Filterbank optimization for robust ASR using GA and PSO

Automatic speech recognition (ASR) systems follow a well established approach of pattern recognition, that is signal processing based feature extraction at front-end and likelihood evaluation of feature vectors at back-end. Mel-frequency cepstral coefficients (MFCCs) are the features widely used in state-of-the-art ASR systems, which are derived by logarithmic spectral energies of the speech signal using Mel-scale filterbank. In filterbank analysis of MFCC there is no consensus for the spacing and number of filters used in various noise conditions and applications. In this paper, we propose a novel approach to use particle swarm optimization (PSO) and genetic algorithm (GA) to optimize the parameters of MFCC filterbank such as the central and side frequencies. The experimental results show that the new front-end outperforms the conventional MFCC technique. All the investigations are conducted using two separate classifiers, HMM and MLP, for Hindi vowels recognition in typical field condition as well as in noisy environment.     

A hierarchical language identification system for Indian languages

Automatic spoken Language IDentification (LID) is the task of identifying the language from a short duration of speech signal uttered by an unknown speaker. In this work, an attempt has been made to develop a two level language identification system for Indian languages using acoustic features. In the first level, the system identifies the family of the spoken language, and then it is fed to the second level which aims at identifying the particular language in the corresponding family. The performance of the system is analyzed for various acoustic features and different classifiers. The suitable acoustic feature and the pattern classification model are suggested for effective identification of Indian languages. The system has been modeled using hidden Markov model (HMM), Gaussian mixture model (GMM) and artificial neural networks (ANN). We studied the discriminative power of the system for the features mel frequency cepstral coefficients (MFCC), MFCC with delta and acceleration coefficients and shifted delta cepstral (SDC) coefficients. Then the LID performance as a function of the different training and testing set sizes has been studied. To carry out the experiments, a new database has been created for 9 Indian languages. It is shown that GMM based LID system using MFCC with delta and acceleration coefficients is performing well with 80.56% accuracy. The performance of GMM based LID system with SDC is also considerable.     

Language identification using phase information

The present work investigates the importance of phase in language identification (LID). We have proposed three phase based features for the language recognition task. In this work, auto-regressive model with scale factor error augmentation have been used for better representation of phase based features. We have developed three group delay based systems, namely, normal group delay based system, auto-regressive model group delay based system and auto-regressive group delay with scale factor augmentation based system. As mel-frequency cepstral coefficients (MFCCs) are extracted from the magnitude of the Fourier transform, we have combined this MFCC-based system with our phase-based systems to exploit the complete information contained in a speech signal. In this work, we have used IITKGP-MLILSC speech database and OGI Multi-language Telephone Speech (OGI-MLTS) corpus for our experiments. We have used Gaussian mixture models for building the language models. From the experimental results it is observed that the LID accuracy obtained from our proposed phase based features is comparable with MFCC features. We have also observed some performance improvement in the LID accuracy on combining the proposed phase-based systems with the state of the art MFCC-based system.     

Human and computer recognition of regional accents and ethnic groups from British English speech

The paralinguistic information in a speech signal includes clues to the geographical and social background of the speaker. This paper is concerned with automatic extraction of this information from a short segment of speech. A state-of-the-art language identification (LID) system is applied to the problems of regional accent recognition for British English, and ethnic group recognition within a particular accent. We compare the results with human performance and, for accent recognition, the ‘text dependent’ ACCDIST accent recognition measure. For the 14 regional accents of British English in the ABI-1 corpus (good quality read speech), our LID system achieves a recognition accuracy of 89.6%, compared with 95.18% for our best ACCDIST-based system and 58.24% for human listeners. The “Voices across Birmingham” corpus contains significant amounts of telephone conversational speech for the two largest ethnic groups in the city of Birmingham (UK), namely the ‘Asian’ and ‘White’ communities. Our LID system distinguishes between these two groups with an accuracy of 96.51% compared with 90.24% for human listeners. Although direct comparison is difficult, it seems that our LID system performs much better on the standard 12 class NIST 2003 Language Recognition Evaluation task or the two class ethnic group recognition task than on the 14 class regional accent recognition task. We conclude that automatic accent recognition is a challenging task for speech technology, and speculate that the use of natural conversational speech may be advantageous for these types of paralinguistic task.     

基于声学特征空间非线性流形结构的语音识别声学模型

从语音信号声学特征空间的非线性流形结构特点出发, 利用流形上的压缩感知原理, 构建新的语音识别声学模型. 将特征空间划分为多个局部区域, 对每个局部区域用一个低维的因子分析模型进行近似, 从而得到混合因子分析模型. 将上下文相关状态的观测矢量限定在该非线性低维流形结构上, 推导得到其观测概率模型. 最终, 每个状态由一个服从稀疏约束的权重矢量和若干个服从标准正态分布的低维局部因子矢量所决定. 文中给出了局部区域潜在维数的确定准则及模型参数的迭代估计算法. 基于RM语料库的连续语音识别实验表明, 相比于传统的高斯混合模型(Gaussian mixture model, GMM)和子空间高斯混合模型(Subspace Gaussian mixture model, SGMM), 新声学模型在测试集上的平均词错误率(Word error rate, WER)分别相对下降了33.1%和9.2%.     

Significance of the Modified Group Delay Feature in Speech Recognition

Spectral representation of speech is complete when both the Fourier transform magnitude and phase spectra are specified. In conventional speech recognition systems, features are generally derived from the short-time magnitude spectrum. Although the importance of Fourier transform phase in speech perception has been realized, few attempts have been made to extract features from it. This is primarily because the resonances of the speech signal which manifest as transitions in the phase spectrum are completely masked by the wrapping of the phase spectrum. Hence, an alternative to processing the Fourier transform phase, for extracting speech features, is to process the group delay function which can be directly computed from the speech signal. The group delay function has been used in earlier efforts, to extract pitch and formant information from the speech signal. In all these efforts, no attempt was made to extract features from the speech signal and use them for speech recognition applications. This is primarily because the group delay function fails to capture the short-time spectral structure of speech owing to zeros that are close to the unit circle in the z-plane and also due to pitch periodicity effects. In this paper, the group delay function is modified to overcome these effects. Cepstral features are extracted from the modified group delay function and are called the modified group delay feature (MODGDF). The MODGDF is used for three speech recognition tasks namely, speaker, language, and continuous-speech recognition. Based on the results of feature and performance evaluation, the significance of the MODGDF as a new feature for speech recognition is discussed     

Isolated word recognition by neural network models withcross-correlation coefficients for speech dynamics

This paper presents an artificial neural network (ANN) for speaker-independent isolated word speech recognition. The network consists of three subnets in concatenation. The static information within one frame of speech signal is processed in the probabilistic mapping subnet that converts an input vector of acoustic features into a probability vector whose components are estimated probabilities of the feature vector belonging to the phonetic classes that constitute the words in the vocabulary. The dynamics capturing subnet computes the first-order cross correlation between the components of the probability vectors to serve as the discriminative feature derived from the interframe temporal information of the speech signal. These dynamic features are passed for decision-making to the classification subnet, which is a multilayer perceptron (MLP). The architecture of these three subnets are described, and the associated adaptive learning algorithms are derived. The recognition results for a subset of the DARPA TIMIT speech database are reported. The correct recognition rate of the proposed ANN system is 95.5%, whereas that of the best of continuous hidden Markov model (HMM)-based systems is only 91.0%     

集成语种辨识的中英文LVCSR系统

为了在未知一段语音所属语言种类的情况下将其转换为正确的字符序列,将语种辨识(language identification,LID)同语音识别集成在一起建立了中、英文大词汇量连续语音识别(large vocabulary continuous speech recognition,LVCSR)系统.为了在中、英文连续语音识别系统中能够尽早的对语音所属的语言种类做出判决以便进行识别,从而降低解码的计算量,对语种辨识过程中的语种剪枝进行了研究,表明采用合理的语种剪枝门限在不降低系统性能的情况下,可以有效的降低系统的计算量及识别时间.     

A clustering based feature selection method in spectro-temporal domain for speech recognition

Spectro-temporal representation of speech has become one of the leading signal representation approaches in speech recognition systems in recent years. This representation suffers from high dimensionality of the features space which makes this domain unsuitable for practical speech recognition systems. In this paper, a new clustering based method is proposed for secondary feature selection/extraction in the spectro-temporal domain. In the proposed representation, Gaussian mixture models (GMM) and weighted K-means (WKM) clustering techniques are applied to spectro-temporal domain to reduce the dimensions of the features space. The elements of centroid vectors and covariance matrices of clusters are considered as attributes of the secondary feature vector of each frame. To evaluate the efficiency of the proposed approach, the tests were conducted for new feature vectors on classification of phonemes in main categories of phonemes in TIMIT database. It was shown that by employing the proposed secondary feature vector, a significant improvement was revealed in classification rate of different sets of phonemes comparing with MFCC features. The average achieved improvements in classification rates of voiced plosives comparing to MFCC features is 5.9% using WKM clustering and 6.4% using GMM clustering. The greatest improvement is about 7.4% which is obtained by using WKM clustering in classification of front vowels comparing to MFCC features.     

一种基于子带处理的PAC说话人识别方法研究

目前,说话人识别系统对于干净语音已经达到较高的性能,但在噪声环境中,系统的性能急剧下降.一种基于子带处理的以相位自相关(PAC)系数及其能量作为特征的说话人识别方法,即宽带语音信号经Mel滤波器组后变为多个子带信号,对各个子带数据经DCT变换后提取PAC系数作为特征参数,然后对每个子带分别建立HMM模型进行识别,最后在识别概率层中将HMM得出的结果相结合之后得到最终的识别结果.实验表明,该方法在不同信噪比噪声和无噪声情况下的识别性能都有很大提高.     

Front end analysis of speech recognition: a review

Automatic speech recognition (ASR) has made great strides with the development of digital signal processing hardware and software. But despite of all these advances, machines can not match the performance of their human counterparts in terms of accuracy and speed, especially in case of speaker independent speech recognition. So, today significant portion of speech recognition research is focused on speaker independent speech recognition problem. Before recognition, speech processing has to be carried out to get a feature vectors of the signal. So, front end analysis plays a important role. The reasons are its wide range of applications, and limitations of available techniques of speech recognition. So, in this report we briefly discuss the different aspects of front end analysis of speech recognition including sound characteristics, feature extraction techniques, spectral representations of the speech signal etc. We have also discussed the various advantages and disadvantages of each feature extraction technique, along with the suitability of each method to particular application.     

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

Many 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%.     

Feature representations and classification procedures for Slovene phoneme recognition

In this paper the comparison of performances of different feature representations of the speech signal and comparison of classification procedures for Slovene phoneme recognition are presented. Recognition results are obtained on the database of continuous Slovene speech consisting of short Slovene sentences spoken by female speakers. MEL-cepstrum and LPC-cepstrum features combined with the normalized frame loudness were found to be the most suitable feature representations for Slovene speech. It was found that determination of MEL-cepstrum using linear spacing of bandpass filters gave significantly better results for speaker dependent recognition. Comparison of classification procedures favours the Bayes classification assuming normal distribution of the feature vectors (BNF) to the classification based on quadratic discriminant functions (DF) for minimum mean-square error and subspace method (SM), which does not confirm the results obtained in some previous studies for German and Finn speech. Additionally, classification procedures based on hidden Markov models (HMM) and the Kohonen Self-Organizing Map (KSOM) were tested on a smaller amount of speech data (1 speaker only). Classification results are comparable with classification using BNF.     

基于差分过零峰值幅度特征的抗噪语音识别

针对多数语音识别系统在噪音环境下性能急剧下降的问题,提出了一种新的语音识别特征提取方法。该方法是建立在听觉模型的基础上,通过组合语音信号和其差分信号的上升过零率获得频率信息,通过峰值检测和非线性幅度加权来获取强度信息,二者组合在一起,得到输出语音特征,再分别用BP神经网络和HMM进行训练和识别。仿真实现了不同信噪比下不依赖人的50词的语音识别,给出了识别的结果,证明了组合差分信息的过零与峰值幅度特征具有较强的抗噪声性能。     

A new optimum feature extraction and classification method for speaker recognition: GWPNN

Speech and speaker recognition is an important topic to be performed by a computer system. In this paper, an expert speaker recognition system based on optimum wavelet packet entropy is proposed for speaker recognition by using real speech/voice signal. This study contains both the combination of the new feature extraction and classification approach by using optimum wavelet packet entropy parameter values. These optimum wavelet packet entropy values are obtained from measured real English language speech/voice signal waveforms using speech experimental set. A genetic-wavelet packet-neural network (GWPNN) model is developed in this study. GWPNN includes three layers which are genetic algorithm, wavelet packet and multi-layer perception. The genetic algorithm layer of GWPNN is used for selecting the feature extraction method and obtaining the optimum wavelet entropy parameter values. In this study, one of the four different feature extraction methods is selected by using genetic algorithm. Alternative feature extraction methods are wavelet packet decomposition, wavelet packet decomposition – short-time Fourier transform, wavelet packet decomposition – Born–Jordan time–frequency representation, wavelet packet decomposition – Choi–Williams time–frequency representation. The wavelet packet layer is used for optimum feature extraction in the time–frequency domain and is composed of wavelet packet decomposition and wavelet packet entropies. The multi-layer perceptron of GWPNN, which is a feed-forward neural network, is used for evaluating the fitness function of the genetic algorithm and for classification speakers. The performance of the developed system has been evaluated by using noisy English speech/voice signals. The test results showed that this system was effective in detecting real speech signals. The correct classification rate was about 85% for speaker classification.     

Autoregressive modeling of speech trajectory transformed to the reconstructed phase space for ASR purposes

Investigating new effective feature extraction methods applied to the speech signal is an important approach to improve the performance of automatic speech recognition (ASR) systems. Owing to the fact that the reconstructed phase space (RPS) is a proper field for true detection of signal dynamics, in this paper we propose a new method for feature extraction from the trajectory of the speech signal in the RPS. This method is based upon modeling the speech trajectory using the multivariate autoregressive (MVAR) method. Moreover, in the following, we benefit from linear discriminant analysis (LDA) for dimension reduction. The LDA technique is utilized to simultaneously decorrelate and reduce the dimension of the final feature set. Experimental results show that the MVAR of order 6 is appropriate for modeling the trajectory of speech signals in the RPS. In this study recognition experiments are conducted with an HMM-based continuous speech recognition system and a naive Bayes isolated phoneme classifier on the Persian FARSDAT and American English TIMIT corpora to compare the proposed features to some older RPS-based and traditional spectral-based MFCC features.     

Spoken Language Recognition Using Ensemble Classifiers

In this paper, we study a novel approach to spoken language recognition using an ensemble of binary classifiers. In this framework, we begin by representing a speech utterance with a high-dimensional feature vector such as the phonotactic characteristics or the polynomial expansion of cepstral features. A binary classifier can be built based on such feature vectors. We adopt a distributed output coding strategy in ensemble classifier design, where we decompose a multiclass language recognition problem into many binary classification tasks, each of which addresses a language recognition subtask by using a component classifier. Then, we combine the results of the component classifiers to form an output code as a hypothesized solution to the overall language recognition problem. In this way, we effectively project high-dimensional feature vectors into a tractable low-dimensional space, yet maintaining language discriminative characteristics of the spoken utterances. By fusing the output codes from both phonotactic features and cepstral features, we achieve equal-error-rates of 1.38% and 3.20% for 30-s trials on the 2003 and 2005 NIST language recognition evaluation databases.     

Features based on filtering and spectral peaks in autocorrelation domain for robust speech recognition

In this paper, a set of features derived by filtering and spectral peak extraction in autocorrelation domain are proposed. We focus on the effect of the additive noise on speech recognition. Assuming that the channel characteristics and additive noises are stationary, these new features improve the robustness of speech recognition in noisy conditions. In this approach, initially, the autocorrelation sequence of a speech signal frame is computed. Filtering of the autocorrelation of speech signal is carried out in the second step, and then, the short-time power spectrum of speech is obtained from the speech signal through the fast Fourier transform. The power spectrum peaks are then calculated by differentiating the power spectrum with respect to frequency. The magnitudes of these peaks are then projected onto the mel-scale and pass the filter bank. Finally, a set of cepstral coefficients are derived from the outputs of the filter bank. The effectiveness of the new features for speech recognition in noisy conditions will be shown in this paper through a number of speech recognition experiments.A task of multi-speaker isolated-word recognition and another one of multi-speaker continuous speech recognition with various artificially added noises such as factory, babble, car and F16 were used in these experiments. Also, a set of experiments were carried out on Aurora 2 task. Experimental results show significant improvements under noisy conditions in comparison to the results obtained using traditional feature extraction methods. We have also reported the results obtained by applying cepstral mean normalization on the methods to get robust features against both additive noise and channel distortion.     

基于短时能量—LPCC的语音特征提取方法研究

为了提高语音信号的识别率,提出了一种基于短时能量—LPCC的语音特征提取方法。该方法在LPCC参数的基础上,增加每帧信号的短时能量信息,使得新参数能够更为准确的表征语音信号。通过仿真实验,说明了新特征参数取得了较高的识别率。     

基于SDC特征和GMM-UBM模型的自动语种识别*

本文提出了一种基于SDC特征和GMM-UBM模型的自动语种识别方法。SDC特征由许多语音帧的一阶差分谱连接扩展而成,与传统的MFCC特征相比,包含了更多的时序特征信息。UBM模型反映了所有待识别语种的特征分布特性,借助贝叶斯自适应算法可以快速得到每个语种的模型。与传统的GMM方法相比,该方法的训练和识别的速度更快。该方法对OGI电话语音库中11个语种进行了测试,其10秒、30秒和45秒句子的最佳识别正确率分别为72.38%、82.62%和85.23%,识别速度约为0.03倍实时。