Note: As the mountains separated, the real guy went from being the “Savior of the Republican Party” (Time magazine cover, and many others) to dead last in all polls.
Let’s hope this humble developer will not suffer similar fate :-
Regards,
-Ramon
ps: Determining the identity of the horizontal fellow -in the context of the NR (left mountain) vs. VR (right mountain) divide- will be left as an exercise to the reader.
Oh! I almost forgot. For context, see the thread
“The Human Voice has been widely characterized, correct?”
in the Usenet comp.dsp newsgroup, plus this:
https://sourceforge.net/p/cmusphinx/discussion/speech-recognition/thread/1177f599/
-RFH
Possibly a more realistic approach would be to start by building Audacity from the source code and studying the current Noise Reduction algorithm to see how it works. Then see if you can improve on it. If I recall correctly I have already given you links to the developer information for obtaining the source code and building Audacity.
It is funny how I got the exact opposite advice from your counterpart in the CMU Sphinx newsgroup. 
https://sourceforge.net/p/cmusphinx/discussion/speech-recognition/thread/1177f599/
IOW: The mountains are separating…
…and I can’t find my wings!!
-Ramon