Abstract: Because of the limitation of Heisenberg's uncertainty principle, the algorithms of time-frequency spectrum decomposition, for example, wavelet transform, and S transform, cannot simultaneously have high resolution in time and frequency domain. In order to meet the higher requirements, there appeared a new method that combined wavelet transform with the time-frequency spectrum rearrangement, which is called synchrosqueezing wavelet transform. In this paper, the related parameters of the algorithm which affect the results of the time-frequency analysis, such as wavelet mother function and its parameter selection, wavelet threshold, are studied. The time-frequency spectrum ambiguity has been analyzed when the variation rate of instantaneous frequency of the signal not equals to zero. The generalized synchrosqueezing wavelet transform algorithm is studied, and the quality of time-frequency spectrum is improved to a certain degree. This result has a certain guiding significance for achieving high resolution time-frequency spectrum.