Abstract
In this paper we have illustrated the
properties of Discrete Cosine Transform (DCT) as
compared to the standard Discrete Fourier Transform
(DFT) in case of noise removal from the speech. The
results shows that DCT has better energy compaction
and less computations as compared to DFT. The
implementation and results of the DCT based minimum
mean square error filter are described. The proposed
algorithm is implemented for the reduction of residual
noise by using probability of speech absence technique.
The proposed techniques use adaptive schemes which
will track the probability of speech absence in a noisy
speech. It estimates the received spectral amplitude by a
binary classification i.e. either the speech is present or
absent state.
Muhammad Safder Shafi, Mansoor Khan, Naveed Abdul Sattar, Muhammad Rizwan, Affan Aziz Baba, Rizwan Ghani. (2012) Transform Based Speech Enhancement Using DCT Based MMSE Filter,& Its Comparison With DFT Filter, Journal of Space Technology , Volume 2, Issue 1.
-
Views
592 -
Downloads
51
