Browse Prior Art Database

Automatic Correction of Voiced Fricatives

IP.com Disclosure Number: IPCOM000076396D
Original Publication Date: 1972-Feb-01
Included in the Prior Art Database: 2005-Feb-24
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Peckels, JP: AUTHOR

Abstract

In a channel vocoder the dichotomous voiced/voiceless decision lacks possibility for correct definition of voiced fricatives, which often call for simultaneous buzz and hiss type information. A solution to this problem is proposed through an artifact, consisting in interleaving voiced and unvoiced excitation modes.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 100% of the total text.

Page 1 of 1

Automatic Correction of Voiced Fricatives

In a channel vocoder the dichotomous voiced/voiceless decision lacks possibility for correct definition of voiced fricatives, which often call for simultaneous buzz and hiss type information. A solution to this problem is proposed through an artifact, consisting in interleaving voiced and unvoiced excitation modes.

Voiced fricatives like zoo, pleasure, seven, then are nearly always mistreated by a pulse excited channel vocoder. The reasons for this are twofold. Due to its limited dynamic and frequency range - the analyzer often misses the weak energy in the higher part of the spectrum, which is one distinctive feature for these sounds, - the dichotomous voicing/voiceless decision does not fully describe these voiced fricatives, which often need simultaneous buzz and hiss type excitation, as is known from their articulatory and acoustic description.

The first difficulty can be overcome, as it is known, by insertion of high- frequency information into the allophone representing the sound to be reproduced. But this insertion is not sufficient to obtain a good quality, due to the second above-mentioned difficulty.

In order to increase the quality, the here described correction system interleaves pitch and statistic excitation modes on a time-sharing basis. Accordingly, when the analyzer has recognized a voiced fricative, it branches to a program issuing a listing according to one of the following forms:

(Image Omitted)

1