[music_man]

Which is greater? 384x32 or 768x24? Should the greater necessarily sound the best? I am having difficulty deciding. They both have merits. I cannot say one is better, but neither is perfect.

 

[msing539]

Are you referring to sample rate in Windows?

 

[music_man]

Yes.

 

[msing539]

My experience is that once you push it too high, Windows starts introducing artifacts into upsampled music. Others might have a different take.

 

[music_man]

Is that windows or cpu limitation? My dac can do 1024 dsd but even the current best I9 poops out on that. You need Foobar to do that too. Windows itself only goes to 32/384. If that setting is actually ruining the sound though I would really like to be told here. I am not really sure if set on 16/44.1 or what is the best setting.

 

[Roseval]

Bit depth is a simple one.
16 bit PCM recording means a dynamic range of 96 dB max.
16 bit in Windows Sound panel means the mixer will convert to float, mix, dither and convert back to 16 bit integer. So you are adding dither even if a single stream is playing.
Obvious this setting is about the arithmetic precision of the data path beteev PC and DAC. Always set it to the max!

Sample rate is more complex.
Unless NOS, all DAC's are of the up- or oversample type. You might wonder if you profit by up-sampling twice, first the PC and then again by the DAC.
My Benchmark DAC one runs internally at the fixed rate of 126 kHz. Using excessive sample rates like 768 will simple cause the DAC to down-sample.
I prefer automatic sample rate switching, each track is played at its native sample rate. Media players supporting WASAPI/Exclusive mode simply bypass the whole Win audio stack.

If you can't use WASAPI/Exclusive, you are forced to use the Win audio. I set it to the sample rate most common (44.1 kHz in my case) and of course to the max bit depth available.

There is a special case, if Windows is resampling and the signal is very close to 0 dBFS, you get measurable distortion.
Eiter avoid resampling or use the pre-amp function of EQ APO
Bit more details: https://www.thewelltemperedcomputer.com/SW/Windows/SRC.htm

 

[jiiteepee]

Which is greater? 384x32 or 768x24? Should the greater necessarily sound the best? I am having difficulty deciding. They both have merits. I cannot say one is better, but neither is perfect.

Hmm... depends on what is sample rate/bit-depth of your source audio?
If it is, lets say, Redbook CD format (16-bit/44.1kHz) or other 44.1kHz base audio then don't expect your output to stay original and therefore, depending on SRC quality, can be even worse compared to original ... if source format is 16-bit/48kHz or 24-bit/48kHz (or other 48kHz base) then your output should stay original (maybe just 8 (24) or 16 (32) zero bits inserted to data length and maybe just 7 (384) or 15 (768) zeros added between each of those original samples) ... this should sound same if other parts of your playback system works correctly.

 

[gregorio]

Which is greater? 384x32 or 768x24?

768x24 is the greater.

Should the greater necessarily sound the best?

No, there is a finite limit beyond which it will sound the same. Go significantly beyond that limit and it will eventually sound worse because more data to process and transfer means more likelihood of introducing errors.

That limit is about 44.1kHz and 16bit, although a setting of 24 or 32bit is likely to be more efficient as you are processing at that bit depth (SRC and/or a digital EQ for example). The exception might be a NOS DAC, as mentioned, in which case upsampling to 96kHz or even 192kHz might be better.

Note Roseval’s “special case”.

G