كبس ملفات الكلام وتشفيرها باستخدام الشبكة العصبية ذات الانتشار العكسي

Abstract

Speech compression is a very important field of internet application or to
transfer through network and telecommunication. This research studies
speech compression using neural network, which is commonly used with
images files compression. It has been focused on the files that include
speech only and do not contain other sounds like music or car sound, or
animal sound etc. In this research it has been used Back Propagation Neural
Network BPNN for speech signals compression. Several experiments have
been performed which differ in configuration of data that entered to the
network. The highest compression ratio obtained is (1:10) from the original
data. The compression files obtained represent as encipher files because they
have little data with compared to original file, and also they night be not
useful if it has been stolen through transmitting operation on the network,
they be able to be decompressed to the original without their weights matrix
can not, so the promising benefit from this compression is to achieve double
aims. Speech security is an important goal for users of many speech
communication systems. To obtain a desired level of security an encryption
scheme should be added for speech signal before transmission.