Enhancing the Sharpening Performance of the Short‐Time Fourier Transform
Hongbo Yu, Liang ZhangABSTRACT
Time‐frequency reassignment and the synchrosqueezing technique are commonly used to sharpen a time‐frequency representation obtained by short‐time Fourier transform (STFT) or others. However, the technique requires the components in the analysed signal being separable in time or frequency with respect to the used analysis window. To weaken the limitation of separability, this letter proposes a method to enhance the sharpening performance of STFT, first setting a narrow window to ensure a high time resolution, then extrapolating signal frames via a linear prediction technique to improve the spectral resolution, and finally sharpening the obtained STFT data. Using a call signal containing both the impulse‐like and frequency‐modulated components, the performance is verified.