Lossy coding and bitrate effects on changes in formant frequencies in Japanese and English speech signals

Abstract

Since speaker recognition and verification became
heavily used technology, both in professional applications like
forensics and more everyday ones, the question arose: what
factors can impact results of those processes? One thing that
may be important with respect to this subject is lossy coding,
as some of the information contained in an original file is lost
in the coding process. In the era of globalization, not only
native languages or languages of neighboring countries are of
interest to researchers, but also those quite far, especially from
Asia – the biggest exporter of goods and services to Europe.
Those economic relationships are usually connected with the
interchange of personnel, which further shortens geographical
distance. The article presents the results that are a continuation
of research on the behavior of Japanese language formants.
Earlier research focused on changes occurring for the first
and second formants. This article presents changes observed
for the third and fourth formants. The knowledge of these
changes is indicated in the process of speaker identification in
forensics using the spectrographic method. At the Department
of Acoustics and Multimedia, Wroclaw University of Science
and Technology and in many centers around the world, the
auditory-spectrographic method is used, which is a combination
of the aural and spectrographic methods. In the spectrographic
part, a person is identified on the basis of a comparison of the
formants’ trajectory.