Should I use 32bit samples inside kontakt? Seen recently kontakt libs bloated to the max using 32bit samples - whats the point using 32bit wordlength waves inside a sampler?

tia - & merry x-mess 4 yall!

 

There is no benefit other than increasing the size. Kontakt's resampling algorithm is great, but introduce harmonics under 120dB, so you won't use the 32bit dynamic range. Also no need for floating point IEEE 754 as kontakt won't use it, well, I can't imagine how that could be of use :dancing:
Better to keep the bit depth the same at all times (24bit), dithering only the end result

 

Hi hpf I do a lot of time stretching and I find if there in 32bit floating point the results are a lot better they dont sound so metallic. and if there was no benefit all sound libs would be in 8bit.

 

If this is true, then Kontakt's algorithm seems to use too low precision. Even with 16bit samples, time stretching wouldn't really introduce audible artifacts if, like in any good algorithm, the signal is processed with at least 32bit or even 64bit internally.

Maybe it has to do with Kontakt's setting for rendering quality which may affect time stretching quality as well?

 

Why use more pixels on cameras ...?

 

Is it just my feeling or are we talking apples and oranges here
I though the original question was about floating point vs. integer, e.g. IEEE 754 vs. PCM and not how bit depth is important.

Otherwise I tend to agree with fiction, Kontakt should process sample of any bit depth as single precision 32bit or double precision 64bit.
I have never experienced metallic sound with Time Machine Pro. Try that, Lukie, well if you aren't using it already. *yes*

As long as still picture is required, more pixels will generate better result, however when recording video, it is always better to have CCD/CMOS sensor with native resolution (2.08Mpix) as no resizing or interpolation is needed. Also those sensors tend to pick up more light.

 

" Why use more pixels on cameras ...?"

You're missing the point (or pixel).

If you want a picture analogy : "why use wider color range?"

- answer is , it's unnecessary, 24 bit is already a WIDE range.

For kontakt (and sampler in general), it make more sense to have a bigger sample rate (for more precise repitching) : 96 or 192k instead of 44.1

 

Excellent analogy, why it didn't occur to me, though it's obvious *yes*
Bit depth - colour depth
aka. 8bit scheme (True Colour - 256^4) vs. 10/12/16bit (Deep Colour)
Number of pixels is exactly - "sample rate"
However in imagery, chroma is sampled differently, by subsampling matrix (YUV 4:2:0 4:2:2 4:4:4, RGB)

Back to audio

If you notice any difference when using 32bit FP in Kontakt then use it. *yes*
I don't hear it, but all ears are different

 

ok so the 32bit type samples used in some 8dio libs are pretty useless and just increase size to let you think you get a lot (bytes) for the money thats what ive expected. tried to figure out a quality difference in 24vs32 but i really couldnt. i totally agree that 96k make sense in very special cases where samples are naturally pitched down by multiple octaves, but thats a more sound design ans sfx thing - in most cases pointless as its pretty common to use at least one sample per note in most modern libs. short 24/48 is deluxe and theres no reason to use higher specs except you want blow up your lib to the max, just to say "but mine is bigger" ^^ or "more bang fo the buck" why they think we are on unlimited cap hdd?

remember the times where patches had few megs, with L/R loops of different length seamless perfection, hard to tell its looping and with added fx even evolving. you could just play lots of stuff directly from ram.

todays lazyboys dont even care and leave the five seconds of silence for you to worry about, this for 88 notes with 8 dynamic levels you get gigabytes of silence but bigger is better isnt it?