Subject: Speech Processing

Topic: Speech Production, Acoustic Phonetics and Auditory Perception

Difficulty: Low

(i) By switching between the voiced and unvoiced excitation generator a changing mode of excitation can be modeled.

(ii) The vocal tract can be modeled in several ways already discussed.

(iii) The model works well or continuation sounds (vowels, diphthongs etc.) since the parameters undergo a slow change.

(iv) The model does not perform very well for transients sound like stops.

(v) Lack of zeros being provided limits its usage for nasals and fricative, since the inclusion of zeros are necessary for nasals and fricatives.

(vi) The zeros can be included if required bu the model.

(vii) The simple division of voiced-unvoiced excitation does not suffice due to the correlation of the frication and peak of glottal flow.

(viii) Therefore a more sophisticated model is required to be designed.