Speech Female Hpsmodel Stochastic

Stochastic approximation of the residual signal obtained after subtracting the harmonics from a female speech sound: http://www. Freesound. Org/people/xserra/sounds/254374/. This was done using the harmonic plus stochastic model implemented in the sms tools: http://mtg. Upf. Edu/technologies/sms. The parameters used for the analysis were:. - window: blackman- window size: 2001- fft size: 2048- magnitude threshold: -100- minimum duration of sinusoidal tracks: 0. 05- maximum number of harmonics: 100- minimum fundamental frequency: 150- maximum fundamental frequency: 250- maximum error in f0 detection algorithm: 5- max frequency deviation in harmonic tracks: 0. 05stochastic approximation factor: 1.