APEX model

APEX: an articulatory synthesis model for experimental and computational studies of speech production.

Writers: Stark, J. and Lindblom, B. and Sundberg, J

Dept. for Speech, Music and HearingQuarterly Progress andStatus Report

Institution: KTH

http://www.speech.kth.se/qpsr

Speech, Music and Hearing
Communication and interaction between humans via speech and music is the unifying theme of research and teaching within the Department of Speech, Music and Hearing. The department is engaged in a diverse set of multi-disciplinary activities, commonly classified into speech communication, speech technology, speech coding, music acoustics, auditory perception and second language acquisition, to mention the largest areas. The department is also the home of the KTH Unit for Language and Communication.

Abstract

This is a preliminary report of a project in progress with the purpose to create anarticulatory synthesis model for studies of speech production. It is realised by acomputer program which may control the lips, the shape of the tongue body andapex, and the mandible. An area function may be computed and displayed graphicallyand numerically. Formant values may be computed and sent to a formantsynthesis model for sound production using a DSP hardware module. Automaticand systematic generation of parameters may be achieved and the results sent to adisk jile. The program keeps all speaker dependent data in a disk file, enablingprocessing of several speakers.

Introduction

The Apex project is aimed at the creation of anarticulatory speech model to be used as a tool instudying an important class of speech sounds:apical speech sounds. The need for such amodel is for example found in phonology:(Browman and Goldstein 1992), speech technology:articulatory speech syntesis: (Lin 1990,Fant 1992), and in general basic phonetic research:(Hardcastle and Marchal 1990). Researchin music acoustics and in singing mayalso be mentioned. Another goal is to gain adeeper understanding of coarticulation ofspeech. This knowledge may hopefully beachieved by comparing speech data from themodel with data gathered by laboratory experiments,e.g. movement data using a movetracksystem (Branderud) and data from signal analysisand spectrograms.

The apex model

The apex model is created by a parabolic function.The parabola is attached by its one leg tothe tongue body. The other leg's end correspondto the tip. The model is rotated in order toacheive a smooth conjunction between body and apex.

http://www.speech.kth.se/prod/publications/files/qpsr/1996/1996_37_2_045-048.pdf

Related reading
THE APEX MODEL AS A TOOL IN THE SPECIFICATION OF SPEAKERSPECIFICARTICULATORY BEHAVIOR
Johan Stark*, Christine Ericsdotter*, Peter Branderud*, Johan Sundberg¤, Hans-Jerker Lundberg^,Jaroslava Lander+
*Stockholm University, ¤Royal Institute of Technology, ^Danderyd Hospital, Stockholm,+Södersjukhuset, Stockholm
http://www.ling.su.se/fon/perilus/1999_19.pdf