Cepstrum Features Research Articles

Real-time edge computing design for physiological signal analysis and classification

Physiological Signals like Electromography (EMG) and Electroencephalography (EEG) can be analysed and decoded to provide vital information that can be used in a range of applications like rehabilitative robotics and remote device control. The process of acquiring and using these signals requires many compute-intensive tasks like signal acquisition, signal processing, feature extraction, and machine learning. Performing these activities on a PC-based system with well-established software tools like Python and Matlab is the first step in designing solutions based upon these signals. In the application domain of rehabilitative robotics, one of the main goals is to develop solutions that can be deployed for the use of individuals who need them in improving their Acitivities-for-Daily Living (ADL). To achieve this objective, the final solution must be deployed onto an embedded solution that allows high portability and ease-of-use. Porting a solution from a PC-based environment onto a resource-constraint one such as a microcontroller poses many challenges. In this research paper, we propose the use of an ARM-based Corex M-4 processor. We explore the various stages of the design from the initial testing and validation, to the deployment of the proposed algorithm on the controller, and further investigate the use of Cepstrum features to obtain a high classification accuracy with minimal input features. The proposed solution is able to achieve an average classification accuracy of 95.34% for all five classes in the EMG domain and 96.16% in the EEG domain on the embedded board.

The Integration of Traditional Music Culture in Modern Informational Music Teaching

Abstract In this paper, in the process of music teaching, the music notes characterized in the time domain are Fourier transformed to the frequency domain, and the resulting notes in the frequency domain are analyzed and processed to obtain the inverted spectral domain features of the notes. On the basis of the cepstrum features, the music recognition model based on the depth confidence network is constructed, and after the training of the depth confidence network, the overfitting phenomenon often occurs for the depth confidence network, and the optimization is carried out by embedding the Dropout method between the implicit layers. From the perspective of music recognition and informationized music teaching, the research on traditional music culture integration of informationized music teaching is designed, and statistical analysis and simulation analysis methods are constructed. The results show that better recognition performance than using music features from different convolutional layers can be obtained using Deep Confidence Networks, with values of 77.5%, 78.8%, and 78.2%, respectively, and that music recognition research based on Deep Confidence Networks is able to better explore and pass on traditional music culture. In the linear regression analysis between the factors of incorporating folk songs into music teaching, the Sig. F and Sig values are 0, which are smaller than the significance level of 0.01 and 0.05, indicating that there is a significant relationship between the factors of students’ gender, age, whether they like folk songs, and the channels of exposure to folk songs and whether folk songs are incorporated into the music teaching program of colleges and universities. This study takes folk songs as representative of traditional music culture, raises people’s awareness of the value of folk songs, and enhances their understanding of the importance of music culture inheritance.

Advancements in Speaker Recognition: Exploring Mel Frequency Cepstral Coefficients (MFCC) for Enhanced Performance in Speaker Recognition

Abstract: Speaker recognition, a fundamental capability of software or hardware systems, involves receiving speech signals, identifying the speaker present in the speech signal, and subsequently recognizing the speaker for future interactions. This process emulates the cognitive task performed by the human brain. At its core, speaker recognition begins with speech as the input to the system. Various techniques have been developed for speech recognition, including Mel frequency cepstral coefficients (MFCC), Linear Prediction Coefficients (LPC), Linear Prediction Cepstral coefficients (LPCC), Line Spectral Frequencies (LSF), Discrete Wavelet Transform (DWT), and Perceptual Linear Prediction (PLP). Although LPC and several other techniques have been explored, they are often deemed impractical for real-time applications. In contrast, MFCC stands out as one of the most prominent and widely used techniques for speaker recognition. The utilization of cepstrum allows for the computation of resemblance between two cepstral feature vectors, making it an effective tool in this domain. In comparison to LPC-derived cepstrum features, the use of MFCC features has demonstrated superior performance in metrics such as False Acceptance Rate (FAR) and False Rejection Rate (FRR) for speaker recognition systems. MFCCs leverage the human ear's critical bandwidth fluctuations with respect to frequency. To capture phonetically important characteristics of speech signals, filters are linearly separated at low frequencies and logarithmically separated at high frequencies. This design choice is central to the effectiveness of the MFCC technique. The primary objective of the proposed work is to devise efficient techniques that extract pertinent information related to the speaker, thereby enhancing the overall performance of the speaker recognition system. By optimizing feature extraction methods, this research aims to contribute to the advancement of speaker recognition technology.

Gas-bearing prediction of deep reservoir based on DNN embeddings

The extraction of gas-bearing information from the deeply underground reservoir is extremely difficult due to the weak seismic response and complicated gas distribution characteristics. To predict gas-bearing reservoirs efficiently, we developed a deep neural network (DNN) embedding-based gas-bearing prediction scheme. First, the cepstrum coefficient that is sensitive to hydrocarbons is computed using the raw seismic data. A DNN model inspired by the x-vector in speech recognition is designed, comprising the long short-term memory (LSTM) networks and two fully connected (FC) networks, stacked from the bottom to the top layer. Then, the cepstrum features are fed into the DNN for training and testing, and DNN embedding is extracted from the top layers after optimized network parameters are determined. Finally, the gas-bearing probability of the reservoir is predicted by calculating the cosine distance between pairs of DNN embeddings. When applied to synthetic seismic data, the proposed method offers greater than 90% accuracy at SNR > 3 dB. Besides, the predicted result applied in deep carbonate reservoirs in China’s Sichuan Basin is in basic agreement with the actual situation, demonstrating the certain feasibility of the proposed scheme.

Improved Multi-Model Classification Technique for Sound Event Detection in Urban Environments

Sound event detection (SED) plays an important role in understanding the sounds in different environments. Recent studies on standardized datasets have shown the growing interest of the scientific community in the SED problem, however, these did not pay sufficient attention to the detection of artificial and natural sound. In order to tackle this issue, the present article uses different features in combination for detection of machine-generated and natural sounds. In this article, we trained and compared a Stacked Convolutional Recurrent Neural Network (S-CRNN), a Convolutional Recurrent Neural Network (CRNN), and an Artificial Neural Network Classifier (ANN) using the DCASE 2017 Task-3 dataset. Relative spectral–perceptual linear prediction (RASTA-PLP) and Mel-frequency cepstrum (MFCC) features are used as input to the proposed multi-model. The performance of monaural and binaural approaches provided to the classifier as an input is compared. In our proposed S-CRNN model, we classified the sound events in the dataset into two sub-classes. When compared with the baseline model, our obtained results show that the PLP-based ANN classifier improves the individual error rate (ER) for each sound event, e.g., the error rate (ER) is improved to 0.23 for heavy vehicle events and 0.32 for people walking, and minor gains are shown in other events as compared to the baseline. Our proposed CRNN performs well when compare to the baseline and to our proposed ANN model. Moreover, in cross-validation trials, the results in the evaluation stage demonstrate a significant improvement compared to the best performance of DCASE 2017 Task-3, reducing the ER to 0.11 and increasing the F1-score by 10% in the evaluation dataset. Erosion and dilation were used during post-processing.

Underground pipeline surveillance with an algorithm based on statistical time-frequency acoustic features

Underground pipeline networks suffer from severe damage by earth-moving devices due to rapid urbanization. Thus, designing a round-the-clock intelligent surveillance system has become crucial and urgent. In this study, we develop an acoustic signal-based excavation device recognition system for underground pipeline protection. The front-end hardware system is equipped with an acoustic sensor array, an Analog-to-Digital Converter (ADC) module (ADS1274), and an industrial processor Advanced RISC Machine (ARM) cortex-A8 for signal collection and algorithm implementation. Then, a novel Statistical Time-Frequency acoustic Feature (STFF) is proposed, and a fast Extreme Learning Machine (ELM) is adopted as the classifier. Experiments on real recorded data show that the proposed STFF achieves better discriminative capability than the conventional acoustic cepstrum features. In addition, the surveillance platform is applicable for encountering big data owing to the fast learning speed of ELM.

Speech Emotion Recognition

Abstract: Speech Emotion Recognition, abbreviated as SER, the act of trying to identify a person's feelings and relationships. Affected situations from speech. This is because the truth often reflects the basic feelings of tone and tone of voice. Emotional awareness is a fast-growing field of research in recent years. Unlike humans, machines do not have the power to comprehend and express emotions. But human communication with the computer can be improved by using automatic sensory recognition, accordingly reducing the need for human intervention. In this project, basic emotions such as peace, happiness, fear, disgust, etc. are analyzed signs of emotional expression. We use machine learning techniques such as Multilayer perceptron Classifier (MLP Classifier) which is used to separate information provided by groups to be divided equally. Coefficients of Mel-frequency cepstrum (MFCC), chroma and mel features are extracted from speech signals and used to train MLP differentiation. By accomplishing this purpose, we use python libraries such as Librosa, sklearn, pyaudio, numpy and audio file to analyze speech patterns and see the feeling. Keywords: Speech emotion recognition, mel cepstral coefficient, neural artificial network, multilayer perceptrons, mlp classifier, python.

Feature Extraction Method for Hidden Information in Audio Streams Based on HM-EMD

Using fake audio to spoof the audio devices in the Internet of Things has become an important problem in modern network security. Aiming at the problem of lack of robust features in fake audio detection, an audio streams’ hidden feature extraction method based on a heuristic mask for empirical mode decomposition (HM-EMD) is proposed in this paper. First, using HM-EMD, each signal is decomposed into several monotonic intrinsic mode functions (IMFs). Then, on the basis of IMFs, basic features and hidden information features HCFs of audio streams are constructed, respectively. Finally, a machine learning method is used to classify audio streams based on these features. The experimental results show that hidden information features of audio streams based on HM-EMD can effectively supplement the nonlinear and nonstationary information that traditional features such as mel cepstrum features cannot express and can better realize the representation of hidden acoustic events, which provide a new research idea for fake audio detection.

Use of Machine Learning for Deception Detection From Spectral and Cepstral Features of Speech Signals

In this research, four unique nonlinear speech features are extracted and analyzed to study the dissimilarity pattern between when the speaker is being deceitful and truthful based on how human speech is perceived. The speaker was under stress in a police interrogation where two ground truth and two deceitful responses were recorded during three different times of the day. Using the audio recordings from all three sessions, the cepstral features and spectral energy features are extracted. Cepstral features are the Mel frequency cepstrum coefficient, from where the delta cepstrum and the time-difference cepstrum features are developed. On the other hand, the spectral energy features are the energy of Bark band energy from where the delta energy and the time-difference energy features are developed. The Levenberg-Marquardt classification method and the long short-term memory classification method are then applied to evaluate the accuracy of detecting deception based on the nine unique training and testing combinations of the three different sessions and their extracted cepstrum and spectral energy features. In addition, the principal component analysis is applied to reduce the dimensionality from the extracted features for further improvement. The projected principal components of the four types of features showed improved accuracy in order to distinguish between truthful and deceptive speech pattern. After incorporating with principal component analysis, the long short-term memory classification method with time-difference spectral energy feature shows the highest recognition rate compared to Levenberg-Marquardt algorithm with other cepstral and spectral features.

Optimisation of training samples in recognition of overlapping speech and identification of speaker in a two speakers situation

Recognition of overlapping speech is still a challenging problem in the area of automatic speech recognition (ASR). In this paper, we have proposed a technique for overlapping speech recognition integrated with firefly optimization technique. Overlapped segments are thoroughly analysed for different dominant frequencies involved in the mixture. We have created an audio splitting function. Split audio signals are converted into mel cepstral coefficients and the intensity variations of signals are indicated by their cepstrum. Phoneme density updated cepstrum (PDUC) features are extracted from both spectrum feature analysis and mel frequency cepstral coefficients (MFCC) coefficients. Further, firefly optimization technique is used for clustering and selecting best relevant features. Datasets of speech separation challenge (SSC) Scopus are used to evaluate the results. From the results, we could conclude that minimum samples of 20% to 30% are sufficient to achieve recognition accuracy of above 90%.

Speaker identification based on deep learning in FX iDeal system

The FX (foreign exchange) iDeal system, which provides instant foreign exchange information for inter bank traders, with the rapidly increased number of users, there are urgently needs to meet the higher security requirements and improve the compliance of transactions in the inter bank market. In this paper, we make a research on speaker recognition to improve the security of the system. According to the historical sound data sent by the user, we extract the Mel cepstrum features, and then construct the speaker’s identification model based on the deep neural network (DNN). Based on the model, we can verify the user. In order to prevent the DNN from over-fitting or falling into the local minimum, we add a dropout mechanism and a dynamic learning rate adjustment algorithm to the network. To maximize the data mining capabilities, we use the restricted Boltzmann machine (RBM) to realize unsupervised training which can make the network initialization parameters are closer to the global minimum, and then optimize the parameters by supervised training. Verification experiment shows that the model can effectively improve the speaker confirmation ability and is more effective than the general BP neural network and GMM model.

Language and Gender Classification of Speech Files Using Supervised Machine Learning Methods

ABSTRACTMany language identification (LID) systems are based on language models using techniques that consider the fluctuation of speech over time. Considering these fluctuations necessitates longer recording intervals to obtain reasonable accuracy. Our research extracts features from short recording intervals to enable successful classification of spoken language. The feature extraction process is based on frames of 20 ms, whereas most previous LIDs presented results based on much longer frames (3 s or longer). We defined and implemented 200 features divided into four feature sets: cepstrum features, RASTA features, spectrum features, and waveform features. We applied eight machine learning (ML) methods on the features that were extracted from a corpus containing speech files in 10 languages from the Oregon Graduate Institute (OGI) telephone speech database and compared their performances using extensive experimental evaluation. The best optimized classification results were achieved by random forest (RF): from 76.29% on 10 languages to 89.18% on 2 languages. These results are better or comparable to the state-of-the-art results for the OGI database. Another set of experiments that was performed was gender classification from 2 to 10 languages. The accuracy and the F measure values for the RF method for all the language experiments were greater than or equal to 90.05%.

A Speech Signal Logarithmic Spectrum Amplitude Envelope Algorithm Analysis

Cepstrum method is a traditional characteristic parameter detection algorithm in the modern in the speech signal processing . it is one of the speech signal using the cepstrum features, to detect the representation glottis incentive cycle pitch information.In order to eliminate the influence of the fundamental frequency harmonic, we utilize the homomorphism solution volume technology, getting smooth spectral envelope.While the artical analyses cepstrum from out of the signal deleted to the short time window , detect speech signal pitch information;Then the signal of Fourier transform, take the logarithmic spectrum amplitude envelope, to analyze the speech signal formant parameters. The speech signal characteristic parameters have higher accuracy than ones from the traditional cepstrum method.

Keyboard acoustic emanations revisited

We examine the problem of keyboard acoustic emanations. We present a novel attack taking as input a 10-minute sound recording of a user typing English text using a keyboard and recovering up to 96% of typed characters. There is no need for training recordings labeled with the corresponding clear text. A recognizer bootstrapped from a 10-minute sound recording can even recognize random text such as passwords: In our experiments, 90% of 5-character random passwords using only letters can be generated in fewer than 20 attempts by an adversary; 80% of 10-character passwords can be generated in fewer than 75 attempts by an adversary. In the attack, we use the statistical constraints of the underlying content, English language, to reconstruct text from sound recordings without knowing the corresponding clear text. The attack incorporates a combination of standard machine learning and speech recognition techniques, including cepstrum features, Hidden Markov Models, linear classification, and feedback-based incremental learning.

Cross-entropic comparison of formants of British, Australian and American English accents

This paper highlights the differences in spectral features between British, Australian and American English accents and applies the cross-entropy information measure for comparative quantification of the impacts of the variations of accents, speaker groups and recordings on the probability models of spectral features of phonetic units of speech. Comparison of the cross-entropies of formants and cepstrum features indicates that formants are a better indicator of accents. In particular it appears that the measurements of differences in formants across accents are less sensitive to different recordings or databases compared to cepstrum features. It is found that the cross-entropies of the same phonemes across speaker groups with different accents (inter-accent distances) are significantly greater than the cross-entropies of the same phonemes across speaker groups of the same accent (intra-accent distances). Comparative evaluations presented on cross-gender speech recognition shows that accent differences have an impact comparable to gender differences. The cross-entropy measure is also used to construct cross-accent phonetic-trees, which serve to show the structural similarities and differences of the phonetic systems across accents.

Micro-controller based Remote Monitoring using Mobile through Spoken Commands

Mobile phone can serve as powerful tool for world-wide communication. A system is developed to remotely monitor process through spoken commands using mobile. Mel cepstrum features are extracted from spoken words. Learning Vector Quantization Neural Network is used for recognition of various words used in the command. The accuracy of spoken commands is about 98%. A text message is generated and sent to control system mobile in form of SMS. On receipt of SMS, control system mobile informs AVR micro-controller based card, which performs specified task. The system alerts user in case of occurrence of any abnormal conditions like power failure, loss of control, etc. Other applications where this approach can be extended are also discussed.

Selecting Good Speech Features for Recognition

This paper describes a method to select a suitable feature for speech recognition using information theoretic measure. Conventional speech recognition systems heuristically choose a portion of frequency components, cepstrum, mel-cepstrum, energy, and their time differences of speech waveforms as their speech features. However, these systems never have good performance if the selected features are not suitable for speech recognition. Since the recognition rate is the only performance measure of speech recognition system, it is hard to judge how suitable the selected feature is. To solve this problem, it is essential to analyze the feature itself, and measure how good the feature itself is. Good speech features should contain all of the class-related information and as small amount of the class-irrelevant variation as possible. In this paper, we suggest a method to measure the class-related information and the amount of the class-irrelevant variation based on the Shannon's information theory. Using this method, we compare the mel-scaled FFT, cepstrum, mel-cepstrum, and wavelet features of the TIMIT speech data. The result shows that, among these features, the mel-scaled FFT is the best feature for speech recognition based on the proposed measure.