Capturing audio in 32 bit is pointless. 32 bit floating point is even more pointless. You would need extremely high quality electronic components within the rest of the recording chain to realize a theoretical advantage, but there would never be an audible advantage over 24 bit, or probably even 16 bit. None of the high end audio manufacturers offer 32 bit capture, and the ones that do offer it, is only for marketing hype. Just like recording at a sample rate higher than 96Khz, it taxes the system more, without any audible advantage at all.