A comparison of several generalized bandwidth compression methods, operating on the output of a high quality channel vocoder is presented. The methods are based on an -entropy- minimizing eigenfilter technique, applied to both a frequency spectrum, energy oriented representation, and to a purely abstract, binary description of the vocoder output. Computer generated graphic representions of resynthesized speech, based on different levels of compression of the minimum entropy data base, are compared with the standard frequency spectrum representations, and audio tapes of the resynthesized speech are presented for judgment of intelligibility and quality.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.