Speech enhancement based on joint time-frequency segmentation | IEEE Conference Publication | IEEE Xplore