Forensic and Automatic Speaker Recognition System

Speech feature extraction is the most significant step in any Automatic speaker recognition system. In the last 60 years a lot of research has gone into parametric representation of these speech features. Several techniques are currently being used for Automatic Speaker Recognition. Yet Automatic Speaker Recognition still remains a confront mainly due to variations in speaker's vocal tract with time and health, varying environmental conditions, disparities in the behavior and quality of speech recorders etc. MFCC is a extensively used technique in Automatic speaker recognition. In this paper the performance of MFCC technique was evaluated in a quiet environment. A speaker database containing 30 male and 30 female speakers was created. Two separate experiments were conducted for the performance evaluation of MFCC technique when applied to K means clustering. In the first case the speech features were directly matched. In the second case a VQ codebook was created by clustering the training features of these 60 speakers. A distortion easure based on the minimum Euclidean distance was used for speaker recognition. The failure rate of speaker recognition in first ase was found to be was found to be 10% while in the second case as found to be 14%. Matlab-7.10.0 was used for this study.

Parameters of speech produced during winter weather are analyzed to determine the potential influence of distorted voices due to common cold on automatic speaker recognition systems. This research examines speech recorded before, during, and after bouts with the common cold. ‘‘Cold-affected’’ speech shows noticeable acoustic and phonetic differences between test sessions. Examining the different states of one male speaker’s voice allows an opportunity to address the speaker intersession problem that exists in speaker recognition technology. Specific emphasis is placed on the contrasts in various sessions of TIMIT inspired sentence ‘‘She had your dark suit in greasy wash water all year.’’ It also analyzes phonetic contrasts and looks at differences in formant patterns. Phonetic transcriptions of ‘‘cold’’ and ‘‘normal’’ sessions reflect changes in place of articulation. Perceptual and acoustic analyses reveal pauses and epenthetic syllables that are not constant throughout all sessions. ‘‘Cold-speech’’ also produces noise caused by hoarseness and coughing. Research presented here shows evidence of variability beyond ‘‘normal’’ speech patterns. Contributing to the information on intraspeaker variability, this work provides a preliminary analysis of ‘‘cold-affected’’ speech in an effort to include ‘‘abnormal’’ speech and speech affected by communications disorders into the framework of automatic speaker recognition.

This paper presents an experimental evaluation of the combined temporal and spectral processing methods for speaker recognition task under noise, reverberation or multi-speaker environments. Automatic speaker recognition system gives good performance in controlled environments. Speech recorded in real environments by distant microphones is degraded by factors like background noise, reverberation and interfering speakers. This degradation strongly affects the performance of the speaker recognition system. Combined temporal and spectral processing (TSP) methods proposed in our earlier study are used for pre-processing to improve the speaker-specific features and hence the speaker recognition performance. Different types of degradation like background noise, reverberation and interfering speaker are considered for evaluation. The evaluation is carried out for the individual temporal processing, spectral processing and the combined TSP method. The experimental results show that the combined TSP methods give relatively higher recognition performance compared to either temporal or spectral processing alone.

The paper presents the architecture and the results of optimization of selected elements of the Automatic Speaker Recognition (ASR) system that uses Gaussian Mixture Models (GMM) in the classification process. Optimization was performed on the process of selection of individual characteristics using the genetic algorithm and the parameters of Gaussian distributions used to describe individual voices. The system that was developed was tested in order to evaluate the impact of different compression methods used, among others, in landline, mobile, and VoIP telephony systems, on effectiveness of the speaker identification. Also, the results were presented of effectiveness of speaker identification at specific levels of noise with the speech signal and occurrence of other disturbances that could appear during phone calls, which made it possible to specify the spectrum of applications of the presented ASR system.

Likelihood ratio calibration is a very important step when applying an automatic speaker recognition system to forensic applications. This paper presents the process and evaluation of a speaker recognition system developed with spoken Indonesian database. The system is developed using MFCC feature, GMM-UBM modeling, and Z normalization. System performances were evaluated based on male and female genders, and two scenarios i.e., natural conversation and interview. System evaluations were done using performance measures for both discrimination and calibration abilities. Results show that based on various indicators, the system behaves very well. This is shown by the best achievable EER values of 4.66%, and Cmc values of 0.04. Therefore, the developed automatic speaker recognition system is now ready to be used for forensic applications in Indonesia.

Real world conditions differ from ideal or laboratory conditions, causing mismatch between training and testing phases, and consequently, inducing performance degradation in automatic speaker recognition systems. Many strategies have been adopted to cope with acoustical degradation; in some applications of speaker identification systems a clean sample of speech, prior to the recognition stage, is needed. This has justified the use of procedures that may reduce the impact of acoustical noise on the desired signal, giving rise to techniques involved in the enhancement of noisy speech. A comparative performance analysis of single-channel (based in classical spectral subtraction and some derived alternatives), dual-channel (based in adaptive noise cancelling) and multi-channel (using microphone arrays) speech enhancement techniques, with different types of noise at different SNRs, as a pre-processing stage to an ergodic HMM-based speaker recognizer, is presented.

This paper analyzes the effects popular audio compression algorithms have on the performance of a speaker recognition system. Popular audio compression algorithms were used to compress both clean and noisy speech before being passed to a speaker recognition system. The features extracted from each speaker were 19-dimensional Mel-Frequency Cepstrum Coefficients (MFCC) and the corresponding features were modeled using a 16 mixture Gaussian Mixture Model (GMM). Our experiments show that compression will have a negative effect on recognition rates if the compressed speech is clean. However, if small amounts of white Gaussian noise are added before the speech is compressed, recognition rates can be increased by as much as 7% with certain compression algorithms.

Recognition of a particular speaker amongst a crowded group of people is one of the current areas of interest of researchers. This paper presents a Text Dependent Automatic Speaker recognition system developed and simulated using Matlab. For a better computational efficiency, the system is trained to store voice of the same person under various physiological conditions such as coughing, shouting, during chewing, mouth covered etc. A dictionary is created to store the signature features of each user's voice. A neural networks is then trained using back propagation and accordingly weights are obtained to recognize voice in the testing phase. The efficiency of the proposed system is then compared to the system implemented using vector quantization.

This chapter describes a method for enhancing the differences between speaker classes at the feature level (feature enhancement) in an automatic speaker recognition system. The original Mel-frequency cepstral coefficient (MFCC) space is projected onto a new feature space by a neural network trained on a subset of speakers which is representative for the whole target population. The new feature space better discriminates between the target classes (speakers) than the original feature space. The chapter focuses on the method for selecting a representative subset of speakers, comparing several approaches to speaker selection. The effect of feature enhancement is tested both for clean and various noisy speech types to evaluate its applicability under practical conditions. It is shown that the proposed method leads to a substantial improvement in speaker recognition performance. The method can also be applied to other automatic speaker classification tasks.

In this research, we present an automatic speaker recognition system based on adaptive orthogonal transformations. To obtain the informative features with a minimum dimension from the input signals, we created an adaptive operator, which helped to identify the speaker’s voice in a fast and efficient manner. We test the efficiency and the performance of our method by comparing it with another approach, mel-frequency cepstral coefficients (MFCCs), which is widely used by researchers as their feature extraction method. The experimental results show the importance of creating the adaptive operator, which gives added value to the proposed approach. The performance of the system achieved 96.8% accuracy using Fourier transform as a compression method and 98.1% using Correlation as a compression method.

Speaker recognition is an advanced method to identify a person from the biometric characteristics of speaking voice samples. Speaker recognition has become a vastly popular and useful research subject with countless essential applications in security, assistance, replication, authentication, automation, and verification. Many techniques are implemented using deep learning and neural network concepts and various datasets for speaker verification and identification. The primary goal of this work is to create improved robust techniques of speaker recognition to identify audio and enhance accuracy to human levels of comprehension. TIMIT and LibriSpeech datasets are used in this paper to develop an efficient automatic speaker recognition system. This work focuses on using MFCC to transform audio to spectrograms without losing the essential features of the audio file in question. We have used a closed set and an open set implementation procedure on these datasets. The closed set implementation uses a standard machine learning convention of utilizing the same datasets for training and testing, leading to higher accuracy. On the other hand, the open set implementation uses one dataset to train and another to test on each occasion. The accuracy, in this case, turned out to be relatively lower. On each dataset, CNN and LSTM deep learning techniques have been used to identify the sound, leading to the observation that implementing CNN resulted in a more significant accuracy.

A system is presented which will identify speakers that it was trained to recognize with 100% accuracy. This system will also classify inputs that it has not been trained to recognize as unknown. The high degree of confidence offered by this system is attributed to careful processing of the input data sets. Only easily distinguishable segments of the speech samples are passed as inputs to the artificial neural network (ANN). Additionally, sufficient input information is provided with the training inputs to enable the ANN to rapidly train and subsequently classify speakers. This system may be readily implemented in hardware using existing technology. Hardware implementation will render near real-time performance. Applications that would be enhanced by the use of this system include electronic surveillance and computer security. The speaker identification system was successfully tested. >

Acoustical mismatch between training and testing phases induces degradation of performance in automatic speaker recognition systems. Providing robustness to speaker recognizers has to be, therefore, a priority matter. Robustness in the acoustical stage can be accomplished through speech enhancement techniques as a prior stage to the recognizer. These techniques are oriented to the reduction of the impact that acoustical noise produces on the input signal. In this paper, several spectral subtraction-derived techniques are used to enhance single-channel noisy speech. Other perspectives, based in dual-channel (adaptive filtering) and multi-channel (microphone arrays) processing are also presented as optimal solutions to speech enhancement needs. A comparative analysis of the proposed techniques, with different types of noise at different SNRs, as a pre-processing stage to an ergodic HMM-based speaker recognizer, is presented.

Automatic speaker recognition systems typically rely on parameters derived from resonance features of the vocal tract. Th is implies that the more similar the geometry of two vocal tracts is, the more similar will be the respective similarity coeffi cients, or likelihood ratios (LRs). Quite obviously this problem is particularly relevant to related speakers, most extremely for identical (monozygotic) twins. Th is paper is about an experiment with 9 male and 26 female pairs of identical twins who produced one read and one spontaneous speech sample. An automatic system for forensic speaker recognition (Batvox 3.1) was used to calculateinter-speaker (non-target), (2) intratwin pair, and (3) intra-speaker (target) LR distributions. Results show that in certain conditions an automatic Bayesian-based system is capable of distinguishing even the vast majority of very similar sounding voices such as those of identical twins. However, the performance of the system used here was superior for male as compared to female voices. Quite obviously the sex-related diff erence was enhanced by the genetic similarity factor.

Context. The questions of adapting the convolution neural network classifier use in automatic speaker recognition system of critical use(ASRSCU) are considered. The research object is the individual features of the human speech process.Objective. Development of means for separating individual features from the speaker’s speech signal, increasing their informativeness asa result of the factor analysis, their visual representation for the use of the convolution neural network classifier, and optimizing itsarchitecture for the needs of ASRSCU.Method. Measures are proposed to optimize the speaker recognition procedure of the ASRSCU, for which the optimal way of informativefeatures representation and the method of increasing their informativeness are theoretically justified, the topology and measures for increasingof the speaker recognition process efficiency are justified. In particular, it is justified the use of power normalized cepstral coefficients (PNCC)for the description of phonograms recorded in noisy environment conditions. We propose to use Gabor filters to represent information thatwill be analyzed by a convolution neural network, an optimal method of factor analysis (a sparse main components analyzing method) toreduce of the features vector length while preserving its informativeness, an improved topology of the convolution neural network in whichthe Gabor filters are integrated in to the convolution layer, which allows them to optimize their parameters during the neural network trainingprocess, and in a fully connected layer a deep neural network with a bottleneck layer is used, whose weights after training are uses as inputs forthe GMM/HMM control classifier.Results. Methods of representation and optimization of the speaker’s individual features, methods for their visual presentation andimprovement of the topology of a convolution neural network for making speaker recognition on their basis.Conclusions. The obtained theoretical results have found empirical confirmation. In particular, the stability of an improved convolutionneural network to the noisy input phonograms proved to be higher than the results of an ordinary convolution neural network and a deep neuralnetwork. With an SNR increase up to 10 dB, the GMM/HMM classifier is more efficient than the neural network, which can be explained by the efficiency of the used UBM models, but it is much more resource-intensive. Also, the parameters of the Gabor filter bank frames thatprovide the most variable individual features from the speech signal for speaker recognition are determined empirically.