Master of Computer Science Project Report

Where To Find Additional Information

Introduction

Digital Sound Compression through the Identification of Symmetrical Components uses simple measurements of the shape of a segment of the input data to characterize the segment. The mesurements of the shape of the data segment are termed its “symmetry”. The SC algorithm segments the audio data into pieces of data that have a measurable period. These segments are called “wavelets”. If the wavelet has similar characteristics to another wavelet that was previously identified, the wavelet is represented as the token of the previously identified wavelet.

A wavelet is characterized using simple approximations of its symmetry or shape. The wavelet is compared with a list of other wavelets that have been characterized as unique. If the wavelet is dissimilar to other wavelets in the list, the wavelet is saved in the wavelet list. The list of wavelets that are considered to be unique to the audio file is called the “component” list. The position where the wavelet occurs in the audio file is also noted and added to a “pattern” list.

Goals of the SC Algorithm:

· Computationally Simple · Encoder and Decoder work in real-time.

· Will Work on all types of Audio Data

· Works best on Voice Files

· Possible candidate for Internet phone or voice e-mail applications