Speech compression & decompression through DSP

• Main Authors: dela Cruz, Linver R., Ng, Fritz Gerald L., Panganiban, Leonardo M., Rayos, Michae D., Rosales, Reginald
• Format: text
• Language: English
• Published: Animo Repository 1996
• Subjects: Speech processing systems; Signal processing > Digital techniques; Coding theory; Digital electronics; Engineering
• Online Access: https://animorepository.dlsu.edu.ph/etd_bachelors/8959
• Institution: De La Salle University

The main purpose of this project is to implement a speech compression algorithm using a digital signal processor. The algorithm that will be developed is the efficient compression and storage of speech signals in real-time. This will be implemented through the use of three components: a host computer, a software program and the digital signal processor. The appropriate signal processing algorithm will be programmed by the group and a host computer will be used as a medium to send and load the necessary instructions of the program to the digital signal processor. The digital signal processor will be the system that will perform the actual signal processing, specifically the encoding of signals, compression, decompression, and the decoding of the decompressed signals. Signal processing will be done by first acquiring speech signals using a microphone which is directly connected to a pre-amplifier. These signals will then be encoded, compressed accordingly in the DSP kit and will be stored into the DSP on-chip RAM. The project will be very substantial for many signal-related applications since most of modern signal related technologies require efficient signal compression. Some of its common applications are computer-based training (CBT), sound editing and recording, speech recognition and voice mail. Computer-based training is a means of acquiring knowledge with the use of computers through recorded topics or lectures whereas sound editing and recording is popularly used by most recording studios and companies to enhance music reproduction. Speech recognition also requires compression to increase its number of templates for signal comparison and it is needed in voice mail to efficiently store as many number of messages as possible.