These sound examples come from the voice separation system developed by Mitch Weintraub while at Stanford, and described in his 1985 Ph.D. dissertation, "A theory and computational model of auditory monaural sound separation" (Stanford University Dept. of Elec. Eng.). You can consult the thesis for a detailed explanation of how the system worked, its goals, and how these examples were produced. (I was able to see a copy through UMI, the thesis microfilm service, via my college library). I got hold of these examples from Malcolm Slaney, who has them on the web here.
The examples (available either as 44kHz 16 bit AIFF, or 22kHz 16 bit MS-WAVE):