TADA: Task Dynamic model of inter-articulator speech coordination

Description

TADA is a software implementation of the Task Dynamic model of inter-articulator speech coordination, incorporating also a coupled-oscillator model of inter-gestural planning, and a gestural-coupling model. Currently, the system will generate articulatory models of English utterances from either phonetic or orthographic text input, and can output HLSyn-compatible speech synthesis files, coupling graphs, gestural scores, and model articulator time functions. The TADA system has also been designed to allow for the development of articulatory models in other languages through the use of language-specific dictionary files and gestural databases.

Developed by: Hosung Nam, Cathe Browman, Louis Goldstein, Michael Proctor, Philip Rubin, Elliot Saltzman
Written by: Hosung Nam
Total number of Downloads : 496
Number of Unique Affiliates : 374

Functionality

TADA is an integrated software suite consisting of three interconnected models:

  1. Syllable Structure-based Gestural Coupling Model:
    Takes as input a text string and generates an intergestural coupling graph that specifies the gestures composing the utterance (represented in terms of tract variable dynamical parameters and articulator weights) and the coupling relations among the gestures' timing oscillators.
  2. Coupled Oscillator Model of Inter-gestural Coordination:
    Takes as input a coupling graph, and generates a gestural score specifying activation intervals in time for each gesture.
  3. Task Dynamic Model of Inter-articulator Coordination
    Takes as input a gestural score and generates the motions of the vocal tract constriction variables and the articulatory degrees of freedom.

Further information about the Gestural Computational model may be found here. More specific information about the implementation of TADA may be found in the TADA manual.

Software

Version 0.982 of TADA is freely available to speech researchers and interested parties. Documentation is available from here, and the program itself from here. The model will run in Windows MATLAB (version 2006a or later). In order to create full synthesized speech output, Sensimetrics HLSyn is also required, although the basic TADA system can be run under MATLAB alone.

Please note that TADA is no longer being actively supported.

TADA is (c) 2001-2022, Haskins Laboratories, Inc., all rights reserved. This software is for non-commercial use only. Please do not distribute this software without our permission.

Gestural score files annotated for the University of Wisconsin X-ray microbeam database.

TADA demo video singing "The Star-Spangled Banner"