>>/33311/
hah, I was going to say it was broken but 'spell text phonetically in english' was in the wrong box.
I like it very much. I didn't like having to import the separate regex library in the python script, plus your solution is much more elegant as it keeps the string rather than pulling bits out so you keep punctuation. However, I have found a bug with eng>bernd 'c' 'k' and 'f' 'ph'. I don't think this would be an issue if you just advise not to spell with 'k' or 'ph'.
I like the interface too but it needs some sort of mascot on the page. Berndese mascot?
tzsz^ mk^w k^ksb