Mel-frequency Cepstral Coefficients Feature Extraction Research Articles

Exploring the Impact of Mismatch Conditions, Noisy Backgrounds, and Speaker Health on Convolutional Autoencoder-Based Speaker Recognition System with Limited Dataset

This paper presents a novel approach to enhance the success rate and accuracy of speaker recognition and identification systems. The methodology involves employing data augmentation techniques to enrich a small dataset with audio recordings from five speakers, covering both male and female voices. Python programming language is utilized for data processing, and a convolutional autoencoder is chosen as the model. Spectrograms are used to convert speech signals into images, serving as input for training the autoencoder. The developed speaker recognition system is compared against traditional systems relying on the MFCC feature extraction technique. In addition to addressing the challenges of a small dataset, the paper explores the impact of a "mismatch condition" by using different time durations of the audio signal during both training and testing phases. Through experiments involving various activation and loss functions, the optimal pair for the small dataset is identified, resulting in a high success rate of 92.4% in matched conditions. Traditionally, Mel-Frequency Cepstral Coefficients (MFCC) have been widely used for this purpose. However, the COVID-19 pandemic has drawn attention to the virus's impact on the human body, particularly on areas relevant to speech, such as the chest, throat, vocal cords, and related regions. COVID-19 symptoms, such as coughing, breathing difficulties, and throat swelling, raise questions about the influence of the virus on MFCC, pitch, jitter, and shimmer features. Therefore, this research aims to investigate and understand the potential effects of COVID-19 on these crucial features, contributing valuable insights to the development of robust speaker recognition systems.

Comparative study of respiratory sounds classification methods based on cepstral analysis and artificial neural networks

In this paper, we investigated and evaluated various machine learning-based approaches for automatically detecting wheezing sounds. We conducted a comprehensive comparison of these proposed systems, assessing their classification performance through metrics such as Sensitivity, Specificity, and Accuracy. The main approach to developing a machine learning-based system for classifying respiratory sounds involved the combination of a technique for extracting features from an unknown input sound with a classification method to determine its belonging class. The characterization techniques used in this study are based on the cepstral analysis, which was extensively employed in the automatic speech recognition field. While MFCC (Mel-Frequency Cepstral Coefficients) feature extraction methods are commonly used in respiratory sounds classification, our study introduces a novelty by employing GFCC (Gammatone-Frequency Cepstral Coefficients) and BFCC (Bark-Frequency Cepstral Coefficients) for this purpose. For the classification task, we employed two types of neural networks: the MLP (Multilayer Perceptron), a feedforward neural network, and a variant of the LSTM (Long Short-Term Memory) recurrent neural network called BiLSTM (Bidirectional LSTM). The proposed classification systems are evaluated using a database consisting of 497 wheezing segments and 915 normal respiratory segments, which are recorded from individuals diagnosticated with asthma and individuals without any respiratory issues, respectively. The highest classification performance was achieved by the BFCC-BiLSTM model, which demonstrated an exceptional accuracy rate of 99.8%.

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.

Wavelet-based Parametric Feature Subset Selection for Speaker and Accent Recognition using Genetic Algorithm

Research on speaker and accent recognition studies using the Malay language in the field of Automatic Speech Recognition (ASR) is limited, with most studies focusing on speech recognition. This study proposes to increase the performance Malaysian speakers and accent recognition using wavelets transform, namely Wavelet Packet Transform (WPT) and Dual-Tree Complex Wavelet Packet Transform (DT-CWPT). A variety of feature extraction combinations, including conventional Mel Frequency Cepstral Coefficients (MFCC), Linear Predictive Coding (LPC) and wavelets transform, were implemented to compare the effectiveness of the proposed method. Although the proposed approach resulted in improved detection rate, it faced challenges in terms of high feature dimensionality and increased computation time. To address these issues, the Genetic Algorithm (GA) approach has been adopted to reduce the number of irrelevant features, accelerate the learning system and achieve better performance. The extracted features were trained using various classifiers, including k-Nearest Neighbors (k-NN), Support Vector Machine (SVM) and Extreme Learning Machine (ELM). The experimental results showed that the best speaker recognition accuracy was 97.33% for English numbers using SVM classifier and 96.02% for Malay words using the ELM classifier with a combination of wavelets, LPC and MFCC features. For accent recognition, the ELM classifier yielded the best performance, achieving 95.28% accuracy for English numbers with a combination of wavelets and MFCC features and 96.72% for Malay words using combined feature extraction of wavelets, LPC and MFCC feature extraction. It can be concluded that Malay words yielded better recognition rates than English numbers. Furthermore, use of GA effectively reduced the overall number of features while maintaining high accuracy level.

A Perspective Study on Speech Recognition

Emotions play an extremely important role in human mental life. It is a medium of expression of one’s perspective or one’s mental state to others. Speech Emotion Recognition (SER) can be defined as extraction of the emotional state of the speaker from his or her speech signal. There are few universal emotions including Neutral, Anger, Happiness, and Sadness in which any intelligent system with finite computational resources can be trained to identify or synthesize as required. In this work spectral and prosodic features are used for speech emotion recognition because both of these features contain the emotional information. Mel-Frequency Cepstral Coefficients (MFCC) is one of the spectral features. Fundamental frequency, loudness, pitch and speech intensity and glottal parameters are the prosodic features which are used to model different emotions. The potential features are extracted from each utterance for the computational mapping between emotions and speech patterns. Pitch can be detected from the selected features, using which gender can be classified. The audio signal is filtered using a method known as feature extraction technique. In this article, the feature extraction technique for speech recognition and voice classification is analyzed and also centered to comparative analysis of different types of Mel frequency cepstral coefficients (MFCC) feature extraction method. The MFCC technique is used for deduction of noise in voice signals and also used for voice classification and speaker identification. The statistical results of the different MFCC techniques are discussed and finally concluded that the delta-delta MFCC feature extraction technique is better than the other feature extraction techniques..

Measurements, Analysis, Classification, and Detection of Gunshot and Gunshot-like Sounds

Gun violence has been on the rise in recent years. To help curb the downward spiral of this negative influence in communities, machine learning strategies on gunshot detection can be developed and deployed. After outlining the procedure by which a typical type of gunshot-like sounds were measured, this paper focuses on the analysis of feature importance pertaining to gunshot and gunshot-like sounds. The random forest mean decrease in impurity and the SHapley Additive exPlanations feature importance analysis were employed for this task. From the feature importance analysis, feature reduction was then carried out. Via the Mel-frequency cepstral coefficients feature extraction process on 1-sec audio clips, these extracted features were then reduced to a more manageable quantity using the above-mentioned feature reduction processes. These reduced features were sent to a random forest classifier. The SHapley Additive exPlanations feature importance output was compared to that of the mean decrease in impurity feature importance. The results show what Mel-frequency cepstral coefficients features are important in discriminating gunshot sounds and various gunshot-like sounds. Together with the feature importance/reduction processes, the recent uniform manifold approximation and projection method was used to compare the closeness of various gunshot-like sounds to gunshot sounds in the feature space. Finally, the approach presented in this paper provides people with a viable means to make gunshot sounds more discernible from other sounds.

Klasifikasi Suara Paru-Paru Berdasarkan Ciri MFCC

The lungs are an important organ in the human respiratory system, which functions to exchange carbon dioxide from the blood with oxygen in the air. Detection of respiratory disorders and lung disorders can be done in various ways; view medical records, physical examination, detection by x-ray and also auscultation of breathing. Digital signal processing can be used as a method to detect lung disorders based on the sound produced. In this study, lung sounds were classified into normal, crackle, wheeze, and crackle-wheeze classes using the Mel Frequency Cepstral Coefficient (MFCC) and Convolutional Neural Network (CNN) methods.Observations were made by varying the MFCC feature extraction using MFCC 8 and 13 coefficients, the number of frames are 50 and 60, and the width of the frames used was 0,1, 0,15 and 0,2 seconds. The result of feature extraction is then applied to the CNN classification system, and the confusion matrix is used to get the accuracy and precision values. The highest accuracy and precision values were obtained at 71,85% and 65,70% on the MFCC 13 coefficient with an average of 71,18%. Based on these results, the system that has been created can classify normal lung sounds, crackle, wheeze and crackle-wheeze quite well.

Frequency domain analysis of MFCC feature extraction in children’s speech recognition system

Abstract —The research on speech recognition systems currently focuses on the analysis of robust speech recognition systems. When the speech signals are combined with noise, the recognition system becomes distracted, struggling to identify the speech sounds. Therefore, the development of a robust speech recognition system continues to be carried out. The principle of a robust speech recognition system is to eliminate noise from the speech signals and restore the original information signals. In this paper, researchers conducted a frequency domain analysis on one stage of the Mel Frequency Cepstral Coefficients (MFCC) process, the Fast Fourier Transform (FFT), in children's speech recognition system. The FTT analysis in the feature extraction process determined the effect of frequency value characteristics utilized in the FFT output on the noise disruption. The analysis method was designed into three scenarios based on the value of the employed FFT points. The differences between scenarios were based on the number of shared FFT points. All FFT points were divided into four, three, and two parts in the first, second, and third scenarios, respectively. This study utilized children's speech data from the isolated TIDIGIT English digit corpus. As comparative data, the noise was added manually to simulate real-world conditions. The results showed that using a particular frequency portion following the scenario designed on MFCC affected the recognition system performance, which was relatively significant on the noisy speech data. The designed method in the scenario 3 (C1) version generated the highest accuracy, exceeded the accuracy of the conventional MFCC method. The average accuracy in the scenario 3 (C1) method increased by 1% more than all the tested noise types. Using various noise intensity values (SNR), the testing process indicates that scenario 3 (C1) generates a higher accuracy than conventional MFCC in all tested SNR values. It proves that the selection of specific frequency utilized in MFCC feature extraction significantly affects the recognition accuracy in a noisy speech.

The Implementation Of Mfcc Feature Extraction And Selection of Cepstral Coefficient for Qur’an Recitation in TPA (Qur’an Learning Center) Nurul Huda Plus Purbayan

There are two approaches to Qur’an recitation, namely talaqqi and qira'ati. Both approaches use the science of recitation containing knowledge of the rules and procedures for reading the Qur'an properly. Talaqqi requires the teacher and students to sit facing each other while qira'ati is the recitation of the Qur'an with rhythms and tones. Many studies have developed an automatic speech recognition system for Qur’an recitation to help the learning process. Feature extraction model using Mel Frequency Cepstral Coefficient (MFCC) and Linear Predictive Code (LPC). The MFCC method has an accuracy of 50% to 60% while the accuracy of Linear Predictive Code (LPC) is only 45% to 50%, so the non-linear MFCC method has higher accuracy than the linear approach method. The cepstral coefficient feature that is used starts from 0 to 23 or 24 cepstral coefficients. Meanwhile, the frame taken consists of 0 to 10 frames or eleven frames. Voting for 300 recorded voice samples was tested against 200 voice recordings, both male and female voices. The frequency used was 44.100 kHz stereo 16 bit. This study aims to obtain good accuracy by selecting the right feature on the cepstral coefficient using MFCC feature extraction and matching accuracy through the selection of the cepstral coefficient feature with Dominant Weight Normalization (NBD) at TPA Nurul Huda Plus Purbayan. Accuracy results showed that the MFCC method with the selection of the 23rd cepstral coefficient has a higher accuracy rate of 90.2% compared to the others. It can be concluded that the selection of the right features on the 23rd cepstral coefficient affects the accuracy of the voice of Qur’an recitation.

Comparison of Mel Frequency Cepstral Coefficient (MFCC) Feature Extraction, With and Without Framing Feature Selection, to Test the Shahada Recitation

Voice research for feature extraction using MFCC. Introduction with feature extraction as the first step to get features. Features need to be done further through feature selection. The feature selection in this research used the Dominant Weight feature for the Shahada voice, which produced frames and cepstral coefficients as the feature extraction. The cepstral coefficient was used from 0 to 23 or 24 cepstral coefficients. At the same time, the taken frame consisted of 0 to 10 frames or eleven frames. Voting as many as 300 samples of recorded voices were tested on 200 voices of both male and female voice recordings. The frequency used was 44.100 kHz 16-bit stereo. This research aimed to gain accuracy by selecting the right features on the frame using MFCC feature extraction and matching accuracy with frame feature selection using the Dominant Weight Normalization (NBD). The accuracy results obtained that the MFCC method with the selection of the 9th frame had a higher accuracy rate of 86% compared to other frames. The MFCC without feature selection had an average of 60%. The conclusion was that selecting the right features in the 9th frame impacted the accuracy of the voice of shahada recitation.

Non-Contact Monitoring and Classification of Breathing Pattern for the Supervision of People Infected by COVID-19.

During the pandemic of coronavirus disease-2019 (COVID-19), medical practitioners need non-contact devices to reduce the risk of spreading the virus. People with COVID-19 usually experience fever and have difficulty breathing. Unsupervised care to patients with respiratory problems will be the main reason for the rising death rate. Periodic linearly increasing frequency chirp, known as frequency-modulated continuous wave (FMCW), is one of the radar technologies with a low-power operation and high-resolution detection which can detect any tiny movement. In this study, we use FMCW to develop a non-contact medical device that monitors and classifies the breathing pattern in real time. Patients with a breathing disorder have an unusual breathing characteristic that cannot be represented using the breathing rate. Thus, we created an Xtreme Gradient Boosting (XGBoost) classification model and adopted Mel-frequency cepstral coefficient (MFCC) feature extraction to classify the breathing pattern behavior. XGBoost is an ensemble machine-learning technique with a fast execution time and good scalability for predictions. In this study, MFCC feature extraction assists machine learning in extracting the features of the breathing signal. Based on the results, the system obtained an acceptable accuracy. Thus, our proposed system could potentially be used to detect and monitor the presence of respiratory problems in patients with COVID-19, asthma, etc.

Isolated Command Recognition Using MFCC and Clustering Algorithm

People who cannot walk by themselves need someone to carry them around on a wheelchair. A voice recognition system was developed to recognize a set of commands used by people with disability to control their wheelchair and devices around them. Voice samples of various commands from 8 different speakers were collected. The features of the collected samples were extracted using MFCC feature extraction technique. The MFCC (Mel-frequency cepstral coefficient) feature extraction technique involves pre-emphasis, framing, windowing, FFT, Mel-scale transformation operations. The major reason for choosing MFCC extraction is that the human ear has a response of a logarithmic scale and not a linear scale. Hence, all the framed voice samples are transformed to the Mel-scale with a logarithmic response and stored in the form of K-means clusters. Features extracted from test data are applied to the models developed for commands, and based on the minimum distance criterion, the model is selected to be closely associated with the respective voice command. The voice recognition system was developed for an 8-command model using MATLAB.

Markov Model Based Real Time Speaker Recognition using K-Means, Fast Fourier Transform and Mel Frequency Cepstral Coefficients

In this study, which was carried out using a combination of machine learning and sound processing methods, a speaker recognition system and application were developed using real-time Mel Frequency Cepstral Coefficients (MFCC) features and Markov chain model classifier. A sound sample was taken from each speaker for the training of the system and these sound samples were processed in Fast Fourier Transform and MFCC feature extraction algorithms. The MFCC features were clustered using the k-means clustering algorithm. A Markov chain model was created for each speaker by using the outputs obtained after clustering. By deducting the characteristic features of the voice of the speaker, the person who was talking in the society and how long and at which time intervals they spoke during the conversation was determined in real time with high accuracy.

Using Auditory Features for WiFi Channel State Information Activity Recognition

Activity recognition has gained significant attention recently, due to the availability of smartphones and smartwatches with movement sensors which facilitate the collection and processing of relevant measurements, by almost everyone. Using the device-embedded sensors, there is no need of carrying dedicated equipment (inertia measurement units or accelerometers) and use complex software to process the data. This approach though, has the disadvantage of needing to carry a device during the monitoring time. WiFi channel state information (CSI) offers a passive, device-free opportunity, for monitoring activities of daily living in non-line-of-sight conditions. In this paper, Mel frequency cepstral coefficient (MFCC) feature extraction, used successfully for audio signals, is proposed for CSI time-series classification. The applicability of the proposed features in activity recognition has been evaluated using three classification methods, convolutional neural networks (CNN), long short term memory recurrent neural networks and Hidden Markov models, in comparison to currently used feature extraction methods (discrete wavelet transform, short time fourier transform). MFCC feature extraction achieves higher accuracy in activity classification than the compared methods, as verified by evaluation with two activity datasets, in particular 95% accuracy precision achieved in activity recognition using MFCC features in combination with CNN classifier.

Speaker Identification for OFDM-Based Aeronautical Communication System

Although a lot of research has been done on speaker identification in the presence of noise and channel variation, to the best of our knowledge, no work has been reported for aeronautical applications. In this paper, we aim to fulfill this goal by developing a Speaker Identification System (SIS) for future aeronautical communications systems. Furthermore, we present a novel feature extraction scheme based on multi-resolution analysis. The proposed features called SMFCC use Mel Frequency Cepstral Coefficients (MFCCs) features of stationary wavelet transform sub-bands. The extracted features are modeled using the i-vector approach, and support-vector machines are adopted as a back-end classifier. The performance of the proposed SIS is evaluated using two publicly available databases. Comparison of the proposed approach with the baseline MFCC feature extraction shows the feasibility and the robustness of the proposed method. Besides the noise reduction, the identification accuracy is improved by about 12% at higher signal-to-noise ratios and reaches 97.33% as compared to 88.33% using MFCC for ATCOSIM database.

Efficient FPGA-based architecture of an automatic wheeze detector using a combination of MFCC and SVM algorithms

This paper proposes a new hardware implementation of an automatic wheeze detector in pulmonary sounds. The proposed system is based on the combination of Mel-frequency cepstral coefficients (MFCC) feature extraction method and a support vector machine (SVM) classifier. It has been implemented on a field programmable gate array (FPGA) chip using the Xilinx System Generator (XSG) programming tool and the Nexys-4 development board, which is built around the low-cost Artix-7 FPGA device. The LIBSVM library has been used to extract the SVM parameters during the training phase in the Matlab environment, then the MFCC feature extraction and the SVM testing phase are performed on the FPGA chip. Two hardware architectures (default and optimized) of the SVM classifier are implemented and compared in terms of resource utilization, maximum operating frequency, and power consumption for the Artix-7 XC7A100T FPGA Chip. The detection rates obtained by the fixed-point XSG implementation are presented and compared to those obtained by the floating-point Matlab simulation. It has been also compared to existing methods in terms of the detection rates and the characteristics of its implementation.

GFCC based discriminatively trained noise robust continuous ASR system for Hindi language

A statistically designed Automatic Speech Recognition (ASR) system extracts features from speech signals using feature extraction methods, links the extracted features with the expected phonetics of the hypothesis using acoustic models, uses language model to add prior information about the structure of the target language. For many years, Mel-frequency Cepstral coefficients (MFCC), n-gram, and Hidden Markov Model (HMM) approaches have been used predominantly for feature extraction, language modeling and acoustic modeling, respectively. However, performance degradation of MFCC in noisy conditions and inaccuracy of HMMs while handling large vocabularies have made researchers to propose more efficient methods. The proposed work uses noise robust method Gammatone Frequency Cepstral Coefficients(GFCC) for feature extraction, trigram language modeling, and HMM-Gaussian mixture model (GMM) based acoustic modeling to implement a continuous Hindi language ASR system. Also, it applies Differential Evolution (DE) technique to refine the GFCC features and discriminative techniques to enhance performance of the acoustic model. The performance of the implemented system has been evaluated by using different feature extraction methods, variants of n-gram language modeling techniques and different discriminative techniques in clean as well as noisy conditions. Initially, the results reveal that DE optimized GFCC with HMM-Gaussian Mixture Model (GMM) acoustic modeling performs better than MFCC, PLP and MF-PLP feature extraction methods. Secondly, the experimental results show that the Minimum Phone Error (MPE) outperforms Maximum Mutual Information (MMI) and Maximum Likelihood Estimation (MLE) and trigram based language modeling gives more accurate results than unigram and bigram language modeling. Finally, it has been concluded that the continuous Hindi language ASR system implemented using DE refined GFFC feature extraction method with MPE discriminative training technique and trigram based language modeling gives better accuracy in clean as-well-as noisy environments.

FPGA-based simultaneous multichannel audio processor for musical genre indexing applications in broadcast band

This paper presents an architecture for real-time music genres classification on broadcast data from the standard Frequency Modulation (FM) radio band. The architecture is composed of an FPGA-based MFCC (Mel-Frequency Cepstral Coefficient) feature extraction stage, followed by a classification procedure. The proposed system enables automatic audio indexing of broadcast data from the standard FM radio band. Using a system-level design approach that reduces overall design time, the system was successfully implemented on Virtex 6 FPGA clocked at more than 150 MHz. The experiments have shown on one hand, a high accuracy between the FPGA-based MFCC calculation and its Matlab-based reference model, and on the other hand, nearly similar result when used in music genre classification with sequential patterns mining techniques. Furthermore, the proposed system satisfies real-time requirements, multichannel scalability and is suitable for indexing applications in embedded systems. Results are also discussed regarding operating frequencies and resources utilization.

Application of a semi-automated vocal fingerprinting approach to monitor Bornean gibbon females in an experimentally fragmented landscape in Sabah, Malaysia

Vocal individuality has been documented in a variety of mammalian species and it has been proposed that this individuality can be used as a vocal fingerprint to monitor individuals. Here we provide and test a classification method using Mel-frequency cepstral coefficients (MFCCs) to extract features from Bornean gibbon female calls. Our method is semi-automated as it requires manual pre-processing to identify and extract calls from the original recordings. We compared two methods of MFCC feature extraction: (1) averaging across all time windows and (2) creating a standardized number of time windows for each call. We analysed 376 calls from 33 gibbon females and, using linear discriminant analysis, found that we were able to improve female identification accuracy from 95.7% with spectrogram features to 98.4% accuracy when averaging MFCCs across time windows, and 98.9% accuracy when using a standardized number of windows. We divided our data randomly into training and test data-sets, and tested the accuracy of support vector machine (SVM) predictions over 100 iterations. We found that we could predict female identity in the test data-set with a 98.8% accuracy. Using SVM on our entire data-set, we were able to predict female identity with 99.5% accuracy (validated by leave-one-out cross-validation). Lastly, we used the method presented here to classify four females recorded during three or more recording seasons using SVM with limited success. We provide evidence that MFCC feature extraction is effective for distinguishing between female Bornean gibbons, and make suggestions for future vocal fingerprinting applications.

Study of MFCC and IHC Feature Extraction Methods With Probabilistic Acoustic Models for Speaker Biometric Applications

Voice is an important human trait in natural human-to-human interaction / communication for identifying a person. So voice can be regarded as a biometric measure for recognizing or identifying the person similar to other biometric measures such as face, iris and fingerprints. Speaker recognition is a class of voice recognition where speaker is identified from the speech rather than the message. Automatic speaker recognition (SR) is an approach to identify people based on features extracted from speech utterances. The major task in any speaker recognition is to extract useful features and allow meaningful patterns of speaker models. This paper compares the performance of two feature extraction techniques Mel Frequency Cepstral Coefficient (MFCC) and Inner Hair Cell Coefficient (IHC) with two different modelling methods Gaussian Mixture Model - Universal background model (GMM - UBM) and i- vector approach. In this experiment speech samples of 600 speakers from TIMIT database with 10 utterances of each speaker are taken for identifying the speaker. A text independent speaker recognition system was implemented and this study resulted in an inference of MFCC feature outperforms IHC feature for both GMM and i vector. The performance of voiced speech IHC feature which simulates the physiological behaviour of human ear is better in terms of accuracy than full speech (voiced and unvoiced) in GMM.