Master volume difference between 44.1kz and 96khz

Found a bug ? Report here

Re: Master volume difference between 44.1kz and 96khz

Postby gimmeapill » Fri Jan 27, 2017 10:56 pm

Yes, it works fine ;-)
The volume of Guitarix Amp is now the same at 44.1k and 96k, even with presence at max.
The sound at 44.1k is also a lot better, I can hardly hear any difference now.

Good job!
User avatar
gimmeapill
 
Posts: 80
Joined: Mon Jan 18, 2016 11:38 am
Location: Czech Republic

Re: Master volume difference between 44.1kz and 96khz

Postby gimmeapill » Sat Feb 11, 2017 10:11 pm

Update: I have not yet tested back Guitarix standalone at 44k, but I found the same problem with the GxAmplifier-X LV2 plugin: In order to use it at 96k wihtout blowing up the meters, I have to keep the master volume just one notch above zero. To get equivalent volume at 44k, I have to raise it to noon.
Could this one get the fix as well?
User avatar
gimmeapill
 
Posts: 80
Joined: Mon Jan 18, 2016 11:38 am
Location: Czech Republic

Re: Master volume difference between 44.1kz and 96khz

Postby brummer10 » Mon Feb 13, 2017 1:33 pm

I'll check it, . . :metal2:
User avatar
brummer10
Site Admin
 
Posts: 678
Joined: Thu Mar 26, 2009 6:57 pm

Re: Master volume difference between 44.1kz and 96khz

Postby gimmeapill » Tue Feb 14, 2017 9:20 am

Update: I finally had some more time to test more patches, and there's still around +8/10db difference between 44.1 and 96k for anything that uses cab sim + tonestacks.
The Guitarix Amp itself seems to be ok now (I need to test a bit more to be 100% certain).
Regarding the tonestacks at 96k: there are also significant volume difference depending on what is chosen. Some of them are very loud (Bassman, Mesa Boogie) and some 10db quieter (Engl, Roland). I assume some of them use the hardwired resampler?

Another subjective observation: With only the Guitarix Amp, sound is about the same at 44.1 and 96k.
But as soon as adding cab sim and tonestacks, I like it way better at 96k. It just sounds more detailed and lively.
User avatar
gimmeapill
 
Posts: 80
Joined: Mon Jan 18, 2016 11:38 am
Location: Czech Republic

Re: Master volume difference between 44.1kz and 96khz

Postby brummer10 » Tue Feb 14, 2017 1:20 pm

gimmeapill wrote:Update: I finally had some more time to test more patches, and there's still around +8/10db difference between 44.1 and 96k for anything that uses cab sim + tonestacks.


Yes, the cabinet needs the same fix then the presence unit. Haven't come around to implement it. Will do this week.
gimmeapill wrote:Regarding the tonestacks at 96k: there are also significant volume difference depending on what is chosen. Some of them are very loud (Bassman, Mesa Boogie) and some 10db quieter (Engl, Roland). I assume some of them use the hardwired resampler?

No, they ain't resample at all. The volume difference between the models is intentional, they are modeled from hardware, were you've the same difference in volume between them. Are you sure that the volume difference is related to the used frequency, or do you mean the difference between the models?

gimmeapill wrote:Another subjective observation: With only the Guitarix Amp, sound is about the same at 44.1 and 96k.
But as soon as adding cab sim and tonestacks, I like it way better at 96k. It just sounds more detailed and lively.


I hope we could solve that when the cabinet use the oversampling as well.
I'll post a notice here when done.

regards
hermann
User avatar
brummer10
Site Admin
 
Posts: 678
Joined: Thu Mar 26, 2009 6:57 pm

Re: Master volume difference between 44.1kz and 96khz

Postby gimmeapill » Tue Feb 14, 2017 8:55 pm

No, they ain't resample at all. The volume difference between the models is intentional, they are modeled from hardware, were you've the same difference in volume between them. Are you sure that the volume difference is related to the used frequency, or do you mean the difference between the models?

I was talking about volume difference between the models at the same frequency (96k in that case), so exactly as you said. I find the volume difference a bit too dramatic for something you switch with a drop down menu. But now that you mention it I should probably test also at 44.1.
User avatar
gimmeapill
 
Posts: 80
Joined: Mon Jan 18, 2016 11:38 am
Location: Czech Republic

Re: Master volume difference between 44.1kz and 96khz

Postby gimmeapill » Tue Feb 14, 2017 9:58 pm

Tested as well at 44.1k with a few patches where I balanced roughly the volumes betwen tonestacks: the behavior is consistent regardless of the frequency, so no issue there.

Another subjective observation: With only the Guitarix Amp, sound is about the same at 44.1 and 96k.

There I spoke too fast: the volume issue is fixed but with headphones I can still hear a tone difference. 99% sure it comes from that Presence control cranked to 11 - but please don't touch it, I Iove it this way ;-)
User avatar
gimmeapill
 
Posts: 80
Joined: Mon Jan 18, 2016 11:38 am
Location: Czech Republic

Re: Master volume difference between 44.1kz and 96khz

Postby brummer10 » Thu Feb 16, 2017 7:56 am

brummer10 wrote:es, the cabinet needs the same fix then the presence unit. Haven't come around to implement it. Will do this week.


So, I've pushed the fix to git, now, the cabinets (mono/stereo) and the ampimpulse using the upsampling as well. Now I need to check gx_amp and jcm_800 LV2 plugs.

regards
hermann
User avatar
brummer10
Site Admin
 
Posts: 678
Joined: Thu Mar 26, 2009 6:57 pm

Re: Master volume difference between 44.1kz and 96khz

Postby gimmeapill » Mon Feb 20, 2017 10:14 pm

Hi Hermann,

Good news, the fix works: I tried a few presets with the tonestack + cab sim, and the volume is now identical at 96k and 44.1k (altough the tone is still slighty different).

Bad news: there are side effects at 44.1k. I thought the change would make the overall volume quieter at 96k, but that's the opposite: nothing changed at 96k and everything is louder at 44.1k. This means that factory presets are now at least 10db too loud... That is not going to be a very popular change ;-)

I'm not sure what should be fixed then. Recalibrate the matser vol? Roll back the change?

BR,

LX
User avatar
gimmeapill
 
Posts: 80
Joined: Mon Jan 18, 2016 11:38 am
Location: Czech Republic

Re: Master volume difference between 44.1kz and 96khz

Postby brummer10 » Fri Mar 03, 2017 6:08 am

Hi

The culprit of this sitting deep in our ffmpeg-convolver implementation.
Best fix would be to forward the samplerate to the ffmpeg-convolver, in order to adjust the normalisation factor between ffmpeg and fftw3 on base of the Samplerate. Yes, they create the convolution data on different levels, and we adjust the difference between them on base of 48khz.
Unfortunately, when we froward the Samplerate to the ffmpeg layer, it would introduce a incompatibility with the original zita-convolver.
Right now, most distros using the zita-convolver implementation, and I'm not sure if I'm willing to skip support for it, even if it leads to some trouble with the LV2 plugs. The fftw3 devs seems to fix the issue with the Thread Safe Planner within the next release, which will reduce the pressure to use ffmpeg at all (reducing dependency's then).
Then, the ffmpeg layer only will be needed to run smooth on ARM devices, which mostly run on 48khz anyway.

regards
hermann
User avatar
brummer10
Site Admin
 
Posts: 678
Joined: Thu Mar 26, 2009 6:57 pm

PreviousNext

Return to Bug reports

Who is online

Users browsing this forum: No registered users and 1 guest