Assessment of Articulation Disorder Using Objective Quality Measures
Smitha Paul, Ben P. Babu, Leena Mary · 2018 International Conference on Control, Power, Communication and Computing Technologies (ICCPCCT) · 2018
Since speech is an integral part of communication, people with various speech disorders feel isolated in the society. The work described in this paper deals with articulation disorder which is a common speech disorder. The difficulty in correctly pronouncing the phonemes is referred to as an articulation disorder. Most of such pronunciation errors seen in children or adults are corrected with the aid of speech therapist, by giving individual training. With increasing population, speech therapists require assistance of signal processing techniques to attend increasing number of patients. As a preliminary work towards automation of Malayalam articulation test, here we investigate substitution type articulation disorder with waveform analysis. Objective evaluation of articulation disorder that rely on techniques such as dynamic time warping (DTW) is introduced. To quantify articulation disorder, objective speech quality measures such as Itakura-saito (IS), log-likelihood ratio (LLR), log area ratio (LAR), log cepstrum distance (LCD), signal to noise ratio (SNR) are computed between normal and disordered speech. Finally these objective measures are combined together using majority voting method to obtain a score. False rejection ratio calculated for these score values shows that these objective measures have good correlation with subjective evaluation results.