Approach For Speech Recognition Research Articles

A HMM Integrated SVM Model for Hindi Speech Recognition

Speech is one of the most useful and effective communication medium used as the biometric feature of human being. Speech Recognition is considered as important part of various applications for biometric recognition and translation. Speech can be acquired inexpensively using some mic or phone device. But because of different noise factors that can be included during acquisition increases the complexity in recognition process. Because of this, there is the requirement of more effective and reliable approach for speech recognition. In this present work, a statistical analysis based predictive model is defined to improve the speech recognition. The presented work is defined in the form of a layered model. In first layer of this model, the speech signal improvement is done. To achieve this, the Discrete Wavelet Transform (DWT) and spectral subtraction based approach is defined for noise reduction. This layer improves the speech features. In second layer, the Hidden Markov Model (HMM) improved statistical model is defined to generate the speech features. This layer uses the segmented mean value, standard deviation, variation analysis as the statistical features. In final stage, Distance based mapping is applied on all these features collectively to perform the recognition. The work is implemented for Hindi word recognition. The work is implemented in Matlab environment.

Speech-to-Speech Translation: A Review

This paper reviews the technology used in Speech-to-Speech Translation that is the phrases spoken in one language are immediately spoken in another language by the device. Speech-to-Speech Translation is a three step software process which includes Automatic speech Recognition, Machine Translation and voice synthesis. This paper includes the major speech translation projects using different approaches for speech recognition, translation and text to speech synthesis highlighting the major pros and cons for the approach being used.

Spoken Content Retrieval—Beyond Cascading Speech Recognition with Text Retrieval

Spoken content retrieval refers to directly indexing and retrieving spoken content based on the audio rather than text descriptions. This potentially eliminates the requirement of producing text descriptions for multimedia content for indexing and retrieval purposes, and is able to precisely locate the exact time the desired information appears in the multimedia. Spoken content retrieval has been very successfully achieved with the basic approach of cascading automatic speech recognition (ASR) with text information retrieval: after the spoken content is transcribed into text or lattice format, a text retrieval engine searches over the ASR output to find desired information. This framework works well when the ASR accuracy is relatively high, but becomes less adequate when more challenging real-world scenarios are considered, since retrieval performance depends heavily on ASR accuracy. This challenge leads to the emergence of another approach to spoken content retrieval: to go beyond the basic framework of cascading ASR with text retrieval in order to have retrieval performances that are less dependent on ASR accuracy. This overview article is intended to provide a thorough overview of the concepts, principles, approaches, and achievements of major technical contributions along this line of investigation. This includes five major directions: 1) Modified ASR for Retrieval Purposes: cascading ASR with text retrieval, but the ASR is modified or optimized for spoken content retrieval purposes; 2) Exploiting the Information not present in ASR outputs: to try to utilize the information in speech signals inevitably lost when transcribed into phonemes and words; 3) Directly Matching at the Acoustic Level without ASR: for spoken queries, the signals can be directly matched at the acoustic level, rather than at the phoneme or word levels, bypassing all ASR issues; 4) Semantic Retrieval of Spoken Content: trying to retrieve spoken content that is semantically related to the query, but not necessarily including the query terms themselves; 5) Interactive Retrieval and Efficient Presentation of the Retrieved Objects: with efficient presentation of the retrieved objects, an interactive retrieval process incorporating user actions may produce better retrieval results and user experiences.

Bounded cepstral marginalization of missing data for robust speech recognition

Spectral imputation and classifier modification can be counted as the two main missing data approaches for robust automatic speech recognition (ASR). Despite their potentials, little attention has been paid to the classifier modification techniques. In this paper, we show that transferring bounded marginalization, which is a classifier modification method, from spectral to cepstral domain would be beneficial for robust ASR. We also propose improved solutions on this transfer toward a better performance. Two such techniques are presented. The first approach still does not need training of any extra model. It benefits from an observed characteristic of cepstral features and raises accuracy of previously proposed method to a comparable level with that of a classic imputation method. The second technique combines our originally proposed method with an imputation technique but replaces spectral reconstruction with a simpler and faster possible range estimation of missing components. We show that the resulting method improves the accuracies of either of the two combined methods. The proposed techniques also show good robustness when implemented with an inaccurate spectrographic mask.

Computer-Aided Training for Quranic Recitation

Computer Aided Language Learning (CALL) systems have gained popularity due to the flexibility they provide in empowering students to practice their language skills at their own pace. Detection/Correction of specific pronunciation error is an important component of an effective language learning system. Learning the correct rules of the Holy Quran recitation is important to every Muslim. In this work, we developed a Computer Aided Quranic Recitation Training system to detect errors in continuous recitation of Holy Quran and increase the accuracy of the error detection. We have integrated Automatic Speech Recognition (ASR) and classifier-based approach to detect recitation errors. Error detection is done in two successive stages: first, an HMM-based ASR recognizes the recitation, detects the insertion, deletion and substitution of phones and provides phonetic time alignments, and then classifier based approach is used to distinguish between confusing phones to achieve improved detection rate. In this implementation we implemented only two classifiers, one to discriminate between emphasized and non-emphasized utterances of the letter “R” in Arabic, and the other to distinguish between closely related, often confused letter pronunciations. The results show, that the system has achieved a 91.2% word-level accuracy.

Modeling Uyghur Speech Phenomena with Morphological Rules

Due to the morphology of Uyghur, it poses a challenge to state-of-the-art speech recognition systems. This paper describes our work on Uyghur speech recognition technology and a Uyghur voice search application with recent patents on them. Firstly, we introduce the morphology of Uyghur and Uyghur speech phenomena. Secondly, we investigate the use of morphemes in Uyghur auto speech recognition. When speech phenomena happen, variant surface forms of the morpheme are produced. Then, we describe a new approach utilizing the morphological rules to model speech phenomena. In this new approach, variant surface forms are replaced by their corresponding original stems. This creates a better pronunciation lexicon and a more robust language model in Uyghur speech recognition experiments. Finally, we apply the new morphemebased speech recognition approach to a Uyghur voice search application. Experimental results show that this new approach gives the best results in both speech recognition experiments and voice search experiments. Keywords: Morpheme, machine learning, pronunciation variation, uyghur, voice search application.

State-Clustering Based Multiple Deep Neural Networks Modeling Approach for Speech Recognition

The hybrid deep neural network (DNN) and hidden Markov model (HMM) has recently achieved dramatic performance gains in automatic speech recognition (ASR). The DNN-based acoustic model is very powerful but its learning process is extremely time-consuming. In this paper, we propose a novel DNN-based acoustic modeling framework for speech recognition, where the posterior probabilities of HMM states are computed from multiple DNNs (mDNN), instead of a single large DNN, for the purpose of parallel training towards faster turnaround. In the proposed mDNN method all tied HMM states are first grouped into several disjoint clusters based on data-driven methods. Next, several hierarchically structured DNNs are trained separately in parallel for these clusters using multiple computing units (e.g. GPUs). In decoding, the posterior probabilities of HMM states can be calculated by combining outputs from multiple DNNs. In this work, we have shown that the training procedure of the mDNN under popular criteria, including both frame-level cross-entropy and sequence-level discriminative training, can be parallelized efficiently to yield significant speedup. The training speedup is mainly attributed to the fact that multiple DNNs are parallelized over multiple GPUs and each DNN is smaller in size and trained by only a subset of training data. We have evaluated the proposed mDNN method on a 64-hour Mandarin transcription task and the 320-hour Switchboard task. Compared to the conventional DNN, a 4-cluster mDNN model with similar size can yield comparable recognition performance in Switchboard (only about 2% performance degradation) with a greater than 7 times speed improvement in CE training and a 2.9 times improvement in sequence training, when 4 GPUs are used.

A wavelet-based thresholding approach to reconstructing unreliable spectrogram components

Data imputation approaches for robust automatic speech recognition reconstruct noise corrupted spectral information by exploiting prior knowledge of the relationship between target speech and background through the use of spectrographic masks. Most of these approaches are model-based techniques that can only provide accurate estimates of the underlying clean speech when the characteristics of the noise corrupted features do not deviate from those of the model. Discrete wavelet transform (DWT) based de-noising methods can also be used for re-estimating the underlying clean speech from a noise corrupted signal, but often require that the background noise is stationary and modeled by a Gaussian distribution. A novel approach is presented here for incorporating the information derived from spectrographic masks in a DWT-based de-noising method. The spectrographic masks are used for deriving thresholds for de-noising wavelet domain coefficients making DWT based de-noising more suitable for non-stationary noise conditions. The results of an experimental study are presented to demonstrate the performance of DWT based data imputation relative to other established techniques on the Aurora 2 noisy speech recognition task. It will be shown that the proposed approach reduces the impact of model mismatch associated with parametric approaches and exploits the robustness of non-parametric wavelet de-noising approach.

Development of Multimodal Biometric Framework for Smartphone Authentication System

Secret knowledge, like remembering password or token based authentication systems are deemed inconvenience and difficult to use for users such as password may be forgotten or token may be lost. So burdens like remembering password and stolen or forged token based authentication have raised a current trend of biometric authentication system. Now in this current tech world, everyone needs security everywhere to protect our personal gadgets. Now-a-days smartphone gradually becoming a vital tool to manipulate enormous applications which were being done in a computer before. So to keep it secured, biometric based approach can be applied for better convenience and ease of use for the user. In this paper, a novel hybrid multimodal approach for ear recognition and speech recognition has been presented for better robustness and efficiency which can be applied in various fields of applications like authentication in banking transactions. Here two techniques DWT (using haar wavelet) and GLCM have hybridized to extract both shape and texture information from ear images. Again MFCC technique has applied to extract features from speech signals. Afterwards fusion is applied to mix both of those ear and speech features. Those features can be easily and efficiently manipulated by applying Euclidean distance and Bhattacharya distance as the similarity or dissimilarity measures. This proposed approach is very convenient and simple to use, thereby its ease of use allows very fast feature extraction. General Terms Image Processing, Biometrics, Security, Authentication

Topic identification techniques applied to dynamic language model adaptation for automatic speech recognition

In this paper we present an efficient speech recognition approach for multitopic speech by combining information retrieval techniques and topic-based language modeling. Information retrieval based techniques, such as topic identification by means of Latent Semantic Analysis, are used to identify the topic in a recognized transcription of an audio segment. According to the confidence on the topics that have been identified, we propose a dynamic language model adaptation in order to improve the recognition performance in ‘a two stages’ automatic speech recognition system. The scheme used for the adaptation of the language model is a linear interpolation between a background general LM and a topic dependent LM. We have studied different approaches to generate the topic dependent LM and also for determining the interpolation weight of this model with the background model. In one of these approaches we use the given topic labels in the training dataset to obtain the topic models. In the other approach we separate the documents in the training dataset into topic clusters by using the k-means algorithm. For strengthening the adaptation models we also use topic identification techniques to group non topic-labeled documents from the EUROPARL text database in order to increase the amount of data for training specific topic based language models. For the evaluation of the proposed system we are using the Spanish partition of the European Parliament Plenary Sessions (EPPS) Database; we selected a subset of the database with 67 labeled topics for the evaluation. For the task of topic identification our experiments show a relative reduction in topic identification error of 44.94% when compared to the baseline method, the Generalized Vector Model with a classic TF–IDF weighting scheme. For the task of dynamic adaptation of LMs applied to ASR we have achieved a relative reduction in WER of 13.52% over a single background language model.

Arabic Audio News Retrieval System Using Dependent Speaker Mode, Mel Frequency Cepstral Coefficient and Dynamic Time Warping Techniques

Recently, audio data has increasingly becomes one of the prevalent source of information, especially after the exponential growth of using Internet, digital libraries systems and digital mobile devices. The currently massive amount of audio data stimulates working on developing custom audio retrieval tools to facilitate the audio retrieval tasks. The most familiar audio retrieval systems are based on searching using keyword, title or authors. This study presents the feasibility of using MEL Frequency Cepstral Coefficients (MFCCs) to extract features and Dynamic Time Warping (DTW) to compare the test patterns for Arabic audio news. The study proposes and implements architecture for content based audio retrieval system that is dedicated for the Arabic Audio News. The proposed architecture (ARANEWS) utilizes automatic speech recognition for isolated Arabic keyword speech mode; template based automatic speech recognition approach, MFCCs and DTW. ARANEWS presents a style of retrieval system that based on modeling signal waves and measuring the similarity between features that are extracted from spoken queries and spoken keywords. One of the major components that compose ARANEWS system is feature Database (ARANEWSDB). ARANEWSDB stores the extracted features (MFCCs) from the spoken keywords that are prepared to retrieve Arabic audio news. ARANEWS supports using Query by Humming (QBH) and Query by Example (QBE) instead of using query by text.

Low SNR Speech Recognition using SMKL

While traditional speech recognition methods have achieved great success in a number of real word applications, their further applications to some difficult situations, such as Signal-to-Noise Ratio (SNR) signal and local languages, are still limited by their shortcomings in adaption ability. In particular, their robustness to pronunciation level noise is not satisfied enough. To overcome these limitations, in this paper, we propose a novel speech recognition approach for low signal-to-noise ratio signal. The general steps for our speech recognition approach are composed of signal preprocessing, feature extraction and recognition with simple multiple kernel learning (SMKL) method. Then the application of SMKL in speech recognition with low SNR is presented. We evaluate the proposed approach over a standard data set. The experimental results show that the performance of SMKL method for low SNR speech recognition is significantly higher than that of the method based on other popular approaches. Further, SMKL based method can be straightforwardly applied to recognition problem of large scale dataset, high dimension data, and a large amount of isomerism information.

Auditory processing-based features for improving speech recognition in adverse acoustic conditions

The paper describes an auditory processing-based feature extraction strategy for robust speech recognition in environments, where conventional automatic speech recognition (ASR) approaches are not successful. It incorporates a combination of gammatone filtering, modulation spectrum and non-linearity for feature extraction in the recognition chain to improve robustness, more specifically the ASR in adverse acoustic conditions. The experimental results with standard Aurora-4 large vocabulary evaluation task revealed that the proposed features provide reliable and considerable improvement in terms of robustness in different noise conditions and are comparable to those of standard feature extraction techniques.

N-Best Re-scoring Approaches for Mandarin Speech Recognition

The predominant language model for speech recognition is n-gram language model, which is locally learned and usually lacks global linguistic information such as long-distance syntactic constraints. We first explore two n-best re-scoring approaches for Mandarin speech recognition to overcome this problem. The first approach is linear re-scoring that can combine several language models from various perspectives. The weights of these models are optimized using minimum error rate learning method. Discriminative approach can also be used for re-scoring with rich syntactic features. To overcome the speech text insufficiency problem for discriminative model, we propose a domain adaptation method that trains the model using Chinese pinyin-to-character conversion dataset. Then we present a cascaded approach to combine the two re-scoring models in pipeline that takes the probability output of linear re-scoring model as the initial weight of the discriminative model. Experimental results show that both re-scoring approaches outperform the baseline system, and the cascaded approach achieves the best performance.

Structured Sparsity Models for Reverberant Speech Separation

We tackle the speech separation problem through modeling the acoustics of the reverberant chambers. Our approach exploits structured sparsity models to perform speech recovery and room acoustic modeling from recordings of concurrent unknown sources. The speakers are assumed to lie on a two-dimensional plane and the multipath channel is characterized using the image model. We propose an algorithm for room geometry estimation relying on localization of the early images of the speakers by sparse approximation of the spatial spectrum of the virtual sources in a free-space model. The images are then clustered exploiting the low-rank structure of the spectro-temporal components belonging to each source. This enables us to identify the early support of the room impulse response function and its unique map to the room geometry. To further tackle the ambiguity of the reflection ratios, we propose a novel formulation of the reverberation model and estimate the absorption coefficients through a convex optimization exploiting joint sparsity model formulated upon spatio-spectral sparsity of concurrent speech representation. The acoustic parameters are then incorporated for separating individual speech signals through either structured sparse recovery or inverse filtering the acoustic channels. The experiments conducted on real data recordings of spatially stationary sources demonstrate the effectiveness of the proposed approach for speech separation and recognition.

Probabilistic speech feature extraction with context-sensitive Bottleneck neural networks

We introduce a novel context-sensitive feature extraction approach for spontaneous speech recognition. As bidirectional Long Short-Term Memory (BLSTM) networks are known to enable improved phoneme recognition accuracies by incorporating long-range contextual information into speech decoding, we integrate the BLSTM principle into a Tandem front-end for probabilistic feature extraction. Unlike the previously proposed approaches which exploit BLSTM modeling by generating a discrete phoneme prediction feature, our feature extractor merges continuous high-level probabilistic BLSTM features with low-level features. By combining BLSTM modeling and Bottleneck (BN) feature generation, we propose a novel front-end that allows us to produce context-sensitive probabilistic feature vectors of arbitrary size, independent of the network training targets. Evaluations on challenging spontaneous, conversational speech recognition tasks show that this concept prevails over recently published architectures for feature-level context modeling.

Improved Bimodal Speech Recognition Study Based on Product Hidden Markov Model

Recent years have been higher demands for automatic speech recognition (ASR) systems that are able to operate robustly in an acoustically noisy environment. This paper proposes an improved product hidden markov model (HMM) used for bimodal speech recognition. A two-dimensional training model is built based on dependently trained audio-HMM and visual-HMM, reflecting the asynchronous characteristics of the audio and video streams. A weight coefficient is introduced to adjust the weight of the video and audio streams automatically according to differences in the noise environment. Experimental results show that compared with other bimodal speech recognition approaches, this approach obtains better speech recognition performance.

Comparison of Linear Discriminant Analysis Approaches in Automatic Speech Recognition

Speech recognition systems are commonly modelled by hidden Markov models with Gaussian mixture models as observation density functions. These models have a significant number of parameters, which usually leads to the problem of data sparsity, especially for under-resourced languages such as Serbian. One of the ways to overcome the problem of data sparsity is the reduction of the number of features. Linear discriminant analysis (LDA) and heteroscedastic LDA (HLDA) are two common ways to reduce the dimensionality in an automatic speech recognition task. The paper compares the properties of speech recognition systems for Serbian in which both techniques are applied with variable types of input features as well as the number of output features of (H)LDA. The best results are obtained in the case of HLDA with input vectors consisting of concatenations of feature vectors across 7 successive frames, where each feature vector contains 12 mel frequency cepstral coefficients (MFCCs) and normalized energy, and the number of output features is 32 or 35.DOI: http://dx.doi.org/10.5755/j01.eee.19.7.5167

Visual Features for Lithuanian Phone Recognition

Paper presents visual features for speech recognition. Visual speech recognition can be applied as support for acoustical recognition process or as stand-alone speech recognition approach in the case of absence of acoustical data. Geometrical features of speaker‘s lips are presented. Statistical analysis of feature values enabled us to select individual feature for each analyzed phone and one combined feature for all phones. The discriminative power was tested using Bayes classification rule. 4th order analysis enabled us to recognize phones with the average rate of 60.3 %, the highest recognition rate of 81.3 % was achieved identifying phone [o]. DOI: http://dx.doi.org/10.5755/j01.eee.19.5.1724

A perceptual masking approach for noise robust speech recognition

This article describes a modified technique for enhancing noisy speech to improve automatic speech recognition (ASR) performance. The proposed approach improves the widely used spectral subtraction which inherently suffers from the associated musical noise effects. Through a psychoacoustic masking and critical band variance normalization technique, the artifacts produced by spectral subtraction are minimized for improving the ASR accuracy. The popular advanced ETSI-2 front end is tested for comparison purposes. The performed speech recognition evaluations on the noisy standard AURORA-2 tasks show enhanced performance for all noise conditions.