asb2m10 / dexed

DX7 FM multi plaform/multi format plugin
GNU General Public License v3.0
2.79k stars 242 forks source link

Feedback level is too high depending on the algo #146

Open MulB opened 6 years ago

MulB commented 6 years ago

Hi,

I installed dexed on my Mac (OS X 10.11.6). I tried the binary distribution as well as the versions that I compiled myself with XCode (BTW: compilation is really straightforward -- thanks for preparing the distribution so well). In all cases, starting from 0.9.3 upward, independently of AU, VST, or standalone, the last note of the initial preset 1. Say Again sounds strange. I found out that for 0.9.2 it sounds fine, but with release 0.9.3 and 0.9.4, the last note is seriously distorted. Can anyone reproduce this?

Thanks, Michael

asb2m10 commented 6 years ago

Hi,

Can you switch engine (from the PARM button) from Mark I to Modern to see if you have the same issue with 0.9.3+ ?

Thanks

PS: thanks for the build praised, it is not me, but rather JUCE that made it possible.

MulB commented 6 years ago

Hi,

thanks for the reply. I tried dexed-0.9.4hf1 binary distribution as well as compiled from source. In all 3 variants (Mark I, Modern (24 bit), OPL Series) the reported issue occurs.

Thanks!

MulB commented 6 years ago

Hi,

if this is helpful I can attach a wav file. The patch ``plays'' the 6 operators in Alg. 32 one after the other. The noisy sound comes from operator 6, which has feedback, and if I reduce the feedback amount from 7 to 4, then it sounds similar to version 0.9.2 with feedback value 7.

PS: the ease of compilation was in comparison, to, e.g., TAL NoiseMaker, which on newer OS X versions does not compile right out of the box (to get it compiled on Mac, it was easier for me to adjust the DISTRHO version).

asb2m10 commented 6 years ago

Sorry for the late reply. I've changed the feedback level from 0.9.2 to be on par with Algo 4 and 6 multi op feedback. This might be a work as design and I need to test this with my DX7. I'll keep you posted.

MulB commented 6 years ago

Thanks. I also tried the patch with Arturia DX7 V. There it sounds like in 0.9.2 and earlier (no distortion, i.e., presumably smaller feedback level). I also noticed that with the modern engine the distortion already occurs in 0.9.2, the other engines sound clean in this version.

Gesendet mit der GMX iPad App

Am 06.07.18 um 04:23 schrieb Pascal Gauthier

Sorry for the late reply. I've changed the feedback level from 0.9.2 to be on par with Algo 4 and 6 multi op feedback. This might be a work as design and I need to test this with my DX7. I'll keep you posted.

-- You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub: https://github.com/asb2m10/dexed/issues/146#issuecomment-402906244

Sentinel77 commented 6 years ago

Hi,

I can also reproduce this, on linux and windows. The feedback for Algorithm 32 is much higher on Dexed than it is on the Yamaha DX7.

asb2m10 commented 6 years ago

For 24-bit or Mark I ? the feedback is tuned for Mark I engine, not the 24-bit one.

Sentinel77 commented 6 years ago

Hi,

All of the engines including Mark I show this behaviour.

I am only seeing this on Algorithm 32 though. This algorithm is unique in that it has the feedback right on the carrier operator and no modulator operator. Maybe this has something to do with it.

I have recorded the Yamaha DX7 for comparison, isolating operator 6 on algorithm 32 with full feedback.

VOICE INIT Turn on only Operator 6. Turn all others off. Operator 6 Output Level = 99 Feedback = 7

dx7.zip

MulB commented 6 years ago

Hi,

 

then my observation is confirmed also on other platforms.

The problem description was given twice now: The (too)

strong feedback in Alg 32 occurs for all engines in the new

version and I am observing it also for the Modern engine in 0.9.2.

Is there a motivation to have the feedback that strong in Dexed?

 

 

Gesendet: Montag, 30. Juli 2018 um 04:00 Uhr Von: Sentinel77 notifications@github.com An: asb2m10/dexed dexed@noreply.github.com Re: [asb2m10/dexed] Strange sound on MacOS X (#146)

Hi,

All of the engines including Mark I show this behaviour.

I am only seeing this on Algorithm 32 though. This algorithm is unique in that it has the feedback right on the carrier operator and no modulator operator. Maybe this has something to do with it.

I have recorded the Yamaha DX7 for comparison, isolating operator 6 on algorithm 32 with full feedback.

VOICE INIT Turn on only Operator 6. Turn all others off. Operator 6 Output Level = 99 Feedback = 7

dx7.zip

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

asb2m10 commented 6 years ago

@Sentinel77 , there is work to do between algo 4-6-32 ; the level must be different, depending on the number of "feed" operator. Will work on this on the next release.

showlabor commented 4 years ago

Hi, I just got an original DX7 Mk1 and almost immediately stumbled across the strange noise produced at the end of "Say Again" in Dexed 0.9.4 but not the original DX7 Mk1. Has anything been achieved in the meantime? Would it be worthwhile to build the current git version or is it still behaving like 0.9.4 in this regard?

mkruselj commented 3 years ago

It still seems to be happening.

waltercruz commented 2 years ago

Still happening.

simonrapaka commented 1 year ago

still happening

showlabor commented 1 year ago

Which version of Dexed are you using? Are you sure you're using the Mk I engine?