Speaker identification enhancement under cochannel conditions using sinusoidal model-based usable speech detection | IEEE Conference Publication | IEEE Xplore