For the synthesized singing voice to sound breathy, aspiration noise is perceptually most important. We have proposed to use the
pitch-synchronous, amplitude-modulated Gaussian noise to model the aspiration component in the glottal excitation. For proper parameterization of the noise residual model, we need to analyze the signal properties of the glottal aspiration noise estimated from the real breathy singing recordings.
Several wavelet thresholding and best basis thresholding methods have been studied and compared in this paper in order to obtain reliable estimates of the glottal aspiration noise. This paper concludes that the best basis hard thresholding method is the most effective way to extract the glottal aspiration noise.
