Utterly no point as there would be no quality gain. The original WAV files are uncompressed (obviously), but they have a low sample rate (11050 Hz mostly) and have a 4 bit sample size. They are rather low quality as a result.
An idea I had was to add Speex support as this would be more suited to voice compression than Vorbis.
Out of interest, an original voice file from FS2 can be compressed 4 times with Speex without any obvious difference. (although this may be because the original files are bad to start with). This was with Quality 3, Complexity 3 and VAD on. (I'm open to better setting suggestions.)