Speech Compression by Analysis-Synthesis
Speech compression is the process of reducing the channel capacity or bandwidth necessary for speech communication. Any reduction of the bandwidth of a given channel necessarily means a reduction in the information-carrying capacity of the channel. A speech signal which normally occupies a channel of 4000 cycles per second and has a 30-decibel signal-to-noise ratio cannot literally be compressed into a smaller channel. To reduce the bandwidth of the speech signal, the degrees of freedom of the signal must be restricted. Some portions of the signal must be removed. The problem of compressing the speech signal while retaining maximum intelligibility and quality of signal is a problem of selection. That portion of speech must be selected which upon removal will have the minimum effect upon speech quality and intelligibility.
This paper costs $20 for non-members, $5 for AES members and is free for E-Library subscribers.