I think I need clarification. I've read at least two different threads that defined 24 bits padded differently. One says it is really 16 bits padded to 24 (so really a 16 bit signal). The other says 24 padded is really 24 bit padded to 32 (so really bit perfect 24).
I am sending 192/24 flacs to the player and I really need to know that I'm getting a bit perfect signal (true 24 bit).
So which is it? Does anyone know for sure?
It is 24 padded to 32.
16-bit can be output directly to any device. 24 cannot, which is why there is an option to pad it to 32-bits.
Padding to 32-bits has zero effect on the audio signal.
Also, if I choose the Marantz in audio devices I get 192/24 padded (this is wasapi direct through hdmi). If I use Realtek digital wasapi (toslink) I get 192/32 - so what is the real difference? Is there any audible difference to me? Are they both playing the bit perfect 192/24 signal? And if they're the same, what's the reason for having two outputs that do the same thing with different names?
If the 32-bit output is actually working, it likely gets truncated to 24-bit at some point.
Proper dithering to 24-bit would give you better quality with a 24-bit DAC.
HDMI can support lossless multichannel audio, while S/PDIF is a lower bandwidth connection and can only support lossless stereo audio.
Multichannel audio has to be compressed using AC3 or DTS with S/PDIF.
For stereo audio, it largely doesn't matter what connection you use. With older devices, S/PDIF may have less jitter than HDMI, but that's generally not an issue any more.
Bit-perfect audio is overrated.
Most audio setups benefit from processing the signal with things like room correction, and volume leveling results in a better user experience even if it doesn't affect audio quality.
Just make sure that dithering in Media Center is set to TPDF.