[Freepats] [Rosegarden-user] swami hard to use, soundfonts hard to use.. Re: restarting qsynth

[Marcos Guglielmetti]

El Martes, 18 de Septiembre de 2007 23:10, Emiliano Grilli escribió:
 | Tuesday, 18 September 2007 alle 17:14:03, Marcos Guglielmetti ha scritto:
 | > El Viernes, 7 de Septiembre de 2007 18:56, Emiliano Grilli escribió:
 | >  | Sunday, 02 September 2007 alle 15:11:08, Rua Haszard Morris ha 
scritto:
 | >  | > I thnk there's a reasonable chance that a lot of the complexity and
 | >  | > pain of editing soundfonts stems from the original sound font spec,
 | >  | > in that it's probably not an easy thing to wrap up nicely... if it
 | >  | > were, there'd probably be more apps that were able to edit/write to
 | >  | > soundfonts :)
 | >  | >
 | >  | > I think that freecycle is able to export soundfonts, but I've never
 | >  | > played with it.
 | >  | >
 | >  | > Has anyone compared how useful swami is compared with Vienna (I
 | >  | > think that's the Creative labs soundfont editor)? I'm guessing it's
 | >  | > no walk in the park either...
 | >  |
 | >  | Sorry for jumping in, I've not read the whole thread.. but having
 | >  | used swami I can confirm that's working (just apt-get installed
 | >  | version 0.9.4 from debian etch). I think it's pretty similar to
 | >  | vienna (used a loong ago). There are some steps to create a soundfont
 | >  | that are not user firendly at all, you'll find them on the "swami
 | >  | tips". Basically you do a file->new, go to the "Samples" folder,
 | >  | right click and "load samples". Then made a new instrument in the
 | >  | same way. Now (here is the obscure maneuvre) to assign the sample to
 | >  | the instrument you have to right click on the instrument *while the
 | >  | sample is selected* and select "paste". Knowing that it should be
 | >  | pretty straight forward to create soundfonts. You can control a lot
 | >  | of parameters and use many samples over the keyboard. I don't know if
 | >  | it's possible to layer samples based on velocity (that would be good
 | >  | of course).
 | >  |
 | >  | To my surprise, this time the whole swami-fluidsynth-jack thing is
 | >  | working (never did before) and this makes much more easy to edit the
 | >  | patch while listening to every change.
 | >  |
 | >  | Cheers
 | >
 | > We, at freepats (http://opensrc.org/listinfo/freepats), are making a
 | > nice Steinway soundfont using Swami, and, yes, it's a pain, but it
 | > works, most of the time...
 |
 | Very good!
 |
 | So, you've managed to build a multi layered soundfont with swami? 

Yes, sure!

 | How 
 | it's done? 

Hehe, I dont remember now, but I will see.... yes, now I remember.

 | How can the sampled be switched following the velocity value? 

Just switch to Velocity Mode (click on a white/blue icon) [swami opens on 
Piano Mode by default], and assign any velocity value you want for each 
sample. You can make plenty of layers. (i can send screenmshots if you need 
them)

Download the Steinway_IMIS2.2.sf2 and you will see it.


ftp://musix.ourproject.org/pub/musix/sf2/

I am working on a middle layer using the ff layer but with a cut-off filter


BTW: using Pysf lets you modify a soundfont without having to upload all the 
entire file, just an xml or so....

 |
 | I contributed a pair of soundfonts to the freepats project long ago
 | (rhodes and mellotron: http://freepats.opensrc.org/sf2/),

Nice!

 | and it would 
 | be wonderful to contribute again, unfortunately my free time is not much
 | these days.

We all have this problem... me too.

 | I can record an acoustic bass (picked),

It would be nice, the project goal is to make hi-quality (well, ok, ok!) 
soundfonts

 | a drumkit, a very nice  
 | accordion, mandolin, flute, glockenspiel and a bunch of percussions
 | (guiro, maracas, darbuka, shaker, triangle, maroqui bongos, claves) 

Wow!

 | but as 
 | I said I don't really have time to cut them or made the soundfonts. If you
 | want I can provide the samples.

Yes! someone will take them and make soundfonts some day!. Some months ago a 
woman (ml at distasis.com) joined us and she is learning how to do this things. 
I am mailing freepats at opensrc.org

freepats has some space in their servers, but if it's not enought, I cant 
provide space at ftp://musix.ourproject.org/pub/musix or other servers.

 | Let me know if you want some samples for 
 | the mentioned instruments, and at which interval they should be sampled
 | and how many velocity layers are needed.
 

Maybe all thoses aspects have something to be with the essential of the 
instrument itself. I dont know if "a very nice accordion" will need 1, 2 or 3 
layers to sound realistic; maybe using 2 layers would be enough, but the 
owner of the instrument should analyse that; or contact some expert. For 
instance, the IMIS Samples recorded at IOWA use to be sampled at pp, mf and 
ff, maybe you can mail the technician who sampled them to ask him about: how 
should we sample mandolin, flute, glockenspiel and a bunch of percussions 
(guiro, maracas, darbuka, shaker, triangle, maroqui bongos, claves) ?


But I would say: 3 layers are enought for me to make something realistic.

 | I did also some multilayered drumkits for hydrogen in the past, but now
 | I have a new drumkit and better microphones/preamps. And I already
 | promise a new drumkit to comix :)
 
Me too!

 | Making good sounding instruments is so time consuming... but we should
 | definately offer a free (as in speech) GM selection to be shipped with
 | audio oriented distros, in my opinion.
 |
 | It's nice to see that the freepat project isn't dead
 |
 | Cheers

It's not dead (but we need more people), just trying to make something at 
hi-quality, and it supposes much more time.

I am finishing Musix 1.0 R3 these days, so I will not have time, but in the 
next months I am planning to make some soundfonts using free samples.

Regards,

Marcos.