Multidimensional analysis of speech (in most cases: time-frequency representation) forms the basic technique for many speech applications, like speaker identification, speech enhancement, but also for automatic speech recognition. Unfortunately, the widely used transform for this analysis, the Short-time Fourier Transform (STFT) has its drawback: in cases when the pitch of the speech signal changes, the classical spectrogram gets distorted. The higher the pitch-variation, the bigger the distortion is.

To eliminate this distortion and to achieve better representation of speech signal even with changing pitch, we defined a new analysis-synthesis method for speech signals, namely the 'The Short-Time Chirp Transform, ie. STCT (STChT)'. The STCT is a special adaptive time-frequency analysis , we could say, a generalized version of the classical STFT.