4th International Conference on Spoken Language Processing

Philadelphia, PA, USA
October 3-6, 1996

A Comparison of Modified k-means(MKM) and NN based Real Time Adaptive Clustering Algorithms for Articulatory Space Codebook Formation

K. S. Ananthakrishnan

School of Electronic Engineering, University of South Australia, Levels, SA, Australia

This paper proposes the use of a neural network based real time adaptive clustering* algorithm for the formation of a codebook of limited set of acoustical representation of finite set of vocal tract shapes from an articulatory space. Modified k-means algorithm (MKM) used for clustering nearly 10000 vocal tract shapes into 1000 cluster centers to form a codebook of articulatory shapes is computationally intensive for our application. An investigative study on the use of NN based algorithm over MKM algorithm at the peripheral level, for our application on Computer Aided Pronunciation-education, suggests the former for less intensive computation, with the possibility of improving the performance of the system by implementing the algorithm using a dedicated neural computer. In this paper, preliminary results of this study are reported.

Full Paper

Bibliographic reference.  Ananthakrishnan, K. S. (1996): "A comparison of modified k-means(MKM) and NN based real time adaptive clustering algorithms for articulatory space codebook formation", In ICSLP-1996, 1253-1256.