GAN You Hear Me? Reclaiming Unconditional Speech Synthesis from Diffusion Models | IEEE Conference Publication | IEEE Xplore