Mutual Information Approach for Predicting Speech Intelligibility |
Author(s): |
| Ranjitha A , Department of DECS, VTU-RC Mysore; Ramyashree M, Department of DECS, VTU-RC Mysore |
Keywords: |
| Intelligibility, prediction entropy and mutual information |
Abstract |
|
The proposed intelligibility prediction model makes use of fundamental principle theoretic tools like entropy and mutual information. It emerges as natural to employ tools developed to characterize data or information transmission. Subsequent to all, the speech communication process be able to view as the method of transmitting a speech signal across a dynamic, transient channel (acoustic channel, auditory periphery and higher stages of the auditory pathway) to arrive at the brain of the receiver. The expression for the AI demonstrates well-built resemblance to the expression for the ability of a memory less Gaussian channel. The fundamental thought of the proposed technique is to compare the critical band amplitude envelopes of the clean and noisy or processed signal for estimating the intelligibility of the noisy or processed signal. |
Other Details |
|
Paper ID: IJSRDV3I40494 Published in: Volume : 3, Issue : 4 Publication Date: 01/07/2015 Page(s): 874-878 |
Article Preview |
|
|
|
|
