upsample / oversample

[krass]

I've seen a few posts extolling the virtues of up sampling, either the source files or in software during playback (specifically some commentary suggests big benefits when using NOS DACs since the up sampling moves the inter-mod distortion products out of audio band).

 

I'm not disputing this at all, it sounds logical.

 

What I don't understand is how up sampling from say 44.1 source file to a non-integer multiple (48,96,192) can achieve anything positive. I can imagine that 44.1 to 88 or 176 in software will bring the benefits reported, but surely 44.1 to 48 etc will just result in source data that exhibits 44.1 48 cross products in the audio, since the 44.1 file is just a series of steps at 44.1 frequency......

 

Can anybody comment or explain in language that is not *too* technical ?

 

thanks, mike

 

[krass]

I should add that I'm considering a Metrum NOS DAC that is limited to 176; for my 44.1 files I can upsample *4, but for future proofing I am concerned that the missing 192 input will restrict me to *2 up sampling for 48 kHz source files and 1* up sampling for 24/96.

 

If the NOS DAC really needs up sampled sources to remove IMD, is the missing 192 input a problem for the future ?

 

sorry I'm rambling so much, thanks for any input,

Mike

 

[ted_b]

that the upsampler PM uses is good enough to upsample without regard to integer math (meaning ok to upsample 24/96 to 24/176k). Several folks here agree that modern SRC's are good enough to do that. Personally, I'd only upsample redbook 4x and leave all the 24 bit native files alone anyway.

 

[Miska]

What I don't understand is how up sampling from say 44.1 source file to a non-integer multiple (48,96,192) can achieve anything positive.

 

For certain class of algorithms, 44.1 -> 48 and similar, are actually also integer multiples. It is just less obvious. So there's no necessarily any difference in quality.