Muffled speech synthesis on linux (mp3 samples provided!) (with linux16audio)

I have setup festival on a linux centos server. It is currently generating wav files with text2wave that are muffled and distorted.

h t t p://67.43.62.154/festival/speech2-clean.mp3 (generated from the online demo at the festival uk site)
h t t p://67.43.62.154/festival/speech2-muffled.mp3

When putting (display "linux16") in the conditional that finds linux16audio in *modules* in init.scm it causes an error so I think festival is defaulting to linux16audio.

I am getting the following output with echo '(*modules*)' | festival -i:

module_description hts_engine MultiSyn clustergen_engine clunits rxp UniSyn_diphone parser donovan UniSyn linux16audio

1. How do I setup Festival and an audio solution so that the synthesized speech is clear?
2. How do I lookup my audio hardware configuration on this server?