It has been shown that visual display systems of intonation can be employed to advantage in teaching intonation to deaf persons and in teaching the intonation of a foreign language. In current training situations, the correctness of a reproduced pitch contour is either rated by the teacher or automatically. In the latter case, an algorithm often estimates the maximum deviation from an example contour or something equivalent. Two issues are addressed in this paper. The first issue is concerned with the question whether visually conspicuous differences between displayed pitch contours correspond with important audible differences. Second, in automatic training, the ratings of the correctness of an imitation must be supplied by an algorithm. These automatically obtained ratings must correspond with perceptual similarity ratings of pitch contours. It is shown that visual and auditory similarity of pitch contours correspond well if pitch contours are displayed in such a way that only perceptually relevant features are represented. Furthermore, for automatic rating, the measure obtained by calculating the maximum deviation from an example contour had a worse correspondence with similarity ratings by phoneticians than three other physical similarity measures.
Bibliographic reference. Hermes, Dik J. (1995): "Measuring the perceptual similarity of pitch contours", In EUROSPEECH-1995, 2051-2054.