Speech visualization with auditory contours
Peter Daniel, Harald Mundt, Matthias Vormann · The Journal of the Acoustical Society of America · 1999
Speech perception is based on several steps of abstraction or contourization. In analogy to primary visual spatial contours Terhardt introduced the spectral pitch as the primary auditory spatial contour for further complex perceptions like phonemes, syllables, and sentences. Spectral pitch can be derived from the part-tone time-pattern (PTTP) as proposed by Heinbach [Acustica 67, 242–256 (1988)]. The PTTP can be seen as a physical equivalent of the spectral pitch contour. It consists of spectral maxima extracted from a spectrum analysis with time and frequency resolution according to the human ear. By resynthesis it can be shown that the PTTP contains all relevant information of the original signal. In addition one can edit and modify the PTTP of speech in a controlled way and auralize the changes through resynthesis. Further applications like the analysis of F0 contours, separation into voiced and unvoiced parts, virtual pitch analysis and segregation of simultaneous signals will be discussed.