Regeneration Model for Esophageal Voices

B. García, J. Vicente, I. Ruiz, E. Loyo, and A. Alonso (Spain)


Signal processing esophageal speech, pitch scaling and jitter reduction, sonority detection


In this paper it is described a new processing for esophageal speech regeneration, based on pitch and jitter modification. Traditional phase vocoder and resampling pitch scaling techniques have been used to develop a new adaptive method which scales the low esophageal speech pitch and applies a variable scaling factor significantly reducing its jitter. Due to the high level of glottal noise in this kind of voices, it is necessary to include in any regeneration model as a preprocessing which deletes this noise. When trying to delete noise from the speech signal, spectral subtraction techniques are not valid due to a double reason, on the one hand the high level of noise present on these signals, and on the other hand the weakness of the harmonics. It is needed a new technique that allows to eliminate glottal noise, which seriously difficults later processing of this speech.

Important Links:

Go Back