Statistical-model-based speech enhancement with musical-noise-free properties | IEEE Conference Publication | IEEE Xplore