Improved vocal tract model for the elongation of segment lengths in a real time. (September 2019)

• Record Type: Journal Article
• Authors: Qureshi, Tahir Mushtaq; Syed, Khalid Saifullah
• Is Part Of: Computer speech & language. Volume 57(2019)
• Page Start: 41
• Page End: 58
• Publication Date: 2019-09
• Subjects: Waveguide -- Vocal tract -- Fractional delay -- Elongation of segment; Speech processing systems -- Periodicals; Automatic speech recognition -- Periodicals; Computers -- Periodicals; Linguistics -- Periodicals; Speech-Language Pathology -- Periodicals; Traitement automatique de la parole -- Périodiques; Reconnaissance automatique de la parole -- Périodiques; Automatic speech recognition; Speech processing systems; Electronic journals; Periodicals; 006.454
• DOI: 10.1016/j.csl.2019.02.001
• Languages: English
• ISSNs: 0885-2308

Abstract

Abstract: In the current work, an efficient approach has been implemented to model the fractional delay in the elongated cylindrical segments of the vocal tract in waveguide modeling. The vocal tract has been divided into uniform cylindrical segments of the variable lengths. In this case, the time taken by the sound wave to propagate through a cylindrical segment in an axial direction may not be an integer multiple of each other. This means that the delay in an axial direction is necessarily a fractional delay for each fractional elongated segment. In the previous work, to accommodate the fractional delay for each elongated cylindrical segment, two extra cylindrical segments of same lengths were added to maintain the even number of segments. In the proposed work, we add only a single extra segment for each fractional elongated segment which reduces memory and computational cost as well. To keep the even number of segments, we assume that the extra and original segments constitute a single long segment. Lagrange interpolation is used for the approximation of the fractional delay. The proposed model has been devised for the elongation of any arbitrary cylindrical segment by a suitable scaling factor. These results are validated with an accurate benchmark model. This model has a single algorithm and there is no need to make sections of the segments for the elongation of the intermediate segments.