DIVISION OF SPEECH SIGNALS INTO VOCALIZED AND UNVOCALIZED SECTIONS ON THE BASIS OF SIMULTANEOUS MASKING
A. A. Konev, R. V. Meshcheryakov, E. Yu. Kostyuchenko
Tomsk State University of Control Systems and Radioelectronics, 634050, Tomsk, prosp. Lenina, 40
Keywords: речевой сигнал, одновременная маскировка, сегментация речевого сигнала, вокализованные и невокализованные участки, speech signal, simultaneous masking, speech signal division, vocalized and unvocalized sections
Abstract
A model of simultaneous tonal masking that isolates speech signal components perceived by a human's auditory system is under study. An algorithm of simultaneous masking on the basis of this model. It is shown that, after simultaneous masking, a signal is represented by a binary structure reflecting the harmonic structure of a vocalized sequence. It is experimentally proved that this structure can be used to isolate key (in terms of perception of the auditory system) speech sections. This structure serves as a basis for the algorithm of high-quality speech signal division into vocalized and unvocalized sections, which does not require training before use. According to the results of testing of the joint use of algorithms for simultaneous masking and division of speech signals, the quality of their performance is obtained.
|
