Analysis of Large-Scale SVM Training Algorithms for Language and Speaker Recognition

Speaker and language recognition in noisy and degraded channel conditions continue to be a challenging problem mainly due to the mismatch between clean training and noisy test conditions. In the presence of noise, the most reliable portions of the signal are the high energy regions which can be used for robust feature extraction. In this paper, we propose a front end processing scheme based on autoregressive (AR) models that represent the high energy regions with good accuracy followed by a modulation filtering process. The AR model of the spectrogram is derived using two separable time and frequency AR transforms. The first AR model (temporal AR model) of the sub-band Hilbert envelopes is derived using frequency domain linear prediction (FDLP). This is followed by a spectral AR model applied on the FDLP envelopes. The output 2-D AR model represents a low-pass modulation filtered spectrogram of the speech signal. The band-pass modulation filtered spectrograms can further be derived by dividing two AR models with different model orders (cut-off frequencies). The modulation filtered spectrograms are converted to cepstral coefficients and are used for a speaker recognition task in noisy and reverberant conditions. Various speaker recognition experiments are performed with clean and noisy versions of the NIST-2010 speaker recognition evaluation (SRE) database using the state-of-the-art speaker recognition system. In these experiments, the proposed front-end analysis provides substantial improvements (relative improvements of up to 25%) compared to baseline techniques. Furthermore, we also illustrate the generalizability of the proposed methods using language identification (LID) experiments on highly degraded high-frequency (HF) radio channels and speech recognition experiments on noisy data.

Relevance factor of maximum a posteriori adaptation for GMM–NAP–SVM in speaker and language recognition

ABSTRACT: Speech is one of the most natural and efficient forms of communication among humans, and it is becoming increasingly integral to human-computer interaction in modern applications such as virtual assistants, smart devices, and automated customer service systems. Among the various technologies that leverage speech, speaker recognition stands out as a biometric method that identifies or verifies individuals based on the unique characteristics of their voice. This approach has gained significant traction due to its non-intrusive nature, ease of integration, and applicability across various domains such as security, forensics, and personalised user experiences. In this paper, we introduce a novel speaker recognition system specifically designed to identify speakers based on utterances in the Marathi language, a linguistically rich and widely spoken regional language in India. The system utilises Mel-Frequency Cepstral Coefficients (MFCCS) to extract distinguishing vocal features that closely mimic the human auditory perception. MFCCS are particularly effective in capturing speech’s phonetic and acoustic properties, making them a preferred choice in speech and speaker recognition tasks. Vector quantisation (VQ) is applied to optimise the feature set and reduce computational complexity. VQ compresses the high-dimensional MFCC feature vectors into representative clusters without significantly compromising accuracy, thereby enhancing system efficiency. These features are then processed using hidden Markov Models (HMMS), which are adept at modelling temporal sequences and dynamic variations in speech. HMMS offer a statistical framework that effectively captures the sequential nature of speech patterns, leading to more reliable speaker modelling and recognition.

Automatic speech and speaker recognition are traditionally treated as two independent tasks and are studied separately. The human brain in contrast deciphers the linguistic content, and the speaker traits from the speech in a collaborative manner. This key observation motivates the work presented in this paper. A collaborative joint training approach based on multitask recurrent neural network models is proposed, where the output of one task is backpropagated to the other tasks. This is a general framework for learning collaborative tasks and fits well with the goal of joint learning of automatic speech and speaker recognition. Through a comprehensive study, it is shown that the multitask recurrent neural net models deliver improved performance on both automatic speech and speaker recognition tasks as compared to single-task systems. The strength of such multitask collaborative learning is analyzed, and the impact of various training configurations is investigated.

Cepstral features and their derivatives have been successfully used in speaker recognitions tasks over the past two decades. Recently new features are emerging that make use of information that is not contained in cepstral features or their derivative, namely the phase spectra and instantaneous frequency. There are numerous approaches to capture instantaneous amplitude and frequency variations in the signal, resulting in an amplitude and frequency modulation (AM-FM) feature set. While the addition of modulation features has been shown to improve speech recognition systems based on cepstral features, their use in speaker recognition is new and has not been fully explored. This paper outlines some of these emerging features and their extraction methods for use in speaker recognition tasks. The paper also talks about the recently proposed Empirical Mode Decomposition (EMD) as an alternative method of extracting instantaneous frequency.

The paper introduces a combination of adaptive neural network systems and statistical method for integrating speech and face image information for person identification. The method allows for the development of models of persons and their on-going adjustment based on new speech and face images. The method is illustrated with a modeling and classification of different persons, when speech and face images are presented in an incremental way. In this model, there are two subnetworks, one for face image and one for speaker recognition. A higher-level layer is applied to make a final decision. In the speaker recognition subnetwork, a text-dependant model is built using evolving connectionist systems (ECOS) [N. Kasabov, 2002]. In the face image recognition subnetwork, composite profile technique is applied for face image feature extraction and zero instruction set computing (ZISC) [ZISC Manual, 2000] technology is used to build the neural network. In the higher-level conceptual subsystem, final recognition decision is made using statistical method. The experiments show that ECOS and ZISC are appropriate techniques for the creation of evolving models for the task of speaker and face recognition individually. It is also shown that the integration of the speech and image information using statistical method improves the person identification rate.

In the biometrics community, face and speaker recognition are mature fields in which several systems have been proposed over the past twenty years. While existing systems perform well under controlled recording conditions, mismatch caused by the use of different sensors or a lack of cooperation from the subject still significantly affects performance, especially in challenging scenarios such as in forensics. Furthermore, existing methods suffer from scalability issues, which prevents them from taking advantage of increasingly large amounts of training data. This is otherwise a promising approach to improve accuracy in such challenging scenarios. In this thesis we address these problems of mismatch and complexity by developing scalable probabilistic models that we apply to face, speaker and bimodal recognition. Our contributions are four-fold. First, we propose a unified framework for session variability modeling techniques based on Gaussian mixture models (GMM), that encompasses inter-session variability (ISV) modeling, joint factor analysis (JFA) and total variability (TV) modeling. Second, we propose a novel exact and scalable formulation of probabilistic linear discriminant analysis (PLDA), which is a probabilistic and generative framework that models between-class and within-class variations. This formulation solves a major scalability issue, by improving both the time complexity of the training procedure from cubic to linear with respect to the number of samples per class, and the complexity of the scoring procedure. Furthermore, the implementations of all the proposed techniques are integrated into a novel collaborative open source software library called Bob 1 that enforces fair evaluations and encourages reproducible research. Fourth and finally, large-scale experiments are conducted with all of the above machine learning algorithms on several databases such as FRGC for face recognition, NIST SRE12 for speaker recognition and MOBIO for bimodal recognition, showing competitive performances.

Speaker Recognition is the ability of the system to recognize the speaker from the set of speaker samples available in the system. It is of 2 types, one uses a keyword, called text-dependent systems, and another one can recognize the voice in any language/text, also called as text-independent speaker recognition. In this paper, a text-independent, language-independent speaker recognition system is implemented using dense & convolutional neural networks. Speaker recognition has found several applications in upcoming electronic products like personal/home assistants, telephone banking and biometric identification. In this paper, we explore a system that uses MFCC along with DNN and CNN as the model for building a speaker recognition system.

Automatic speaker recognition determines whether a person is who he/she claims to be without the intervention of another human being, from their voiceprint. In recent years different methods of voice analysis to fulfill this objective have been developed. At the moment, the development of computational auditory models that emulate the physiology and function of the inner ear have added a new tool in the field of speaker recognition, including the Triple Resonance Nonlinear filter. This paper studies the behavior of a bio-inspired model of inner ear applied to the speakers's voice analysis. Their ability on speaker recognition tasks was statistically analyzed. We conclude, given the results obtained, that this system is an excellent tool, which shows a high rate of sensitivity and specificity in speaker recognition.

Extraction and representation of prosodic features for language and speaker recognition

A method is described for estimating telephone handset nonlinearity by matching the spectral magnitude of the distorted signal to the output of a nonlinear channel model, driven by an undistorted reference. This "magnitude only" representation allows the model to directly match unwanted speech formants that arise over nonlinear channels and that are a potential source of degradation in speaker and speech recognition algorithms. As such, the method is particularly suited to algorithms that use only spectral magnitude information. The distortion model consists of a memoryless nonlinearity sandwiched between two finite-length linear filters. Nonlinearities considered include arbitrary finite-order polynomials and parametric sigmoidal functionals derived from a carbon-button handset model. Minimization of a mean-squared spectral magnitude distance with respect to model parameters relies on iterative estimation via a gradient descent technique. Initial work has demonstrated the importance of addressing handset nonlinearity, in addition to linear distortion, in speaker recognition over telephone channels. A nonlinear handset "mapping," applied to training or testing data to reduce mismatch between different types of handset microphone outputs, improves speaker verification performance relative to linear compensation only. Finally, a method is proposed to merge the mapper strategy with a method of likelihood score normalization (hnorm) for further mismatch reduction and speaker verification performance improvement.

Deep CNN models have become state-of-the-art techniques in many application, e.g., face recognition, speaker recognition, and image classification. Although many studies address on speedup or compression of individual models, very few studies focus on co-compressing and unifying models from different modalities. In this work, to joint and compress face and speaker recognition models, a shared-codebook approach is adopted to reduce the redundancy of the combined model. Despite the modality of the inputs of these two CNN models are quite different, the shared codebook can support two CNN models of sound and image for speaker and face recognition. Experiments show the promising results of unified and co-compressing heterogeneous models for efficient inference.

RESUMEN Currently, many speaker recognition applications must handle speech corrupted by environmental additiv e noise without having a priori knowledge about the charact eristics of noise. Some previous works in speaker r ecognition have used Missing Feature (MF) approach to compensate for noise. In most of those applications the spe ctral reliability decision step is done using the Signal to Noise Ratio (SNR) criterion. This has the goal o f enhancing signal power rather than noise power, which could b e dangerous in speaker recognition tasks, because u seful speaker information could be removed. This work proposes a new mask estimation method based on Speaker Discriminative Information (SDI) for determining spectral reliabil ity in speaker recognition applications based on th e MF approach. The proposal was evaluated through speaker verification experiments in speech corrupted by additive no ise. Experiments demonstrated that this new criterion ha s a promising performance in speaker verification t asks.

As one of biometric identification technologies, speaker recognition shows better application prospects in many fields. At present, the implementation of speaker recognition algorithm on the hardware is mostly based on System on a Programmable Chip(SOPC) platform of Field Programmable Gate Array(FPGA) with Nios II Intellectual Property(IP) core. And the algorithm can be selected and optimized effectively on SOPC. However the high-speed and parallel operation of FPGA can't be fully utilized. By researching and analyzing the speaker recognition algorithm, a FPGA-based speaker recognition method is presented in this paper. This method includes the Voice Active Detection (VAD), Mel Frequency Cepstrum Coefficient (MFCC) extraction and Vector Quantization (VQ) recognition algorithm. By using the technology such as Ping-pang operation, Pipeline and module reuse, it makes full use of FPGA's speed. After testing, this method can effectively meet the requirement of real-time data processing.

To evaluate the performance of speaker recognition systems, a detection cost function defined as a weighted sum of the probabilities of type I and type II errors is employed. The speaker datasets may have data dependency due to multiple uses of the same subjects. Using the standard errors of the detection cost function computed by means of the two-layer nonparametric two-sample bootstrap method, a significance test is performed to determine whether the difference between the measured performance levels of two speaker recognition algorithms is statistically significant. While conducting the significance test, the correlation coefficient between two detection cost functions for two algorithms, respectively, is taken into account. Examples are provided.