dubstepforum.com
worldwide dubstep community

so i understand that if i want to pitch shift and play back a sound lower and slower to record at a higher sample rate

but my question is this : Will sample rate EVER affect the quality of .wavs after multiple resampling jobs (that dont speed or slow the file), or is that solely up to the bit depth?
and also, will 24 bit depth become "lofi" after 4 or 5 resamplings?

a lot of the information on both sample rate and bit depth only talks generally and definitely not at the level of a bored producer running something through an effects chain 1000 times. any good links will be appreciated

it depends on how the fx chain processes the original file and what kind of artifacts/affect it has on the source file while processing
so it really varies situation to situation and what type of engine each piece in the signal chain is using as it's only as strong as it's weakest link

deadly habit wrote:it depends on how the fx chain processes the original file and what kind of artifacts/affect it has on the source file while processing
so it really varies situation to situation and what type of engine each piece in the signal chain is using as it's only as strong as it's weakest link

understood..shit will beget shit

but lets assume that all effects are native to the program and at 64 floating point.

i know we all love to give the "it varies" but thats an easier answer to understand if you know what it varies to and from. theres no information as far as i can find as to how audio degrades under certain effects repeated.
What I do know for certain that 16 bit 44.1khz does not last very long on any type of repeated resampling, it quickly turns to a tinny hollow POS sample in my experience.
But 24 bit 44.1khz doesn't seem to be cutting it in the long run either, or maybe its just me.

i mean the best way to test it is through experimenting either just with your daw's engine or just load up the plug you want to test and zero out the settings so it's doing nothing, but still routed through it

deadly habit wrote:i mean the best way to test it is through experimenting either just with your daw's engine or just load up the plug you want to test and zero out the settings so it's doing nothing, but still routed through it

oh i planned on doing that, no worries there, but i just wanted to know a couple of examples and some personal experience before delving into it.
so lets put it this way: when was there one instance for you when any one effect was affected not only by bit depth but by sample rate as well?

also whats a good standalone program for analyzing waves only? no editing, just different spectrum views and such? the waveforms drawn in ableton aren't always superbly accurate and i want to test the A/D converters on some hardware playing a sample versus the computer playing it.

honestly i've never been that concerned or worried about it to judge beyond what my ears hear
just use whatever audio editor, oscilloscope and spectral tools you want. most offer zoom in/out features

deadly habit wrote:honestly i've never been that concerned or worried about it to judge beyond what my ears hear

typically i haven't been either, but i tend to get into the zone when sound designing disregarding all questions of technicality or sound degradation so im trying to learn more. ive had some great resampled sounds that ended up pretty much unusable with out serious layering and disguising, that i suspect if i started with higher bit depth/sample rate would have been more usable.
especially if im working with my sp and kp, any samples on them get resampled so many times that sometimes the recorded the result is a much lower quality from the sound playing directly out of the boxes and into the monitors.

gotcha
might want to take you cable and power situation into account too man
the joys of hardware and variables it tosses into the equation

I'd guess that all the degregation is happening at the sp kp level. That is the bottle neck in terms of sound quality.

When you are done with that sp, lemme know.

nowaysj wrote:When you are done with that sp, lemme know.

NEVARRRR

"Resampling", well, that depends what you mean. Bouncing audio, no, not perceptibly at 24 bit, because all you're doing is quantizing each sample's value by some tiny amount. Any sample rate conversion, though *can* cause some degradation towards the upper frequency ranges if a poor interpolation method is used, but most DAWs are better than that these days (as long as you tick "Hi-Q" in ableton when warping clips for example). Artificial pitch-shifting or time stretching (as in, changing pitch without stretching the tempo, or vice versa) always causes degradation.

Simply bouncing to 24 bit WAV, I wouldn't worry about. The quantization noise floor is way below the level where it matters. I wouldn't ever do anything at 16 bit though, except for the final CD master. Anything else (stems, mixdowns, master to vinyl, etc) 24 bit.

BTW, EQs will usually have a different frequency curve in the top end if you run at 44k or 96k, but this isn't really a problem as long as you adjust the curve however you see fit. Distortion/waveshapers almost always oversample the signal internally and downsample the result, otherwise you get pretty nasty aliasing. Well-designed FX should sound more or less the same at all sample rates, but that's not always the case, especially with freeware.

The degradation from 4 or 5 resamplings vs the degradation caused by the processing itself is fairly insignificant. If you take a sound and do stuff to it repeatedly, you can't help but damage it. All processing has its detrimental effects, you're more than likely hearing that accumulate, as opposed to the direct effects of SR/bit depth.

That's not to say that higher sample rates can't help, particularly with effects that introduce harmonic distortion, cos of reduced aliasing. Having an aliased result processed by something causing more aliasing etc etc will of course give you a shitty result. Most EQs these days are decrampled in some way so as to have correct curve shapes at the top end, so oversampling in EQs isn't as prevalent or necessary as it was a few years ago, but obviously the same doesn't hold for harmonic distortion-inducing plugs.

Either way, processing a sound repeatedly at whatever sample rate will still mess it up.

Your proposed ADC test is flawed, by the way.

God this post is dry as bollix innit. Bit monged from 2 days on the lash, sorry

More important than op's ot, is what the heck does "2 days on the lash mean?"

nowaysj wrote:More important than op's ot, is what the heck does "2 days on the lash mean?"

One can not describe "2 days on the lash" in words alone.
It must be experienced.

JemGrover wrote:

nowaysj wrote:More important than op's ot, is what the heck does "2 days on the lash mean?"

One can not describe "2 days on the lash" in words alone.
It must be experienced.

it means spending two days out on the piss getting into all sorts of mischief like eating out of public toilets and fingering young girls.

French girls, otherwise fairly accurate.

the plot thickens
sounds like someone had fun

That really doesn't even cover a quarter of it

from my experiences, it doesn't seem like it's an issue anymore.

when i used to bounce stuff "wet" (with all the fx on) in logic 7, i definitely noticed a loss of warmth and presence in the sound.

these days on logic 9 i don't hear that same loss in quality. i don't know if my skill improved, or somehow logic improved it's audio bouncing..but it's just not there anymore.

i generally bounce my audio dry then put the fx back on the audio file afterwards. i don't really resample anymore unless i really do want to fuck up the sound...which means a loss of quality isn't really a massive problem anyways as the sound itself isn't going to be a main focal point in the mix.