The development of high performace speech input and output devices has produced the need for sensitive and standardised assessment procedures to determine the relative merits and possible applications of different systems. The major approach to this problem has relied on statistical evaluation using percent correct identification by listening crews for processed speech, typically using measures related to the Diagnostic Rhyme Test, and percent correct recognition by a speech input device using a standardised speech data base. Statistical evaluation is essential and serves a dual role, at least in the case of speech input devices, by providing a training data base for speech recognition. However, it is also important to develop analytical, diagnostic measures of performance, preferably based on models of human perception. The studies reported in the present paper describe some developments in the use of analytical techniques for speech technology assessment.
Cite as: Dermody, P., Mackie, K. (1989) Development of analytical speech input/ouput assessment procedures. Proc. Speech Input/Output Assessment and Speech Databases, Vol.2, 155-156
@inproceedings{dermody89_sioa,
author={Phillip Dermody and Kerrie Mackie},
title={{Development of analytical speech input/ouput assessment procedures}},
year=1989,
booktitle={Proc. Speech Input/Output Assessment and Speech Databases},
pages={Vol.2, 155-156}
}