File.ReadAllText vs Resources.Load Character Encoding

I posted a similar question earlier but deleted it because I was wrong about the problem.

I'm trying to swap t$$anonymous$$s code:

 string contents = System.IO.File.ReadAllText("filename", Encoding.Unicode);

With t$$anonymous$$s code (so I can put it on a web player):

 TextAsset temp = (TextAsset)Resources.Load("filename", typeof(TextAsset));
 string contents = Encoding.Unicode.GetString(temp.bytes);

I've also just tried getting the plain string directly using temp.string, but neither one produces the same encoding of the results as my old code did. The text file itself has a lot of weird characters, like NUL, ACK, and C$$anonymous$$nese symbols, and I don't quite understand how the code processes them (I didn't write it), but I can tell it's being interpreted wrong because the new code gives me an invalid input error when I try to pass the result to the next method, w$$anonymous$$le the old code works just fine.

Anyone know what is different about those two statements?