Dissertation > Excellent graduate degree dissertation topics show
Speaker Recognition on the Base of Time-Varying Characteristics of Speech Signal
Author: XuLiangJun
Tutor: FeiWanChun
School: Suzhou University
Course: Textile Engineering
Keywords: Time-varying characteristics Characteristic frequency Nonstationarity TVPAR model Speaker Recognition
CLC: TN912.34
Type: Master's thesis
Year: 2010
Downloads: 83
Quote: 0
Read: Download Dissertation
Abstract
|
Speaker recognition is a special kind of speech recognition. In recent years, the rapid development of this technology , and text dependent speaker recognition systems require identity verification in some places has been applied. But there are still some problems to be solved , where the key question is whether to use the speech signal which features describing speak talent is effective and reliable. Speaker recognition , including speaker verification and speaker identification , this paper is the study of the text relating to speaker identification problem . Based on time-varying characteristics of speech signal , the mean MEL cepstrum extracted based on the characteristic frequency variation with time (including the time-varying pitch frequency ), whereby the speech signal by the characteristic frequency of the respective inverted sequence of spectral values ??of the time series . Using time series preprocessing and mathematical statistics method , the separation time series trends and fluctuations in the amount of the amount of random fluctuations . The amount is zero-mean random fluctuations autocovariance nonstationary time series , the use of full- order time-varying parameter autoregressive (Time-Varying Parameter Autoregressive) model, the amount of sequence analysis of stochastic volatility , further extraction speaker speech feature parameters. Fluctuations in the amount of random sequence and with a full order TVPAR models were based on the analysis of speaker recognition . This selects the smallest BIC (Bayesian Information Criterion) law analysis to determine the order of regression model , the last speaker using Mahalanobis distance discriminant . Experimental results show that the model with a full order TVPAR recognition , the recognition rate than the amount of random fluctuations on the sequence recognition rate has improved greatly . In the full -order TVPAR based on the model , take a characteristic frequency of 97.3% recognition rate , two characteristic frequency identification rate of 98.6% .
|
Related Dissertations
- Complex channel speaker recognition technology,TN912.34
- Research on Speaker Recognition System under the Visual C++ 6.0,TN912.34
- Design and Implementation of Voice Screening System for Public Security Based on Speaker Recognition,TN912.34
- Speaker Recognition on the Basis of Time Series Analysis,TN912.34
- The Study of Support Vector Machine and Its Application on Speaker Recognition,TN912.3
- Graph transformation method for solving the time-dependent with time windows Chinese Postman Problem,F616
- Research of Feature Extraction for Voice Based on Wavelet Transform,TN912.34
- Submicron complementary bipolar process of key components,TN432
- EIS Study on the Failure Process of Three Kinds of Composite Coatings and Rapid Detection Methods,TG174.4
- EMD-based speaker recognition,TN912.34
- Research on Text-independence of Open-set Speaker Recognition,TN912.34
- The Development of Speaker Recognition System Based on Support Vector Machine,TN912.34
- Design and Implementation of Speaker Recognition System Based on Windows CE,TN912.34
- Mixing characteristics and Gaussian mixture model - based speaker recognition,TN912.34
- The Study of Hierarchical Speaker Segmentation and Relative Algorithms,TN912.3
- HMM - based speaker identification to improve the research and application,TN912.3
- With the text on Speaker Recognition Method,TN912.3
- Research on Technology of Speaker Recognition Based on VQ and HMM,TN912.34
- Speaker recognition based on support the phasor machine stability key,TN912.34
- Study of Speaker Recognition System,TN912.34
CLC: > Industrial Technology > Radio electronics, telecommunications technology > Communicate > Electro-acoustic technology and speech signal processing > Speech Signal Processing > Speech Recognition and equipment
© 2012 www.DissertationTopic.Net Mobile
|