We actually have someone working on this. The user will record his voice with MacRecorderΓäó (right in the stack). An XFCN will analyze the recording using fast fourier transforms, comparing it against a
"model" sound.
This will be useful for practicing foreign language pronunciation, tones, etc.