when the match is above a predefined threshold. The speech used for these tasks may be either text-dependent or text-independent. The article uses Bacterial Foraging Optimization Algorithm (BFO) for accurate speech recognition through Mel Frequency Cepstral Coefficients (MFCC) model using DNN. Speech recognition efficiency is compared to that of the conventional system.

Improving short utterance speaker verification by combining MFCC and Entrocy in Noisy conditions

Short utterance and background noise represent great challenging for speaker verification due to the mismatch and limited training and/or retrieve data. A remarkable performance using matched training and testing conditions generally could be achieved in automatic speaker verification. However, mismatched noisy and short utterances conditions attend to drop the results significantly. Furthermore, the performance is significantly affected by the features extraction. The most common features in this field of the study are Mel-Frequency Cepstral Coefficients (MFCCs). With a noise presents in the background and short utterances, MFCC performance could not be reliable without a support feature. To address this, a new feature ‘Entrocy’ for accurate and robust speaker verification under limited data and noisy environments is proposed and employed to support MFCC coefficients. Entrocy feature represents the Fourier Transform of the Entropy that calculates the fluctuation of the information in the sound segments over time. The resulting Entrocy features are combined with MFCC functionality to generate a composite feature, which is tested using the Gaussian Mixture Model (GMM) recognition method. The suggested method was conducted out over a range of signal/noise ratios and utterances were truncating into shorts (2, 3, 4, 5, 6, 8, and 10s) for verification. The proposed method has shown strong robustness in the challenging of background noise and limited testing data and they consistently perform better than the well-known MFCC.

Feature Extraction Techniques with Analysis of Confusing Words for Speech Recognition in the Hindi Language

The research work presents experimental work to build a speaker-independent connected word Hindi speech recognition system using different feature extraction techniques with comparative analysis of confusing words. Comparative analysis of confusing words is essential to understand the reason for the speech recognition errors. Based on the error analysis, different feature extraction techniques, classification techniques, acoustic models, and pronunciation dictionaries can be selected to improve the speech recognition system's performance. Earlier studies for Hindi speech recognition lack detailed comparative analysis of confusing words for different feature extractions methods. As speaker-independent systems are developed for all, comparative analysis of confusing words is also presented for all feature extraction techniques. Speaker independent system was proposed with five states monophone based hidden Markov model (HMM) using HMM-based tool kit HTK. A Self-created data set of Hindi speech corpus has been used in the experiment. Feature extraction techniques such as linear predictive coding cepstral coefficients (LPCCs), mel frequency cepstral coefficients (MFCCs), and perceptual linear prediction coefficients (PLPs) were applied using delta, double delta, and energy parameters to evaluate the performance of the proposed methodology. The system was assessed by using different feature extraction techniques for speaker-independent mode. Research findings reveal that PLP coefficients show the highest recognition score, while LPCCs got the lowest recognition scores.Investigations also reveal that both PLP and MFCC coefficients are better than LPCC in speech recognition. Comparative analysis of confusing words shows that PLPs and MFCCs show fewer confusions than LPCCs and exhibit mostly the same pattern in the confusion analysis. Research outcomes also reveal that substitution errors are a significant cause of low recognition. It was also found that some words were recognized with individual feature extraction techniques only. Confusion analysis of the words indicates that words which have nasals, liquid, and fricative sound in first place exhibit more confusions. The investigation could improve speech recognition by choosing an appropriate feature extraction method and mixing the various feature extraction methods. The research outcomes can also be utilized to build linguistic resources for improving speech recognition. The results show that the developed recognition framework achieved the highest recognition word accuracy of 76.68% with PLPs for the speaker-independent model. The proposed system was also compared with existing similar work available.

Evaluation of Perceptual and Multi Sub-Band Energy Features for Classification of Normal, Pathological and Noisy Phonocardiogram

Murmurs in the heart sound signal convey crucial diagnostic information on valvular defects. It is a significant challenge to identify murmurs in heart sound signals due to its narrow bandwidth and the short time duration. In the clinical examination, these sounds are categorized based on human auditory perception. In this paper, we explore the popular perceptual features mel frequency cepstral coefficients (MFCC) and rhythm pattern (RP), and various multiple sub-band energy (MSBE) features for evaluation and quantification of murmurs and noises. After a comprehensive analysis of MFCC at a different number of coefficient values, it is observed that four MFCC coefficients are optimal and more discriminative than other conventional features. This is further compared with the proposed MSBE feature. The proposed feature performs better achieving an average M Acc score of 68.27%, 79.83% and 94.71% for respective classes. M Acc is the mean value of sensitivity and specificity. The estimation of sub-band energy values from log magnitude spectrogram and across larger frequency intervals improve the quantification of low amplitude high-frequency components introduced into the heart sound signals.

Discriminating Parkinson diseased and healthy people using modified MFCC filter bank approach

In this paper a modified Mel scaled filter bank-based approach to discriminate people suffering from Parkinson disease (PD) in their early stages from healthy people using speech samples is proposed. Parkinson’s disease not only affects the muscular activities of the human body but also affects the speech of the diseased. So, the speech features of Parkinson affected people tend to vary and hence differ from the speech features of healthy people. In this paper, the speech feature used for discriminating the two groups is the Mel frequency cepstral coefficients (MFCC) extracted from speech samples of both the PD and healthy people. The traditional way of computing the MFCC coefficients involves the design of the Mel filter bank. These filters are usually designed according to the auditory or acoustic system of human ear which follows the Mel scale. In this study, modification to this Mel scaled bank of filters is done by varying its bandwidth in the region of interest to compute the feature, MFCC and its performance is then compared with the conventionally designed MFCC filter bank for the said application. The performance is compared in terms of classification accuracy using radial basis network classifier. The results show an improvement of 6.3% in the classification accuracy obtained using the proposed method.

Evaluation of Mel-Frequency Cepstrum for Wheeze Analysis.

Monitoring of wheezes is an integral part of managing Chronic Respiratory Diseases such as asthma and Chronic Obstructive Pulmonary Disease (COPD). Recently, there is a growing interest in automatic detection of wheezes and the use of Mel-Frequency Cepstral Coefficients (MFCC) have been shown to achieve encouraging detection performance. While the successful use of MFCC for identifying wheezes has been demonstrated, it is not clear which MFCC coefficients are actually useful for detecting wheezes. The objective of this paper is to characterize and study the effectiveness of individual coefficients in discriminating between wheezes and normal respiratory sounds. The coefficients have been evaluated in terms of histogram dissimilarity and linear separability. Further, a comparison between the use of single coefficient against other combinations of coefficients is also presented. The results demonstrate MFCC-2 coefficient to be significantly more effective than all the other coefficients in discriminating between wheezes and normal respiratory sounds sampled at 8000 Hz.

Diagnosis of Parkinson’s Disease based on Wavelet Transform and Mel Frequency Cepstral Coefficients

The aim of this study presented in this paper is to determine the choice of the appropriate wavelet analyzer with the method of extraction of MFCC coefficients for an assistance in the diagnosis of Parkinson's disease. The analysis used is based on a database of 18 healthy and 20 Parkinsonian patients. The suggested processing is based on the transformation of the speech signal by the wavelet transform through testing several sorts of wavelets, extracting Mel Frequency Cepstral Coefficients (MFCC) from the signals, and we apply the support vector machine (SVM) as classifier. The test results reveal that the best recognition rate, which is 86.84%, is obtained by the wavelets of level 2 at 3rd scale (Daubechie, Symlet, ReverseBior or BiorSpline) combination-MFCC–SVM.

Significance of incorporating excitation source parameters for improved emotion recognition from speech and electroglottographic signals

The work presented in this paper explores the effectiveness of incorporating the excitation source parameters such as strength of excitation and instantaneous fundamental frequency ( $$F_0$$ ) for emotion recognition task from speech and electroglottographic (EGG) signals. The strength of excitation (SoE) is an important parameter indicating the pressure with which glottis closes at the glottal closure instants (GCIs). The SoE is computed by the popular zero frequency filtering (ZFF) method which accurately estimates the glottal signal characteristics by attenuating or removing the high frequency vocaltract interactions in speech. The arbitrary impulse sequence, obtained from the estimated GCIs, is used to derive the instantaneous $$F_0$$ . The SoE and the instantaneous $$F_0$$ parameters are combined with the conventional mel frequency cepstral coefficients (MFCC) to improve the recognition rates of distinct emotions (Anger, Happy and Sad) using Gaussian mixture models as classifier. The performances of the proposed combination of SoE and instantaneous $$F_0$$ and their dynamic features with MFCC coefficients are compared with the emotion utterances (4 emotions and neutral) from classical German full blown emotion speech database (EmoDb) having simultaneous speech and EGG signals and Surrey Audio Visual Expressed Emotion database (3 emotions and neutral) for both speaker dependent and speaker independent emotion recognition scenarios. To reinforce the effectiveness of the proposed features and for better statistical consistency of the emotion analysis, a fairly large emotion speech database of 220 utterances per emotion in Tamil language with simultaneous EGG recordings, is used in addition to EmoDb. The effectiveness of SoE and instantaneous $$F_0$$ in characterizing different emotions is also confirmed by the improved emotion recognition performance in Tamil speech-EGG emotion database.

Feature extraction from smartphone inertial signals for human activity segmentation

This paper proposes the adaptation of well-known strategies successfully used in speech processing: Mel Frequency Cepstral Coefficients (MFCCs) and Perceptual Linear Prediction (PLP) coefficients. Additionally characteristics like RASTA filtering or delta coefficients are also considered and evaluated for inertial signal processing. These adaptations have been incorporated into a Human Activity Recognition and Segmentation (HARS) system based on Hidden Markov Models (HMMs) for recognizing and segmenting six different physical activities: walking, walking–upstairs, walking-downstairs, sitting, standing and lying.All experiments have been done using a publicly available dataset named UCI Human Activity Recognition Using Smartphones, which includes several sessions with physical activity sequences from 30 volunteers. This dataset has been randomly divided into six subsets for performing a six-fold cross validation procedure. For every experiment, average values from the six-fold cross-validation procedure are shown.The results presented in this paper overcome significantly baseline error rates, constituting a relevant contribution in the field. Adapted MFCC and PLP coefficients improve human activity recognition and segmentation accuracies while reducing feature vector size considerably. RASTA-filtering and delta coefficients contribute significantly to reduce the segmentation error rate obtaining the best results: an Activity Segmentation Error Rate lower than 0.5%.

Singer identification using perceptual features and cepstral coefficients of an audio signal from Indian video songs

Singer identification is a difficult topic in music information retrieval because background instrumental music is included with singing voice which reduces performance of a system. One of the main disadvantages of the existing system is vocals and instrumental are separated manually and only vocals are used to build training model. The research presented in this paper automatically recognize a singer without separating instrumental and singing sounds using audio features like timbre coefficients, pitch class, mel frequency cepstral coefficients (MFCC), linear predictive coding (LPC) coefficients, and loudness of an audio signal from Indian video songs (IVS). Initially, various IVS of distinct playback singers (PS) are collected. After that, 53 audio features (12 dimensional timbre audio feature vectors, 12 pitch classes, 13 MFCC coefficients, 13 LPC coefficients, and 3 loudness feature vector of an audio signal) are extracted from each segment. Dimension of extracted audio features is reduced using principal component analysis (PCA) method. Playback singer model (PSM) is trained using multiclass classification algorithms like back propagation, AdaBoost.M2, k-nearest neighbor (KNN) algorithm, naive Bayes classifier (NBC), and Gaussian mixture model (GMM). The proposed approach is tested on various combinations of dataset and different combinations of audio feature vectors with various Indian male and female PS’s songs.

Emotion Detection Using MFCC and Cepstrum Features

Abstract A tremendous research is being done on Speech Emotion Recognition (SER) in the recent years with its main motto to improve human machine interaction. In this work, the effect of cepstral coefficients in the detection of emotions is performed. Also, a comparative analysis of cepstum, Mel-frequency Cepstral Coefficients (MFCC) and synthetically enlarged MFCC coefficients on emotion classification is done. Using a compact feature vector, our algorithm depicted better recognition rates of identifying seven emotions from Berlin speech corpus compared to the earlier work by Firoz Shah where only four emotions were recognized with good accuracy. The proposed method has facilitated a considerable reduction in the misclassification efficiency which outperforms the algorithm by InmaMohino, where the feature vector included only synthetically enlarged MFCC coefficients.

Classification of fricative consonants for speech enhancement in hearing devices.

ObjectiveTo investigate a set of acoustic features and classification methods for the classification of three groups of fricative consonants differing in place of articulation.MethodA support vector machine (SVM) algorithm was used to classify the fricatives extracted from the TIMIT database in quiet and also in speech babble noise at various signal-to-noise ratios (SNRs). Spectral features including four spectral moments, peak, slope, Mel-frequency cepstral coefficients (MFCC), Gammatone filters outputs, and magnitudes of fast Fourier Transform (FFT) spectrum were used for the classification. The analysis frame was restricted to only 8 msec. In addition, commonly-used linear and nonlinear principal component analysis dimensionality reduction techniques that project a high-dimensional feature vector onto a lower dimensional space were examined.ResultsWith 13 MFCC coefficients, 14 or 24 Gammatone filter outputs, classification performance was greater than or equal to 85% in quiet and at +10 dB SNR. Using 14 Gammatone filter outputs above 1 kHz, classification accuracy remained high (greater than 80%) for a wide range of SNRs from +20 to +5 dB SNR.ConclusionsHigh levels of classification accuracy for fricative consonants in quiet and in noise could be achieved using only spectral features extracted from a short time window. Results of this work have a direct impact on the development of speech enhancement algorithms for hearing devices.

Pattern recognition methods applied to respiratory sounds classification into normal and wheeze classes

In this paper, we present the pattern recognition methods proposed to classify respiratory sounds into normal and wheeze classes. We evaluate and compare the feature extraction techniques based on Fourier transform, linear predictive coding, wavelet transform and Mel-frequency cepstral coefficients (MFCC) in combination with the classification methods based on vector quantization, Gaussian mixture models (GMM) and artificial neural networks, using receiver operating characteristic curves. We propose the use of an optimized threshold to discriminate the wheezing class from the normal one. Also, post-processing filter is employed to considerably improve the classification accuracy. Experimental results show that our approach based on MFCC coefficients combined to GMM is well adapted to classify respiratory sounds in normal and wheeze classes. McNemar's test demonstrated significant difference between results obtained by the presented classifiers ( p < 0.05 ) .

Research and realization of speech segmentation in MP3 compressed domain

This article proposed an approach for detecting the voice change of speakers by employing improved Bayesian Information Criterion(BIC) algorithm in MPEG1-layer3(MP3) compressed domain.According to the process of MFCC calculation in raw audio,a new Mel-Frequency Cepstral Coefficients(MFCC) algorithm by utilizing Modified Discrete Cosine Transform(MDCT) coefficient in MP3 domain was presented.Based on these coefficients,the improved BIC algorithm was employed to decide which point was the voice change point of speakers.The experimental results show that using the MFCC coefficients extracted in MP3 domain for speech segmentation,similar segmentation precision can be obtained as that in uncompressed domain.