|
|
|
|
|
|
Title
|
Evaluation of Feature Recombination Technique...
 |
|
|
|
Abstract |
|
Given a speech signal, there are two most important information that can be extracted from it. One being the linguistic information (about what is being said) and other being the speaker specific information (about who is speaking). This report is about the task of speaker recognition where the goal is to determine the speaker identity, from a group of known speaker, which closely matches with input sample. This problem become even more tough when there is limited amount of test and train data, a mismatch between the surrounding conditions while recording the test and train data, or in noisy environments.In this thesis, we consider the problem of speaker identification in noisy and bandlimited telephonic environments using the Gaussian Mixture Model approach combined with sub-band based feature extraction. We implement a sub-band based Posteriori Union Model described by Reynolds. Then, we extend sub-band based approach to combine the phase based feature ModGDF and Magnitude based feature MFCC using several feature recombination techniques described in this thesis. These sub-band based feature recombination methods gives 46% identification accuracy, in best case, on NTIMIT database with little or no increase in computation. |
| |
|
| |
|
|
|
|
|
Title
|
|
Evaluation of phase and magnitude based featu...
|
|
|
|
Abstract |
|
Given a speech signal, there are two most important information that can be extracted from it. One being the linguistic information (about what is being said) and other being the speaker specific information (about who is speaking). This report is about the task of speaker recognition where the goal is to determine the speaker identity, from a group of known speaker, which closely matches with input sample. This problem become even more tough when there is limited amount of test and train data, a mismatch between the surrounding conditions while recording the test and train data, or in noisy environments.In this thesis, we consider the problem of speaker identification in noisy and bandlimited telephonic environments using the Gaussian Mixture Model approach combined with sub-band based feature extraction. We implement a sub-band based Posteriori Union Model described by Reynolds. Then, we extend sub-band based approach to combine the phase based feature ModGDF and Magnitude based feature MFCC using several feature recombination techniques described in this thesis. These sub-band based feature recombination methods gives 46% identification accuracy, in best case, on NTIMIT database with little or no increase in computation. |
| |
|
| |
|
|
|
|
|
Title
|
|
Blind De-convolution of BPSK and QPSK modulat...
|
|
|
|
Abstract |
|
Signal processing is the most important part of most of the electronics and software system designed for electronics. Signal processing plays a very important role in most of the wireless communication, image processing and Speech processing problems. Most of the applications of wireless communication involve transmission of data accurately from source to destination. Most often, the transmitted signal suffers from noise addition, interference and other medium of transmission (also known as channel) factors. In most cases, we can approximate the channel by a Linear Time-Invariant (LTI) system. Thus, a signal can be assumed to pass through a LTI system with impulse response of channel before reaching destination.At the receiver end, we need to perform some kind of post-processing to nullify the effect of channel LTI filter. So, the problem reduces to find the input given the output and channel LTI filter. However, in most of the cases, channel LTI filter is not known before hand. So, the problem of finding input becomes more difficult. There have been a lot of algorithms introduced in literature which try to find LTI filter based on knowledge of input and output signals. Since, the whole motive of the problem is to find the input s... |
| |
|
| |
|
|
|
|
|
Title
|
|
Adaptive Equalization Techniques using Recurs...
|
|
|
|
Abstract |
|
In this project, we extend the use of methods of least squares to find a recursive algorithm solution of adaptive transversal filter. Given the LS solution at any time instant n-1, we find the solution at time n recursively using past solution and newly arrived data. This algorithm is known as Recursive Least Squares (RLS) algorithm. We show the convergence rate of RLS algorithm is faster than LMS algorithm by comparing the learning curves of two algorithms for specified channel response. |
| |
|
| |
|
|
|
|
|
Title
|
|
Adaptive Equalization Techniques using Least ...
|
|
|
|
Abstract |
|
Adaptive equalization is the technique used to reliably transmit data through a communication channel. Ideally, if the channel is ideal (without and channel distortion and additive noise), we can demodulate the signal perfectly at the output without causing any error. However, in practice, all the channels are non-ideal and noisy in nature. So, to recover the original signal after demodulation, our aim is to find an equalization filter which will minimize the error between original transmitted signal and demodulated signal passed through equalization filter. Several algorithms like Least Mean Square (LMS), Recursive Least Mean Square (RLMS), Normalized Least Mean Square (NLMS) etc., has been proposed to perform this operation of equalization. In this project, we study the adaptive equalization technique with the use of least mean Square algorithm. |
| |
|
| |
|
|
rolex replica
|
Go to document page