A proposed system for sound retrieval using MAS and ANN

Abstract EN

As the use of sounds for computer interfaces, electronic equipment and multimedia contents, has increased, the role of sound design tools has become more important.

In sound retrieval, picking one sound out from huge data is troublesome for users because of the difficulty of simultaneously listening to plural sounds and sometimes there are difficulties with speech and sound recognition.

Consequently, an efficient retrieval method is required for sound databases.

This research proposes a system aim to deal with sound retrieval in both two cases : authenticity and normal.

In the first case, authenticity, two algorithms has been develop one for building the authentication database and the second deal with user sound sample to retrieve the matched authenticated samples.

In the second case normal we develop algorithm to deal with user sound sample to retrieve all the matched samples.

Many techniques used in this proposed system such as Artificial Neural Network (ANN), Data Encryption Standard (DES), Multi Agent System (MAS) and Fourier transformation (FT).

Using these combinations of advanced and adaptive techniques supports the system to be reliable, secure and parallel.