A sparse representation of the excitation source characteristics of nonnormal speech sounds

The impulse-sequence representation of the excitation source information in normal speech signal has been explored for speech coding. Such a representation, if can be developed for paralinguistic and emotional speech sounds, would help in their acoustic analyses. This paper proposes a sparse representation of the excitation source characteristics of nonnormal speech sounds signal, in terms of a time-domain sequence of impulses or impulse-like pulses. Using a recently proposed modified zero-frequency filtering method, an impulse sequence representation is obtained for the excitation source characteristics of nonnormal sounds in three categories: emotional speech, paralinguistic sounds and expressive voices. Validation of effectiveness of the proposed representation is carried out by analysis-synthesis approach and perceptual evaluation of Noh voices. This representation can potentially help significant reduction in the signal storage and processing requirement. It can also be helpful in speech coding of nonnormal speech sounds.