Xite @ 96kHz: some questions...

[at0m]

Hey gentlemen, I am not interested in the pro and cons of 96kHz here, let everyone subjectively deceide for their own, but in Xite's capability to do 96kHz. Since it is practically impossible to run full-blown projects on Scope PCI at 96kHz, if Xite could manage, it would be the only reason for me personally (and probably I'm not alone in that) for getting one. Because apart from being able to load bigger projects, Xite isn't offering nothing new for me if it can't do 96kHz.

XITE-1/4LIVE wrote: Alfonso's 96k version of MoJo is the only recently made device that I am aware of that was designed to be used in 96k.

MoJo's circuit makes use of the inherent delay caused by specific samplerate. This is the reason why the 44.1 version sounds different at 96kHz, and why Alfonso released a 96kHz version that makes up for the different latency: Now both versions consistently sound the same.

99% of the devices out there are not influenced like that by the samplerate, their only difference at 96kHz is that they benefit from the higher resolution for their oscillators (cfr. aliasing), filters etc. But Mojo needed to be re-tuned for the smaller (PCI?) delay that comes with the higher samplerate - or it just wouldn't have sounded the same... Why not just try the original version at 96kHz?

[w_ellis]

Thanks for the reply at0m. I had heard things about the internal benefits for synths of running at 96khz for filter aliasing etc, but I was interested as to whether anyone had experienced these benefits in the real world.

garyb: I'm totally onboard with 96khz being a waste of resource for mixdowns etc, especially for electronic music. I'd like to know whether it helps internally with devices though.

For now, I can't see it being a big deal, so I'll carry on at 44.1, but I would welcome any other thoughts on this.

[garyb]

some 3rd party developers have already begun optimizing for XITE. dsp loading has also been imroved several times recently. PTHD does not allow 100% use of dsps, the same with XITE. one thing is certain from tests done recently, XITE will load a ridiculous crapload of devices. 96k works fine from the reports i've gotten, as i said, 96k is more than double the use of all computer resources from 44.1k, but it does work, for the most part, the same as with PCI, there may be a few devices which do not function properly at 96k, but that would not include core devices.

[spacef]

just wanted to point out that all and any spacef-device plugin works @ 96khz without a single difference in sound or function compared to 44.1...this inlcudes all delays since Echo 1 and with all bpm functions (tempo sync, ulli, etc). if a spacef-plugin doesn't work, it is because 96khz makes it too big for the hardware, but then you need more dsp and you will be able to load it. You will not find a dramatic difference in sound of the plgins between 44.1 and 96khz inside scope, it is only when it is bounced to audio track that your samplerate and bitdepth will make a difference (especially bitdepth). it is the case of 99.9% of the scope plugins anyway... I don't know what the purists say, but it is more clever to begin to use higher bitdepth in your daw, it will upgrade your sound already. I love the sound I get from a 24bits cubase project mixed through asio16bits of scope (and recorded through asio 16bits): thick sound.... very good recorded sound, i just have to pay attention to levels in cubase (-3/6dB on most outputs) music professionals really like my sound (i am trying to get there money, so i will not change this very weird combination any time soon ). If you don't get a very good sound at 44.1/24bits (which you hear at scope internal 32bits resolution), then it may not be better at any higher samplerate... I am pretty sure it is a better idea (and more expensive) to upgrade to new studio monitors to get a better sound than you already have. But it is also true that i don't really go to 96khz for technical reasons rather than pure "educated opinion" (i mean, you need double dsp and high HDD speed/RAID if you really want 96khz).
(edit: i highlight because someone is trying to prove to me that a sample @ 96khz is better than 44.1, which is not really my point, while it also confirms what i just said)....

[iSiStOy]

MoJo's circuit makes use of the inherent delay caused by specific samplerate. This is the reason why the 44.1 version sounds different at 96kHz

Wishes for discovering new sound frontiers? That's a point

it is more clever to begin to use higher bitdepth in your daw, it will upgrade your sound already. I love the sound I get from a 24bits cubase project mixed through asio16bits of scope (and recorded through asio 16bits): thick sound.... very good recorded sound, i just have to pay attention to levels in cubase (-3/6dB on most outputs)

Hmm... The difference really gets to my ears if I do record with 24bit asio and then mixdown to 16bit. Can't do this in real time though. I also have to pay attention to volume as it seems you get more drastic amplitude changes (dynamic?) in recorded sounds.
48KHz recording changes a bit as well, when compared.

[siriusbliss]

So far 96K is working great here in initial tests (Samplitude) on acoustic instruments.
I really only see advantages in acoustic material.
I still think it gets redundant with internal instruments.
I'm normally all 48K/32 most of the time anyways.

DAS and Spacef devices are rapidly being optimized for DSP distribution in Xite as well.
Overall DSP distribution in latest 64-bit release has noticably improved overall, and the project 'ceiling' is higher.
Just to see how many bxDigital V2's are loading now is pretty staggering.

Atze, Xor, and Flexor overhead is much better.
Just waiting on Bowen's updates for better DSP distribution in Xite (without having to manually assign DSPs).

Going for live performance/tracking using ModIV, Atze, P5, Flexor, and Vectron, as well as Superior2 and Kontakt running in one project is fairly insane.

Overall I'm happy.

Greg

[spacef]

Hmm... The difference really gets to my ears if I do record with 24bit asio and then mixdown to 16bit. Can't do this in real time though. I also have to pay attention to volume as it seems you get more drastic amplitude changes (dynamic?) in recorded sounds.
48KHz recording changes a bit as well, when compared.

in my example, there is no mixdown at 16bits, except when i make the final CD (export audio 16bits).
In my example, i have a better sound than using the 24bit asio. Theonly difference is gain at cubase outputs (less headroom) and I see this when i bounce back from cubase to cubase through scope, for prematsering for example. I have never had such a good sound - a sound that i like much more than using 24bit asio driver. May be it is me, it is very possible that i prefer this sound over the other one for subjective reasons. It is the only thing that really counts at the end, at least in my field (which is not movie scoring but dancehall stuff).

Of course i didn't say 48khz or 96khz is no difference; it is no difference inside scope for 99.99999% scope plugins, it is only when bounced as audio tracks that you will hear a difference between the 96khz and the 44.1khz samples (because the 96khz one was sampled at higher resolution, so obviously, there is a diffrerence in sample qualities).

gotta finish the 4th track of the mini LP
my sound setup is the following: lots of experience in mixing-->DAW 44.1khz/24bits->asio2-16bits source->scope32bits->asio2 16bits dest -> record back daw 24 bits-> final mixdown 44.1/16bits or 320kbs mp3 .

[iSiStOy]

Spacef, do you mean using 16bit asio with a 24bit cubase project, recording and mixing?

[spacef]

yes... 16 bits asio2 source/dest.... cubase project 44.1 / 24bits.
You know, I still love the sound that goes out of my EPS16+... most of them are 22khz/16bits... but through the EPS hardware, and fed into scope, I have barely heard better sounds (now, define what is "better" and we can go on for years ).... I also think aliasing is part of the sound, and it is in most of the vintage analog synths (noise, aliasing, etc, it is part of a sound). May be it is subjective and just coresponds to what i wanted to hear in this world.... what is clear, is that my mixes are much better now than they were before (when i was on 44.1/16bits everywhere). It sounds 24bits.

[at0m]

That's a lot of talk. Apart from all that, the proof is in the pudding. I'm posting a 30sec. audio file, containing 4 snippets, of a single saw oscillator with a slow LFO modulating the pitch: 2 notes played and recorded at 44.1kHz, and 2 notes played and recorded at 96kHz. I did not make anything complex, the recorded modular patch is as simple as can possibly be. Using more complex synths only makes the side-effects worse, and they are more noticeable at lower sample-rates.

For playback convenience I then converted the 96kHz snippets to 44.1. I know the sample-rate conversion makes the 96kHz sound worse than it would sound before the conversion.

To avoid further misconceptions, I do understand that for recording analog inputs, it does not make any noticeable difference. Guitars don't scream at such high frequency, and your singer isn't a bat. But I am not a guitarist, nor a singer (you'd love that one huh;) - I'm a synthesist. I live in the realm of oscillators, filters, shapers and envelopes. So apart from pre-occupied theoretical assumptions, or in case you've never tried to A/B yourself, just listen and I trust you will notice the difference immediately. Playing on the upper octave of the MIDI spectrum, you will still hear some artefacts, but the difference is quite noticeable. You will understand why Bowen and Klaus made Solaris run at 96kHz.

[The extension mp3 has been deactivated and can no longer be displayed.]

[spacef]

Thanks for taking the time atomkstudio
As I said, anything sampled at 96khz will probably sound better than the same at 44.1...(whatever the source, even guitars and analog stuff). But having heard the Solaris hardware a few times at past musikmess, I am pretty sure you cannot get the sound of solaris hardware just by putting a scope/xite on 96khz ? as you noticed, you still have the aliasing on high notes (on the FM oscillators for example, which aliaes a lot, i didn't find a difference when i tested it - which was the reason why i wanted to test (do i get rid of high notes aliasing @ 96khz? my findings were "no, i don"t hear a difference, so the loss in dsp power didn't seem worth it - or i would have gone 96khz probably) . I never achieved the clarity of the Solaris hardware with the scope atoms, nor heard anything close to the solaris hardware in scope plugins or in vstis (or in other hardware) - (but it is also true that i haven't used any scope synths for quite a long time now, may be i missed some stuff). I was under the impression that Klaus had to redo most of the atoms, starting with the scope models from John. But i haven't participated at all in the making nor have a clear memory of what we talked about so i could be wrong and don't want to talk for other people. Now, if i could go 96khz all the way (hard disks setup, soundcard etc) I wouldn't refrain doing so

[dawman]

That clarity is surely going to be noticable for giving better focused Low end.
I am sure that Modular patches will also get a cleaning in the high end too.
This was a nice way to prove a point..Thanks for time taken.
What effect would this have on Modules inside of Modular that carry audio...?

And yes I am a moron, so is there any effect on our effects, or envelopes like the AHD Loop or MultiSeg..? They loop nicely but the start and stop points are noticable if listened to closely.
I guess I am asking if this is a great all around upgrade for Modular...

BTW Brotha' Man at0m, I am immensely relieved you have Modular IV.
I look forward to even more patches using new modules that I can hardly pronounce much less understand their practical use.
RoyT has really helped out there, butthe more the merrier.

[Warp69]

You will understand why Bowen and Klaus made Solaris run at 96kHz.

It's of course possible to create beautiful bandlimited osc's in 44.1KKz - it's not a 44.1KHz issue as such - you'll still need bandlimited osc's in 96KHz.

[astroman]

... To avoid further misconceptions, I do understand that for recording analog inputs, it does not make any noticeable difference. Guitars don't scream at such high frequency, and your singer isn't a bat. But I am not a guitarist, nor a singer (you'd love that one huh;) - I'm a synthesist. ...

imho the 96k advantage (in the analog domain) isn't about more precise definition of high frequencies, but about avoiding the fold-back of aliasing artifacts. Those show up in much lower parts of the spectrum than a bat may use to find it's way to prey...
Of course it's questionable if that makes a (huge ?) difference on a miced guitar amp anyway... but the recording would gain clearity.
Certainly on vocals, but then... what do they sell as 'vocal recordings' today ?

cheers, Tom

[spacef]

on a miced guitar ..

http://d3b9cwalzc5eko.cloudfront.net/il ... guitar.jpg


highest sampling rates is about having more resolution, more samples per seconds, ie less "holes" in your samples (less missing information), whatever the frequency, it will necessarily sound better, especially on acoustic recording where there is no "resolution" issue because there is no samplerate or binary data involved at the source (is it "+ infinite khz" ?) .. your brain receives sounds through multiple inputs (ears, belly, skin vibrations) , then does the compression and conversion, at least on 3 dimensions, based on complex electric-organic impulses, not binary digital sampling.

even a "miced guitar" sampled at 192khz is still a "sliced" guitar ...

[Warp69]

highest sampling rates is about having more resolution, more samples per seconds, ie less "holes" in your samples (less missing information), whatever the frequency, it will necessarily sound better..............

That's incorrect - http://en.wikipedia.org/wiki/Nyquist–Sh ... ng_theorem

A samplerate of 44.1KHz can perfectly reconstruct all frequencies upto just below samplerate/2 (just below 22050Hz) - higher samplerate will not improve an already perfect reconstruction.

[spacef]

I will not discuss the theory, but then it would mean that 44.1 is enough for a "perfect" "reconstruction" of a guitar (a simple electric guitar for example, polugged direrctly in a scope with an adequate hi-z), but i will not agree one this from my own experience. There is a difference in the recorded sound and what you can do with it, it has always been a problem to me to amp a guitar with a software; i find a big difference between amping the input sound ("live") and amping the sampled sound afterword with the same preset. I can "hear" holes in the reamped sampled sound (it's like a bad noise like "crrrrrrrrrrrr" , it is awful, especially in dsp guitar amps by the way, it is impossible for me to reamp a guitar inside scope, even if the direct sound is ok through the very same plug. but i know that many people will disagree because they don't hear it for various reasons that i can't really explain.. sounds to me it is impossible not to hear this). I haven't tried alfonso's though...
I always thought that was due to a lack of resolution of 44.1, that would vanish at higher samplerates.
it would means also that, for most people hearing ability, you don't need more than, say 30khz sampling rate? as said, i can easily compare for ex, 22, 32, and 44.1 amongst others thanks to the EPS (has a lot of samplerates to choose from). there is a big difference in the resulting sounds on simple stuff, like a 808 snare drum for example which is not supposed to have tremendous high or lows (evnthough I tend to prefer the lower samplerates personally, because i like the sowehow harsh color it gives to drum samples, but that's just a matter of taste).
also in the article, at a moment, they seem to relate the "perfect reconstruction" to the interpolation formula used to achieve the result.
what do you think ?

[Warp69]

Im only talking about sampling and reconstruction, not DSP processing. Many DSP functions related to saturation and distortion benefits from higher samplerates.

If you have an analogue signal that don't have any information above 18KHz - then it won't matter if you sample the signal at 48KHz or 96KHz - there's no benefit in using 96KHz. But the processing afterwards can easily benefits from higher samplerates.

The interpolation is done by filtering in the AD/DA converter.

signal -> filtering (bandlimiting) -> AD -> xxxxxxx -> DA -> filtering (recontruction) -> signal

[spacef]

I feel releived that maths tell me that i do not have to upgrade my system to 96khz(my initial thought)
I guess i just need to get to 32 bitdepth drivers now...
cheers
Mehdi

[garyb]

what are we dealing with? audiophile classical piano and orchestra and fools with $100,000 turntables or is it pop music?...