Feature: please consider replace fftw3 with vkFFT or other FFT-in-GPU lib

[StrideZhou]

Modern computers have graphics cards, even cell phones, and by utilizing vkfft, you can put GPU resources to use to increase decoding performance and get a higher SNR . Also nowadays, 24bit, 48k sample rate sound cards are very common on the market, We can put it to good use.
VkFFT - Vulkan/CUDA/HIP/OpenCL/Level Zero/Metal Fast Fourier Transform library