|
The Speaker Recognition Based on Linear Predictive Coding and Adaptive Algortihms.
In this study, feature extraction method by way of adaptive approximation and feature extraction method by way of high level statistics method are suggested for speaker identification. Each Coefficients of the Linear Predictive Coding and Adaptive Algorithm has been computed for each participants in order to use in ‘speaker recognition process’ and then a database has been created with these related data. Then, the LPC coefficients of one participant have been computed with intended method and have been compared with related coefficients, which had been stored in the database, by the method of Euclid Distance. Thus, personal description process has been accomplished. Also, the probability of success has been calculated for every single method.
In the General Parts of this study, the definitions of voice, speech and speaker have been given. The methods of speaker recognition and the concepts of open and closed sets have been examined. In the part of Material and Method, linear predictive and adaptive methods, which had been used in the analysis, have been scrutinized.
In the Evidences Part, application of speaker recognition has been indicated. Obtaining of speaker recognition coefficients, comparing of its results and success ratio of performed applications have been evaluated in the part of Argument and Result.
TOMBAZ Sibel
Danışman : Prof. Dr. Hakan Ali ÇIRPAN
Anabilim Dalı : Elektrik-Elektronik Mühendisliği
Mezuniyet Yılı : 2009
Tez Savunma Jürisi : Prof. Dr. Hakan Ali ÇIRPAN
Prof. Dr. Aydın AKAN
Prof. Dr. Sıddık YARMAN
Prof. Dr. Ahmet SERTBAŞ
Doç Dr. Halim ZAİM
Dostları ilə paylaş: | 
