Bitwig Sampler Cuts frequencies > 15k (majkel was right)

[reflekshun]

Thanks for the replies guys! Let me address some of your concerns:

I'm 35, and I can hear the difference in the sound in this case without question. It's not a maybe, its a definitely. 15k isn't out of the range of hearing for most people. Remember, what we hear doesn't have an ON/OFF switch above a certain frequency. There is a roll off, meaning we can still hear it, but at a much lower volume. At the same time I acknowledge at a certain point it is so soft that we essentially can't hear it.

The argument that everything above 15k isn't worth worrying about, I can't say I agree with that, but that's just my opinion Show an engineer that every layer on their DAW is gonna have that cut and let me know how you go

@meta-redundant: Good observation, the weighting of my freq analyzer is set by default to have that slant - it's tuned to pink noise so that a good mix will have a flat curve, but pink noise itself will end up having that upwards slope. Sorry that it's a confusing thing for this test! But I find the relative difference between two cases is still very clear.

@hez: I think that theory is the most likely, a way of anti-aliasing. Kinda strange way to do it if you ask me, I'm pretty sure other digital samplers don't do this, but then again I haven't tested that!

OR maybe it's a freaking amazing high end oldskool modelling of the classic EMU sampler.

[meta-redundant]

You asked

This is what I did:

1: go to http://www.audiocheck.net/testtones_pinknoise.php
2: download audiocheck.net_pinknoise.wav
3: load up a sampler
4: drag in audiocheck.net_pinknoise.wav
5: load Voxengo SPAN
6: play the audiocheck.net_pinknoise.wav with a midi note
7: look at the same results OP and majkel gets. Trust me, I don't want majkel to be right.

So what gives?

Thanks for posting this. When I do it that way, with that specific file, I do get a slight rolloff. When I use any of the other pink noise files on their site, or I use a pink noise file from another source (including one I print myself), it doesn't happen. There is something about the content of that file which is interacting with the sampler. I suspect the file itself has a slight rolloff that is compounded by the sampler's antialiasing filter. Can you try it with other pink noise?

[majkel]

i did..
with pink noise 88,8khz doesnt roll off

[Hez]

There's some recent research emerging regarding the importance of ultrasonic frequencies to humans' perception of pitch/timbre/harmony. As I understand it, most of the tests back in the day regarding the range of human hearing were done on sine fundamentals/harmonically not particularly rich waveforms. I think the latest research is related to our ability to detect subtle changes in timbre of brass instruments and cymbals where we appear to be deriving information from partials above 20k.

Could all be BS, I don't think I properly fact checked it at the time (oops ;p), and obviously a lot of digital formats aren't capable of accurately representing information above that cutoff point anyway (44.1Khz renders for example), but I found it interesting conceptually nonetheless.

[mosska]

it was mentioned here a few years ago. Just load a kick drum into sampler and bounce it. Then compare transients (enlarge it) of the original kick and bounced one. you will see the difference.

[reflekshun]

I think the conclusion is this:

Everyone can see the difference, many people can hear the difference. Whether that frequency discrepancy matters to you, that is a personal thing. Either way those of you who care about it, it's better to be aware of the details of your tools when making mixing or production decisions.

[JHernandez]

Can you try it with other pink noise?

Generating my own @44.1 gives me the same results. @ 48 it's much better but still a bit of roll off way up high. The 88.2 and 96 files at that site are good.

I used two random generators that google came up with, pink and mda testtone FWIW.


I'm going to give it another shot when I get a chance as I rushed through those, life called. I think I noticed something in the low end when I bounced that 48 file. Not sure though.

[stamp]

Poise sampler does the same and developer told me it was because of antialiasing.
I'm 35, like the OP, and my hearing goes up to 16.6khz (tested). And I hear a difference.

This should be fixed.

Cheers!

[strovoknights]

I'm try HOFA System PinkNoise generator with Pro-Q2 is working fine.
After that,HOFA PinkNoise to audio convert for BW Sampler.

PinkNoiseGenerator_WorkingFine.JPG

PinkNoise_Samppler.JPG

[craigtumps]

So the average adult cant hear 17k, only teenagers. So the roll off doesn't matter?
Well this is quite offensive to teenagers, they listen to music too you know. Why shouldn't they have crystal clear highs!

[Netsu]

I'm sorry, I didn't realise I was speaking to a toddler.

The "mosquito" device used in UK train stations emits a 17khz frequency, to deter teenagers from loitering there. Teenagers, not adults, because the average adult cannot hear a 17khz tone.

https://www.thetimes.co.uk/article/scot ... -djxst00c6

Please do a tiny bit of research before insulting me further. My hearing is fine for an adult of my age.

Even though I hear a sharp loudness rolloff at 17khz I can still hear sounds up to 19khz and I'm 29. I'd say most people would notice when a track has everything above 15khz cut off.

[Hez]

I'm sorry, I didn't realise I was speaking to a toddler.

The "mosquito" device used in UK train stations emits a 17khz frequency, to deter teenagers from loitering there. Teenagers, not adults, because the average adult cannot hear a 17khz tone.

https://www.thetimes.co.uk/article/scot ... -djxst00c6

Please do a tiny bit of research before insulting me further. My hearing is fine for an adult of my age.

Even though I hear a sharp loudness rolloff at 17khz I can still hear sounds up to 19khz and I'm 29. I'd say most people would notice when a track has everything above 15khz cut off.

I've thought a lot about this recently as I've been listening to a lot of very soft/warm/lofi dub type techno that barely has anything above 10k in some instances.

If you balance your mix right, then even the odd hi hat or cymbal at 10-12k can 'feel' like a really pristine, open top-end element. Likewise, super slammed EDM where nothing is high-cut can sound really congested and like it actually has less high end than some of the dub stuff.

Recently I've been paying a lot of attention to low passing elements where required etc. to try to move the energy deeper into the mids when I mix, and I'm quite happy with the way it's progressing.

Obviously, a good mastering engineer critically listening between two tracks in front of good monitors will easily be able to identify which track is rolling off lower than the other. Likewise, there are some genres of music that make good use of the full frequency spectrum and don't just pack it out like overly-clipped dance music.

Don't really have much of a point here but I guess in summary:

1. It's good that people know how their tools are working and therefore I appreciate the fact that majkel dug this up as it's interesting to know.

2. Don't overestimate the importance of full spectrum information when it comes to releasing clean and pleasant sounding mixes. Just because some people can physically hear over 15k doesn't mean that retaining/adding information up there necessarily adds anything to their experience of the music.

[meta-redundant]

Generating my own @44.1 gives me the same results. @ 48 it's much better but still a bit of roll off way up high. The 88.2 and 96 files at that site are good.

Thanks for testing JHernandez. I don't see anything anomalous when I print my own internally - although my whole signal chain is always running at 48k (I have to deliver for picture).

Some perspective: I think it's important to realize that we are talking about the sampler's prefiltering, not anything related to audio tracks, mixing, summing, etc. I've been delivering things for mix for several months now without issue.

I think reflekshun put it best, a few posts up:

Everyone can see the difference, many people can hear the difference. Whether that frequency discrepancy matters to you, that is a personal thing. Either way those of you who care about it, it's better to be aware of the details of your tools when making mixing or production decisions.

In my case, I'm not using the in-built Sampler for anything other than the occasional one-shot, sometimes for chopping a pulse loop. YMMV.

[majkel]

i hope we will get a bug fix with 2.3.3
a simple , lightweight sampler its a basic stuff .. a sampler is a base of drumachine !! soo, its good to hear a clear and 'opened' hihats and rides
ive been a test 3rd part sampler 'grace'.. also very similar situation - roll off in hi freq

[pdxindy]

i hope we will get a bug fix with 2.3.3

My impression is that it is a design choice... not a bug