Re: [fluid-dev] Trying another reverb ?

[Ceresa Jean-Jacques]

Hi ,

Here is the source file freemverb-demo.c of the reverb announced. It contains the two reverbs (freemverb, and freeverb).

To do the test the file must replace the original fluid_rev.c of any Fluidsynth version from v1.1.3.

At the top of the file you can set the variable freemverb (char freemverb = 1; // 1: use freemverb   0:use freeverb)

> Now you have the possibility to build tow version of the FluidSynth library:

> - the one with 'freemverb = 1' using freemverb  (stand for 'free modulated reverb')

> - the other with 'freemverb = 0'. using freeverb.

When you run the freemverb version you will see the message:Fluidsynth: using freemverb-demo (in process mixing only !).

1)You can use the FluidSynth console application with a MIDI keyboard on input (if possible).

To do comparison test between both reverb, i have used the GeneralUser GS 1.471 soundfont from S;Christian Collins

but it isn't mandatory.

2)To make an effective comparison,following settings for both reverbs are recommened.

 - Use headphone only otherwise fluidsynth reverb  will be masked by the natural reverb of the room your are in.

 - roomsize : 1(max)     damp: 0 (min)

 - width: default    level: default

 - SoundFont Preset: The more percussive the preset is (fast attack and than decay)(i.e piano ,...) the more the reverb will be stressed.

 - Your playing for the comparison:

    - use low , medium and high dynamics ranges.

    - use low, medium and high key range too.

Note: Fast attack and staccato playing contributes to the reverb stressing (i.e making "ringing").

Note: please note that reverb settings range are 0:Min 1: Max for both reverb. But freemverb is not calibrated the same way as freeverb.

Freemverb have higher 'reverb time' than freeverb.  Consequently to get the same roomsize effect for example you will need 0.2 for freemverb against 0.7 for freeverb.

Similarly, you will need different value for 'damp' setting to produce the same damp effect. For this reason, to do a pertinent comparaison test please follow the above recommendations (see 2). Later it will be easier to use your own settings.

Remember: The goal of freemverb is to diminish the "ringing" tendency (resonant frequencies) produced by any artificial reverb.

Objectives tests would be appreciated.

cheers.

 

jjc

 

> Message du 23/10/17 02:44
> De : "Ceresa Jean-Jacques" <address@hidden>
> A : "FluidSynthmailinglist" <address@hidden>
> Copie à :
> Objet : Re: [fluid-dev] Trying another reverb ?
>
>

> Hi, Marcus and GrahamG

>  

> Thanks for your interrest.

>  

>I would also be very interested in a better sounding reverb.

I don't pretend that freemverb is a better reverb. This is ears dependant. The goal was to obtain a less "ringing" sound that with freeverb.To obtain this result the inner structure off freemverb is completely different and the necessary comb filters are not fixed delays lines but modulated delay lines. The effect of modulation is to diminish local resonnance.  The "ringing" is really diminished with only 8 combs filters !. However now we obtain a sound a bit "chorused" wich is a bit unnatural for reverb. For this reason somes ears may not accept this sound. By the way the best is to hear the result.

 

> >So maybe what we need is a system where we can have multiple reverb implementations in core FluidSynth and a setting that can be used to select one >of the implementations.

>  

> >Let me explain that idea in a bit more detail. What I have in mind is a better defined interface (basically an internal API) for the reverb and chorus >effects. Something that works a little like LADSPA[1], only less generic and more specific to FluidSynth.

> The fact that it is API specific to FluidSynth make the things OS independant. I like this idea.

>  

>  

> Yes, but to keep things simple at a first time and to give you the possibility to ear the result quickly . I propose the following that is simple for me:

> 1)I give  a file called fluid_mrev2.c. It will contain the two reverb (freemverb, and freeverb) and of course the API expected by the FluidSynth library(defined in fluid_rev.h) .

> -At the top of the file the variable: char freemverb = 1; // 1: use freemverb   0:use freeverb

>  

> 2)The file fluid_mrev2.c is intended to replace the original fluid_rev.c

> 3)Now you have the possibility to build tow version of the FluidSynth library,

> - the one with 'char freemverb = 1' using freemverb  (stand for 'free modulated reverb')

> - the other with 'char freemverb = 0'. using freeverb.

>  

> Note:fluid_mrev2.c is actually a working reverb and the code is a draft not yet intended to be easly understandable. It works and will be used

> with usual shell reverb commands.

>  

> This could be done in a couple of days.

> cheers.

> jjc

>  

> Message du 22/10/17 21:14
> De : "Marcus Weseloh" <address@hidden>
> A : "FluidSynth mailing list" <address@hidden>
> Copie à :
> Objet : Re: [fluid-dev] Trying another reverb ?
>
>

Hi,

>

I would also be very interested in a better sounding reverb.

>

I've written a different reverb as well, implementing a "sympathetic string reverb" using tuned comb filters. Currently I maintain it in a custom fork, but I wanted a maintainable solution. So I started working on the LADSPA plugin system: getting it working properly and in a stable manner, documenting it and tuning the performance. The largest part of the changes is already merged in the master branch on GitHub. So adding a different reverb would be as easy as writing a LADSPA plugin, disabling the internal reverb and configuring the new one as a plugin. But I guess that won't be much use for people using FluidSynth on non-Unix platforms...

>

So maybe what we need is a system where we can have multiple reverb implementations in core FluidSynth and a setting that can be used to select one of the implementations.

>

Cheers,

>

  Marcus

>

2017-10-22 19:14 GMT+02:00 Graham Goode <address@hidden>:
>

Hi jjc
>
> Yes, there would be great interest in this, particularly from the
> jOrgan users group as many of use us use fluidsynth with the current
> reverb engine.
>
> Kind regards,
> GrahamG
>

> On 10/22/17, Ceresa Jean-Jacques <address@hidden> wrote:
> > Hello,
> >
> > For another application than FluidSynth i have build a reverb (intended to
> > be called freemverb) that sounds less "ringing" that freeverb.
> >
> > Like freeverb , freemverb is a "late" reverb and have a low cpu load
> > (sligtly above freeverb) and low memory cost.
> >
> > Both freeverb and freemverb aren't high quality reverb, but it seems that
> > freemverb gives better results (at least on my ears).
> >
> >
> >
> > It would be easy to build a version of freemverb for fluidsynth with the
> > same Reverb' API (v 1.0.6 or above if versions above have the same Reverb
> > API than v1.0.6 ?).
> >
> > Before doing that, i wish to know:
> >
> > 1) If there are any interests for FluidSynth ?.
> >
> > 2) Also it is necessary that others peoples than me care objectives tests
> > (by ears) ?.
> >
> >
> >
> > Let me know your opinion.
> >
> > Regards
> >
> > jjc
> >
> >
> >
>
>

_______________________________________________
> fluid-dev mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/fluid-dev
>

>

_______________________________________________
fluid-dev mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/fluid-dev

_______________________________________________
fluid-dev mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/fluid-dev

freemverb-demo.c
Description: Text document