Voice Intonation Transformation Using Segmental Linear Mapping of Pitch Contours

Voice transformation converts the voice of a speaker to an intended person, such that the listener is deceived for the target speaker. Voice transformation has many application like low bandwidth speech encoding, voice disguise, entertainment, voice morphing. The proposed methodology transforms the pitch contour of one speaker to another by segmental linear mapping of the pitch contour. This method considers sound as an one-dimensional continuous signal. A linear mapping function is considered to transform pitch contour i.e. y(t) = f(x(t)), where y(t), x(t) are two different pitch contours of sound signal and f is the mapping function. The segmentation of the pitch contour is done using linguistically motivated parameters, which are extracted automatically from the audio signals to capture the inotation of the voice. The proposed methodology do not need any large speech corpus for training or learning the speech intonation generation, making the methodology computationally efficient. The simulation result proves the efficiency of the transformation of the pitch contour.