S. Jothilakshmi

Speaker diarization using autoassociative neural networks

This paper addresses a new approach to speaker diarization using autoassociative neural networks (AANN). The speaker diarization task consists of segmenting a conversation into homogeneous segments which are then clustered into speaker classes. The proposed method uses AANN models to capture the speaker specific information from mel frequency cepstral coefficients (MFCC). The distribution capturing ability of the AANN model is utilized for segmenting the conversation and grouping each segment into one of the speaker classes. The algorithm has been tested on different databases, and the results are compared with the existing algorithms. The experimental results show that the proposed approach competes with the standard speaker diarization methods reported in the literature and it is an alternative method to the existing speaker diarization methods.

Unsupervised speaker segmentation with residual phase and MFCC features

This paper proposes an unsupervised method for improving the automatic speaker segmentation performance by combining the evidence from residual phase (RP) and mel frequency cepstral coefficients (MFCC). This method demonstrates the complementary nature of speaker specific information present in the residual phase in comparison with the information present in the conventional MFCC. Moreover this method presents an unsupervised speaker segmentation algorithm based on support vector machine (SVM). The experiments show that the combination of residual phase and MFCC helps to identify more robustly the transitions among speakers.

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.

Automatic system to detect the type of voice pathology

Abstract Acoustic analysis is a noninvasive technique based on the digital processing of the speech signal. Acoustic analysis based techniques are an effective tool to support vocal and voice disease screening and especially in their early detection and diagnosis. Modern lifestyle has increased the risk of pathological voice problems. This work focuses on a robust, rapid and accurate system for automatic detection of normal and pathological speech and also to detect the type of pathology. This system employs non-invasive, inexpensive and fully automated measures of vocal tract characteristics and excitation information. Mel-frequency cepstral coefficients and linear prediction cepstral coefficients are used as acoustic features. The system uses Gaussian mixture model and hidden Markov model classifiers. Cerebral palsy, dysarthria, hearing impairments, laryngectomy, mental retardation, left side paralysis, quadriparesis, stammering, stroke, tumour in vocal tract are the types of pathologies considered in our experiments. From the experimental results, it is observed that to classify normal and pathological voice hidden Markov model with mel frequency cepstral coefficients with delta and acceleration coefficients is giving 94.44% efficiency. Likewise to identify the type of pathology Gaussian mixture model with mel frequency cepstral coefficients with delta and acceleration coefficients is giving 95.74% efficiency.

Segmentation of Continuous Speech into Consonant and Vowel Units using Formant Frequencies

This paper addresses the issues in segmentation of continuous speech into sub-word units of speech using Formants and support vector machines (SVMs). Many studies have been conducted to identify and discriminate vowels and consonants using acoustic/articulatory differences. In this study the continuous speech is segmented into smaller speech units and each unit is classified either consonant or vowel using the Formant frequencies. This process when further combined with recognition of each unit will form a complete speech recognition system. The proposed detection strategy is tested with the speech signals recorded from the television broadcast.

Unsupervised Speaker Segmentation using Autoassociative Neural Network

In this paper we propose an unsupervised approach to speaker segmentation using autoassociative neural network (AANN). Speaker segmentation aims at finding speaker change points in a speech signal which is an important preprocessing step to audio indexing, spoken document retrieval and multi speaker diarization. The method extracts the speaker specific information from the Mel frequency cepstral coefficients (MFCC). The speaker change points are detected using the distribution capturing ability of the AANN model. Experiments are carried out on different audio databases, and the method is capable of detecting speaker changes with short duration of speech in an unsupervised manner.

A novel spoken keyword spotting system using support vector machine

Spoken keyword spotting is crucial to classify expertly a lot of hours of audio stuffing such as meetings and radio news. These systems are technologically advanced with the purpose of indexing huge audio databases or of differentiating keywords in uninterrupted speech streams. The proposed work involves sliding a frame-based keyword template along the speech signal and using support vector machine (SVM) misclassification rates obtained from the hyperplane of two classes efficiently search for a match. This work framed a novel spoken keyword detection algorithm. The experimental results show that the proposed approach competes with the keyword detection methods described in the literature and it is an alternative technique to the prevailing keyword detection approaches.

Tamil Sign Language to Speech Translation

Sign Language Recognition is one of the most growing fields of research today. Most researches on hand gesture recognition for HCI rely on either Artificial Neural Networks (ANN) or Hidden Markov Model (HMM). There are many effective algorithms for segmentation, classification, pattern matching and recognition. The main goal of this paper is to compare the classifiers for translating Tamil sign language to speech, which will definitely help the researchers to attain an optimal solution. The most important thing in hand gesture recognition system is the input features and the selection of classifiers. To increase the recognition rate and make the recognition system resilient to view-point variations, the concept of shape descriptors from the available feature set is introduced. K-Nearest Neighbor (KNN), Proximal Support Vector Machine (PSVM) and Naive Bayesian are used as classifiers to recognize static Tamil words. The performance analysis of the proposed approach is presented along with the experimental results. Comparative analysis of these methods with other popular techniques shows that the real time efficiency and robustness are better. Experimental results demonstrate the effectiveness of the proposed work for recognizing efficiency 78% for KNN classifier, 91% for PSVM classifier and 93% for Naive Bayesian classifier.

An Efficient Continuous Speech Recognition System for Dravidian Languages Using Support Vector Machine

This paper mainly focuses on developing a novel speech recognition system for Dravidian languages such as Tamil, Malayalam, Telugu, and Kannada. This research work targets to afford a well-organized way for human to interconnect with computers absolutely for people with disabilities who facade variety of stumbling blocks while using computers. This work would be very helpful to the native speakers in various applications. The proposed CSR system comprises of three steps namely preprocessing, feature extraction, and classification. In the preprocessing step, the input signal is preprocessed through the steps such as pre-emphasis filter, framing, windowing, and band stop filtering in order to remove the background noise and to enrich the signal. The best-filtered and the enriched signal from the preprocessing step is taken as the input for the further process of CSR system. The speech features being the most essential segment in speech recognition system. The most powerful and widely used short-term energy (STE) and zero-crossing rate (ZCR) are used for continuous speech segmentation, and Mel-frequency cepstral coefficients (MFCC) and shifted delta cepstrum (SDC) are used for recognition task. Feature vectors are given as the input to the classifier such as support vector machine (SVM) for classifying and recognizing Dravidian language speech. Experiments are carried out with real-time Dravidian speech signals, and the results reveal that the proposed method competes with the existing methods reported in literature.

A GMM-BASED HIERARCHICAL AUTOMATIC LANGUAGE IDENTIFICATION SYSTEM FOR INDIAN LANGUAGES

Automatic spoken language identification (LID) is the task of identifying a language from a short utterance of speech by an unknown speaker. This article describes a novel two-level identification system for Indian languages using acoustic features. In the first level, the system identifies the family of the spoken language; the second level aims at identifying the particular language within the corresponding family. The proposed system has been modeled using Gaussian mixture model (GMM) and utilizes the following acoustic features: mel frequency cepstral coefficients (MFCC) and shifted delta cepstrum (SDC). A new database has been created for nine Indian languages. It is shown that a GMM-based LID system using MFCC with delta and acceleration coefficients is performing well, with 80.56% accuracy. The performance accuracy of the GMM-based LID system with SDC is also considerable.