Re: shufps bug ? (And interpolation)
Posted: Sun Aug 28, 2022 3:27 pmThanks, very interesting, make me want to do some more interpolation !)
But it will be hard for me to base on the interpolation utility..
I was using c++ with unity not so long ago..
But i have the tendency to focus on only one language at a time and to forget other's.
Now that i'm on asm, it's hard for me to do dsp. And c++ is even more difficult..
Also, the formula provided use a lot of if-else, or loop. But in asm it's not so easy or convenient to use those.
(i also need to learn more how to do a loop in asm)
I'm not sure how much the quality of the interpolation count in a "normal/classic" dsp effect ?
If the pitch modulation is low, (and the sample rate at 44100 or more) it's very difficult for me to hear a difference, even with a linear one.
But when i turn the delay knob largely (and maybe more specially causing a pitch down) i think that i hear a difference.
I imagine that there's some application where the best interpolation is needed.
But finally, once we do not know where could be the real unknown point of the signal, how could we say that this or this is more precise ? (unless maybe an intelligent interpolator that do some fractal calculation on the real point to find a logic in the signal ?)
For example, let's say that i have those point recorded, but the signal in real was this :
How could i know the real thing and the best interpretation ?
But it will be hard for me to base on the interpolation utility..
I was using c++ with unity not so long ago..
But i have the tendency to focus on only one language at a time and to forget other's.
Now that i'm on asm, it's hard for me to do dsp. And c++ is even more difficult..
Also, the formula provided use a lot of if-else, or loop. But in asm it's not so easy or convenient to use those.
(i also need to learn more how to do a loop in asm)
I'm not sure how much the quality of the interpolation count in a "normal/classic" dsp effect ?
If the pitch modulation is low, (and the sample rate at 44100 or more) it's very difficult for me to hear a difference, even with a linear one.
But when i turn the delay knob largely (and maybe more specially causing a pitch down) i think that i hear a difference.
I imagine that there's some application where the best interpolation is needed.
But finally, once we do not know where could be the real unknown point of the signal, how could we say that this or this is more precise ? (unless maybe an intelligent interpolator that do some fractal calculation on the real point to find a logic in the signal ?)
For example, let's say that i have those point recorded, but the signal in real was this :
How could i know the real thing and the best interpretation ?