Multitaper adaptive short-time Fourier transform with chirp-modulated Gaussian window and multitaper extracting transform

The adaptive short-time Fourier transform with chirp-modulated Gaussian window (ASTFT-CMGW), a recently developed time-frequency analysis method, presents a high energy concentration in the time frequency plane by using two signal-dependent parameters, i.e. fractional Fourier transform (FRFT) angle and Gaussian variance. Nonetheless, it is difficult to obtain the optimal time-frequency-varying parameters. In the paper, the ASTFT-CMGW containing a fixed variance and a time-frequency-varying FRFT angle is investigated, and it is abbreviated as ASTFT-CMGW-FVVA. For the time-frequency-varying FRFT angle, a simple calculation method is proposed, which significantly improves the previous method based on the chirp rate of the input signal. However, the fixed variance hinders the complete adaptivity of the ASTFT-CMGW-FVVA. To alleviate this problem, multiple ASTFTs-CMGW-FVVA are combined to get the multitaper ASTFT-CMGW. For the purpose of further sharpening the time-frequency representation, the multitaper extracting transform is developed, which is available by utilizing the solution of a system of equations that the instantaneous frequency of the signal satisfies. Furthermore, the relationship between the proposed extracting transform and the extracting transform based upon the ASTFT-CMGW-FVVA is given. Experimental results demonstrate the effectiveness of the multitaper ASTFT-CMGW and the proposed extracting transform. (C) 2022 Elsevier Inc. All rights reserved.

Automated summary

This paper investigates the improved adaptive short-time Fourier transform with chirp-modulated Gaussian window (ASTFT-CMGW) method and proposes a method called ASTFT-CMGW-FVVA with fixed variance and time-frequency-varying parameters. The multitaper ASTFT-CMGW and multitaper extracting transform are developed to further enhance the time-frequency representation precision.