Dolby's suggested bitrate for 2.0 in AC3 in 192 kbps. This is transparent to me - I cannot tell the difference between uncompressed PCM and a 192 kbps AC3. This is not opinion, this is fact: proven via a double-blind ABX test.

This is not to say that 192 kbps is transparent to everyone; it's possible that there are some people, maybe with better ears or better equipment, who can tell the difference. I also suspect it's quite likely that some people think they would easily be able to tell the difference, but if they actually tried to prove it via an ABX they would fail.

When it comes to audio, opinions are quite often bullshit. I would advise you to spend some time reading on listening tests at the hydrogenaudio.org forums, and try some tests of your own.

It is even possible that some people would be able to distinguish between 640 kbps AC3 and the uncompressed original. The only way to be sure your audio is transparent to everyone is to keep it uncompressed, or use lossless compression. As far as I know there is no free or open source software that can compress to DTS-HD MA or Dolby TrueHD, but if you wanted to release your project as an MKV instead of a structured BD you can easily include lossless FLAC audio.

You wouldn't even need to do a 44.1 to 48 kHz conversion, saving another potentially non-transparent stage, but I'm not sure how well hardware players would support playback of an MKV containing 44.1 kHz FLAC audio.