32/172? O.o
Can you even hear the difference between that and regular 24/96? Bit depth just affects the dynamic range/noise floor and even a 20-bit bit depth is plenty for music. Sampling rate...well the Nyquist Theorem. A sampling rate of 48 kHz should give plenty of headroom for music.
I spent over two months comparing different ADCs, mostly pro gear from Apogee, RME, EMU, etc... then spent another month experimenting on sampling rates and bit rates. I could tell the difference between the 24/96 and 32/176 files every time. Not as large of a difference as 16/44 to 24/96, but a difference none the less. Since this was a massive project - I only wanted to do it once. Besides once the analog rig was sold there would be no chance to repeat.
Funny the 32/192K samples were not quite as good as the 32/176k at least to my ears. These were all compared A/B to the LP playing - across all music (Rock, Alt, Emo, Jazz, New Age, Reggae, Pop, R&B, and Classical (Chamber and Symphonic)) Approximately half my collect is Rock, Alt and Emo.
We now have DxD as an emerging standard - that is PCM at 24/352.8k. Many are finding it superior to DSD.
Here is a great whitepaper on digital sampling rates.
http://www.merging.com/uploads/assets//Merging_pdfs/dxd_Resolution_v3.5.pdf
It's not just the noise floor - but the pulse response rate. And my own theory as to the capture of inter-phase relationships.
I used Steinberg WaveLab 6.0 the best audio software package available at the time.