Login

VNXT EMT Sound Bad?....

Officially Licensed 3rd Party Developer Libraries
Free 3rd Party Programs

VNXT EMT Sound Bad?....

Postby Jozelo » Fri May 28, 2010 2:44 am

I usually work at 44.1kHz / 32 bit, but I was looking at several discussions on the internet that many said that the plugins work better in higher clock, 88.2kHz and 96kHz sample. I use AlexB libraries and all are 96Khz and he recommended to mix in 96Khz, I give it a try.
I did not see a significant change mixing 88.2 or 96 (Yes, more cpu usage), but I note that the EMT VNXT library at 44.100Khz doesn't sound anything like 96Khz. I used sample EMT140 VNXT 1s and 96Khz sounds like a 1s instead when in 44.100Khz sounds much more than 1 s. Further there is a difference of 6db and sound detail. For more detailed information you can download the examples that I upload.

My simple question is why :?:
I use:
Windows 7 64 Bit
Reaper
Nebula 1.3.401

Saludos
Attachments
Test VNXT EMT_44100.wav
VNXT EMT 140 sample rate 44.100
(2.43 MiB) Downloaded 148 times
Test VNXT EMT_96000.wav
VNXT EMT 140 sample rate 96
(5.28 MiB) Downloaded 140 times
Test VNXT EMT_96000_MatchingLevel.wav
VNXT EMT 140 sample rate 96 Matching levels
(5.28 MiB) Downloaded 168 times
Jozelo
User Level I
User Level I
 
Posts: 12
Joined: Tue Mar 30, 2010 6:43 pm

Re: VNXT EMT Sound Bad?....

Postby zaminx » Fri May 28, 2010 8:34 am

Is your rate convert in the nebula reverb Master page set high enough?
zaminx
User Level XI
User Level XI
 
Posts: 182
Joined: Sun Mar 28, 2010 6:57 pm

Re: VNXT EMT Sound Bad?....

Postby fuseburn » Fri May 28, 2010 2:43 pm

I can reproduce this, I think it's a bug.
Same setup here (Win7 64, Reaper 64, Nebula 1.3.401 x64)

MAST-Page: Try increasing L FREQD to like 6 seconds (that's the length of reverb tails to be computed), then increase RATE CNV to...higher value (not sure what that does exactly, but I set it to 2000 ms, just to be sure). Don't forget to hit save and then reinstantiate nebula (Ctrl-Shift Click on the plugin to make it offline, then do it again to bring it online again, settings should be recalled now). All parameters should be fine in order to play back the EMT program.

Now do the following: Open an empty project, make sure you're running on 44,1 kHz (Change samplerate with alt-enter project settings to your desired value if needed)
Now open any eq program which is sampled @ 44,1 khz (check on prog-page, it should say FRT: 44100 Hz -> 44100 Hz), dial in a narrow boost which is clearly audible. Listen to it, get used to it.
Now change the samplerate to 96000 Hz (alt-enter, Reapers project settings)
Hit Play again - Prog page still says FRT: 44100 Hz -> 44100 Hz; the boost is more than 1 octave higher, so youre listening to a wrong representation of the preset.
Now reinstantiate Nebula (as explained above, 2x ctrl-shift-click, set it offline and online again). Now Prog page says FRT: 44100 Hz -> 96000 Hz - that's correct, and the eq boost is exactly where it's supposed to be.
From here you can do the same game backwards (set samplerate back to 44100 Hz and listen to the EQ-boost which is more than an octave too low), but you get the image. Always refresh all Nebula instances after you've changed the samplerate. Closing and reopening should do that automatically.
-> Nebula's resampling seems to be working.

BUT now to your problem with the EMT which is in fact a program sampled @ 96000 Hz: The case is different here, because if you're running your session on 44,1k, you're actually DOWN-sampling the program - and that seems to be faulty in this particular case.
Open the EMT-plate demo (2s 96k) in a 44,1k-session. Play back a short signal and measure how long the tail is - something around 3,7 seconds in my case. Sounds exactly like your first sample.

Now try to do it in the right order: Set Nebula offline, change SR to 96k, activate Nebula again. Hit play. Tail is now around 2 s, way brighter, way shinier, in fact exactly like your 2nd sample. Better sound details, like a real Plate.
I took it further and switched to 192k - and its even brighter, and tail is 1s long.

I tried the same with some other Reverb programs @ 96k (CDSoundMaster Producer Pack 2 Demo TrueHall @ 96k) - and it resamples fine (if you do your homework as I described, nebula offline-online-procedure). It sounds more or less identical on any samplerate.

The case is clear to me: VNXT EMT plate doesn't resample at all. It only works properly on 96k. We might want to verify that on other hosts and x32-systems etc.
Funny, because in every comparison against UAD-Plate and L*****n PCM on Gearslutz etc., where the VNXT-plate often came out as the winner, they definitely used the sound of the 44,1k nebula EMT plate - which is actually downsampled ! :lol:
User avatar
fuseburn
User Level X
User Level X
 
Posts: 133
Joined: Mon Apr 05, 2010 6:19 pm

Re: VNXT EMT Sound Bad?....

Postby david1103 » Fri May 28, 2010 5:57 pm

well, your files had a huge difference :)!

i did my own test with 32 bit nebula 1.3.277, its set to the highest conversion rate number.

board attachment limit run out, here's my direct link (3meg):

http://www.sendspace.com/pro/dl/wkox52

i didn't follow your exact procedure, just 1st set daw to 44.1 and exported the 1st file. then set to 96, restarted the daw and re-loaded nebula. then exported the 2nd file. last i converted the 96k file to 44.1 with the free version of r8brain.

nothing like what you had, but it is very interesting that the 44.1 downsampled version is quite a bit inferior.

listen to the strange bass range distortion sound vs the 96k version. also lack of overall quality. what do people make of this difference? i am listening on benchmark dac1 with hd580 headphones.

does anyone think the r8brain downsampled version is better?

its possible that when cubase upsampled my 44.1 snare sound it changed it to create the bass distortion effect in the reverb, but afraid i run out of test time now. seems doubtful :)
User avatar
david1103
Beta Tester
Beta Tester
 
Posts: 516
Joined: Wed Mar 31, 2010 12:26 am

Re: VNXT EMT Sound Bad?....

Postby futur2 » Fri May 28, 2010 6:42 pm

wow that's quite a difference! :shock: i have to confess i didn't notice it as i have not used it so much and never tested the emt against anything. it always sounded good to me. but i want my 44.1 to sound as good as the 96 :evil:

nextaz?

i wonder if the alexb and cdsoundmaster programs behave the same :?
futur2
Member
Member
 
Posts: 303
Joined: Sun Mar 28, 2010 8:16 pm

Re: VNXT EMT Sound Bad?....

Postby fuseburn » Fri May 28, 2010 9:21 pm

I also think that 96k sounds best. R8brain is the worst (don't like R8brain at all). 44,1k is quite a difference, yes.

Btw. I managed to get rid of the faulty behaviour that Jozelo described (the pitched tails). Jozelo, just set RATE CNV to 9000 ms and you'll be set - the plate will sound identical on all samplerates now (apart from the sample-rate conversion itself, which the other guys are talking about :D )
User avatar
fuseburn
User Level X
User Level X
 
Posts: 133
Joined: Mon Apr 05, 2010 6:19 pm

Re: VNXT EMT Sound Bad?....

Postby Jozelo » Fri May 28, 2010 10:13 pm

Zaminx and fuseburn Thanks for the help, I set the Rate CNV to 9000 and everything works great.

Saludos
Jozelo
User Level I
User Level I
 
Posts: 12
Joined: Tue Mar 30, 2010 6:43 pm

Re: VNXT EMT Sound Bad?....

Postby futur2 » Fri May 28, 2010 10:48 pm

fuseburn wrote:I also think that 96k sounds best. R8brain is the worst (don't like R8brain at all). 44,1k is quite a difference, yes.

Btw. I managed to get rid of the faulty behaviour that Jozelo described (the pitched tails). Jozelo, just set RATE CNV to 9000 ms and you'll be set - the plate will sound identical on all samplerates now (apart from the sample-rate conversion itself, which the other guys are talking about :D )


thank you! i now see it's in the manual but there it says: "If You have any problems with loading programs when You are not working 96khz – go to MAST page and change increase RATE CNV to min 6000 (i suggest max 9000)." as there was no problem with loading i maybe would never have realized there IS a problem at all ;)

i had the rate cnv set to 3000 before as suggested for the alexb libraries. even after reading the FAQ i'm not really sure what the rate cnv parameter does. this are the kind of parameters/settings that sometimes make nebula an unnecessarily difficult experience. :roll:
futur2
Member
Member
 
Posts: 303
Joined: Sun Mar 28, 2010 8:16 pm

Re: VNXT EMT Sound Bad?....

Postby studiorecording » Thu Jun 03, 2010 4:57 am

Thanks for the input.
Shifting to 96k really helps the sound. I did raise conv. to 9000 in the mast tab but when I raised the L FreqD above my original 2000 setting my computer stalled on EMX140 reverb and if I lowered the LFreqD it would work.
So now I am confused.
What am I missing?
studiorecording
User Level 0
User Level 0
 
Posts: 4
Joined: Sat Apr 03, 2010 7:10 am

Re: VNXT EMT Sound Bad?....

Postby david1103 » Thu Jun 03, 2010 12:41 pm

having long L FreqD (length of reverb) at 96k you need a super computer! my quad core 6700 struggles.

i have 2 nebula reverbs, one with dsp buffer 1024 and the other about 8192. you need the big buffer to use stuff at 96k but you get the extra latency. you change this setting in the xml file, its mentioned in the forums.
User avatar
david1103
Beta Tester
Beta Tester
 
Posts: 516
Joined: Wed Mar 31, 2010 12:26 am

Next

Return to 3rd party libraries

Who is online

Users browsing this forum: No registered users and 2 guests