Adaptive chirp-based time–frequency analysis of speech signals

In this paper, a new method for time–frequency analysis of speech signals is proposed. Given that the fundamental frequency of voiced speech often undergoes rapid fluctuation and in these cases the classical spectrogram suffers from blurring and artifacts, an adaptive analysis basis composed of quadratic chirps is what we consider. The analysis basis of the proposed Short-Time Fan-Chirp Transform (FChT) is defined univocally by the analysis window length and by the frequency variation rate, this parameter being predicted from the last computed spectral segments. The prediction algorithm is based on time tracking the joint trajectory of the harmonic contours, this process also provides a voiced/unvoiced detection parameter. Comparative results between the proposed Short-Time FChT and popular time–frequency techniques reveal an improvement in spectral and time–frequency representation. Since the signal can be synthesized from its FChT, the proposed method is suitable for filtering purposes.