The Upsampling Your Mix Thread

[Kingston]

or Waves R-verb??

the worst reverb on the market currently? uhmh.. well go right ahead.

Kingston did you mean on Voxengo tapebus.

yes. (of course) I experienced a bit of a renaissance quite recently with my sound when I emulated a tape multi track set up at 96khz with it. Some of it had to do with the fact I managed to match it *very* close to an Otari 5050 mkII mastering quarter inch (with emtec tape to be more precise). It sounds perfectly convincing.

[metamorphosis]

I've just spent the past day doing SRC comparison tests between R8brain, the EXCELLENT converter in Adobe Audition/Cool Edit, and another freeware/opensource effort called "Audiomove".
R8brain is excellent, and beats the lot-
I would rather use it than anything included in the standard hosts/audio engines, including Sound Forge (which is sh*t for sample rate conversion).
Particularly I wouldn't recommend using any of the VST plugs listed above - there's only a certain amount you can do with realtime processing before you have to start cutting corners in terms of audio fidelity. I think the SRC engine in Izotope Ozone (I own it) is not up to par with something like Cool Edit or R8brain.
Cheers-
m@t

[Kingston]

R8brain is excellent, and beats the lot-

if you meant the pro version, you'd be right.

but the free version is not that great. Not as good as audiomove for example: http://www.sg-acoustics.ch/digital_audi ... arison.pdf

the high frequency response is pretty average. Read the charts, and see the very round, ie. non-optimal low-pass filter of r8brain. The noise floor is marginally better than audiomove (meaning under -150dB). It's just that our ears are more sensitive to high frequency/transient response in this context.

Audiomove is also a lot faster since it's able to use multiple cores while processing (more than one file at a time as well).


In that chart, have a look at the weiss saracon graph. That's world class performance.

[bmanic]

bManiac, thanks for the example. Yes, that's the kind of artifacts that can occur in extreme, non real world examples, but I've still never heard one real world mixdown that actually benefits, and that's why I say it is placebo.

[Sam@Megablastic]

@Shy

A quote from the sonalksis forum. Posted Thu Feb 10, 2005. Trying to demonstrate how upsampling helps EQ plugins, in this case the Sonalksis EQ. Original file was 44.1kHz, upsampled with r8brain free to 192kHz, applied EQ, then downsampled again to 44,1kHz. The example .zip file has both the upsampled EQ version and the normally applied, at 44.1kHz, EQ version. The settings are extreme to show a clear difference so this is not a "real world" test. However, you can clearly hear the difference (listen to the transients and the "smoothness" of the audio around the cut/boost) so just imagine what upsampling a whole project can do for you. This is just one single plugin on a single audio clip.

Just to demonstrate the point, check these small snippets out.

2 files, both with identical EQ settings on the SV-517. One at 44.1khz and one at 192khz downsampled with voxengo's free r8tebrain to 44.1khz at maximum quality.

http://cmt.siba.fi/nisilen/eq_quality_demo.zip

Listen to the snare. It shows very clearly some nasty artifacts at 44.1 which are much smoother at 192khz. I admit that this is a way radical eq setting (+12dB highshelf from 10khz and -18dB mid band at 1.17khz with 0.40 Q). Also, it's not just the high frequencies that get better, the bass also benefits at higher sampling rates. The transients keep a bit tighter and more precise at radical boosts/cuts.

Remeber that all these artifacts add up. Using 20 sonalksis EQs in a track and rendering the whole thing at 192khz makes for quite a big difference to 44.1khz. Is upsampling needed? IMHO, yes, it would help but as I can always choose myself the samplerate when rendering this is not such a big issue.

Cheers!
bManic

He he..I've been preaching this for nearly 2 years and no one ever believed me. It's so obvious that working at 88/96khz has so many advantages. For all the non believers, do this simple test. Take a saw or square wave and put it in your favourite sampler with the resample algo set to normal at 44.1khz(i.e no oversampling).

Now play a really high octave until you can really hear the aliasing...now set your project sample rate to 88/96khz. Play the same sampler patch and ..oh noes...the aliasing has gone

Bmaniac is one of the few people I have ever seen notice that running at higher frequencies reall helps the bass end as well. For me it's just as important as the highs. 44.1 just not cut it if you use ANY plugins or vitual instruments. Not for me anyhow.

[BosseJo]

He's not that shy apparently.

[Colonel Flashback]

in a real world example, e.g. a typical band setup (unicorn playing trombone, leprechaun on the harp, goat as lead singer) it still would NOT make a difference...

[dusted william]

I thought tape buss uses it's own internal IR.

dw

[Kingston]

I thought tape buss uses it's own internal IR.

dw

you can turn it off. and in fact you *must* turn it off to get the best sound out of it. I'm personally responsible for a few of those IRs (the tape A, ie Studer b67) and quite frankly they just aren't up to todays standards.

[kmonkey]

I thought tape buss uses it's own internal IR.

dw

you can turn it off. and in fact you *must* turn it off to get the best sound out of it. I'm personally responsible for a few of those IRs (the tape A, ie Studer b67) and quite frankly they just aren't up to todays standards.

Why what is wrong with internal impulses (only one idea is on my mind and that is that they are recorded at 44khz )...

[egbert]

I thought tape buss uses it's own internal IR.

dw

you can turn it off. and in fact you *must* turn it off to get the best sound out of it. I'm personally responsible for a few of those IRs (the tape A, ie Studer b67) and quite frankly they just aren't up to todays standards.

So were you using no tape impulse at all with Tapebus or can you insert your own somehow?

[xander]

or Waves R-verb??

the worst reverb on the market currently? uhmh.. well go right ahead..

Waves IR it is then

I don't agree with you on R-verb (not sure of your reasons for saying such) -- it has some EXCELLENT plates and rooms, IMHO, with good CPU performance.

[Lagrange]

I guess it would not make a difference to upconvert the IR's then reload em?

BTW, still wondering how this process should effect vocoders and audiomorphers..

L

[xander]

I guess it would not make a difference to upconvert the IR's then reload em?

I think someone was saying that it turns them into mush or something. I haven't tried that but it would be interesting -- and it would be cool if it worked. Right now I'm playing with the Waves IRs at 96Khz.

BTW, still wondering how this process should effect vocoders and audiomorphers..

L

Yeah -- be interesting to see what's up with those -- if your reference wav file for the vocorder was already high-sampled, it would seem to make a diff but maybe it depends on the vocorder itself?

[Lagrange]

right