Text-dependent Speaker Recognition System Research Articles

Text Dependent Speaker Recognition using MFCC features and BPANN

Mel-Frequency Cepstral Coefficients are spectral feature which are widely used for speaker recognition and text dependent speaker recognition systems are the most accurate in voice based authentication systems. In this paper, a text dependent speaker recognition method is developed. MFCCs are computed for a selected sentence. The first 13 MFCCs are considered for each frames of duration 26ms and each coefficient is clustered to a 5 element cluster centres and finally to a form a 65 element speech code vector for the entire speech. The speech code is trained using a multi-layer perceptron backpropagation gradient descent network and the network is tested for various test patterns. The performance is measured using FAR, FRR and EER parameters. The recognition rate achieved is 96.18% for a cluster size of 5 in each coefficient. General Terms Speaker Recognition, Artificial Neural Networks, Clustering

Development of a Text-Dependent Speaker Recognition System

Speaker recognition is the ability of recognizing a person based on his voice. Many modalities and techniques have been applied to achieve the task of authentication, ranging from retina scans to finger prints. This paper presents a TextDependent model for automatic Speaker Recognition. Features extractor are based on Mel-Frequency Cepstrum Coefficients (MFCCs) and Linear Predictive Coding (LPC) analysis technique which aids the translation of incoming speech into a feature value. Also, a recognizer block, employs the two techniques (MFCC and LPC) to get an hybrid features for speaker identification/verification system. An experiment was carried out on the threshold from 75% to 95% at 5% differential to know the performance of the MFCCs and LPC identification/verification system. From the experiment carried out, the result shows that LPC outperforms MFCCs and a combination. The text-dependent speaker recognition system was implemented using Java Programming Language (Java Speech Application Programming Interface (JSAPI) ). The system is developed for access control into computer systems and could be used for access control where security is considered to be of utmost important.

An Educational Text-Dependent Speaker Recognition System

Speech is one of the most natural forms of communication. Recently there have been many developments and research in the field of speech processing. Speaker recognition, an important application of speech processing, is the process of determining the identity of a speaker by comparing speech produced by a speaker to a number of speech samples stored in a database. This paper describes the development of an educational text-dependent speaker recognition system. Speech processing systems are highly complex and teaching students this subject with the underlying technologies can be a challenging task. The aim of this work was to give a hands-on-experience to students via the development of a speaker recognition system that can be used in laboratory sessions. The developed system is based on MATLAB and can easily be used to teach the basic principles of speaker recognition, including the effects of noise on the recognition process.

Research of Vowel Mapping Theory for Speaker Identification of Chinese Mandarin

This paper presents a novel framework of speaker identification based on Chinese vowel mapping theory. The base of this framework is the mapping from Chinese multi-phthong to monophthong phonemes. According to the contrast of spectrum, features, monophthong phoneme glide statistical distribution and the performance of vowel classification, we confirmed that Chinese vowels could be decomposed into several monophthong phonemes based on the short time characteristics. Then we developed a new mapping table from multi-phthong to monophthong phoneme to assist the latter research of a great deal of experiment and theory analysis. The novel framework added a special model to implement the decomposition and organized several monophthong classifiers to replace the traditional classification module, which can avoid the disturbance of semantic information and achieve higher performance. In this new framework, it adopts short time frame as the basic identify unit, which makes it more compatible to real time system. The sufficient theory analysis and experimental results showed that the presented model and algorithms based on novel framework have achieved higher accuracy, speed and enhanced the robustness in different conditions compared with many traditional methods. Especially, we succeed in separating personal identification information from semantic information based on classifying the Chinese vowel, which will be a new way to transform the text-independent system into text-dependent speaker recognition system.