Year: 2020
Type: article
Abstract: Speech is an important essential function of human beings.Humans strive to recover data from the external world or to communicate with everyone.For many reasons, speech protrudes as the resourceful and suitable.Speech signal constitutes both vowels and consonants and consonants are more susceptible towards the noise than vowels as they have less energy.Hence when speech signal is communicated it is necessary to reduce the noise and this process is known as speech enhancement.In noisy environment listening task is very difficult at the destination, so reduction of noise is important research field in speech enhancement.In speech enhancement various filter banks are used to reduce the variety of noises.The proposed methods give better quality of speech and a minimized musical noise and background noises.Auditory filter bank, QMF banks, Warped Low Delay Filter-Banks (WLDFB), Unimodular Filter Banks (UMFB), Cascaded filter Banks, DFT modulated Filter Banks (DFTMFB) are the filter bank algorithms which very effective in noise reduction even in low delay and low SNR situations.
Author: Jagadish S.Jakati
Language: English
Topic: Speech and Audio Processing
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
Funder: Basaveshwar Engineering College
OpenAlex ID: W3085706397