Analysis of hypernasality by synthesis

Speakers with velopharyngeal incompetence, produce hypernasal speech across voiced elements. During the production of vowels, due to the defective velopharyngeal mechanism, oral cavity gets coupled with the nasal cavity. This coupling may introduce resonances of nasal cavity in the resulting vowel sound. In the present study, an attempt is made to convert pure vowels in to its nasalized counterpart, by introducing nasal resonances in full/part, in addition to the oral tract resonances. Further, an analysis is done by varying the formant bandwidths of all the resonances. By considering nasalized vowel of normal speakers as reference, a perceptual experiment is conducted and validated. The speech data of 10 patients with cleft palate, who are expected to produce hypernasal speech were analyzed. The presence of nasal formants in their Linear Prediction spectra, validated the above results. Presence of nasal formants in the hypernasal speech, may be a cue for detection of hypernasality. Another experiment is also conducted in the time domain, to verify whether the widening of formant bandwidth is due to the distribution of energy between oral and nasal cavities.