Subject: Re: [linux-audio-dev] New release of the Smurf Sound Font Editor
From: Josh Green (jgreen_AT_users.sourceforge.net)
Date: Sun Sep 17 2000 - 01:39:28 EEST
Benno Senoner wrote:
> On Sat, 16 Sep 2000, Josh Green wrote:
>
> hehe , it seems that I made the right assumption that users want
> software emulation of SF capable hardware.
> I looked at the SF docs, and the modulation , filtering and
> enveloping capabilities and LFO are quite simple.
> I will take this into account when working the sample playback engine of EVO,
> so that all parameters (hopefully) can be emulated.
> And the funny thing is that thanks to the disk streaming capabilites,
> you can play Soundfonts which are larger than your available RAM.
> But notably SF aren't that big (32MB would be quite a big SF instrument).
> That means that in EVO you will be able to load a large amount of SF instruments
> simulaneously.
> For now I am justfocusing on a flexible sample playback engine.
> I sketched out a scheme where you can modulate all parameters
> like volume,pan, pitch,filter cutoff etc (except voices with voices)
> by every MIDI source and by almost arbitrary formulas.
> (eg pitch = base_pitch * pitchbend_value + velocity ).
> Since these formulas are not computed for every sample but only
> every 32-64 samples or when a MIDI event occurs, the added overhead
> is minimal. (the formula is computed calling a list of function-ptrs (or switch
> statements which is even faster because it translates into jmp tables).
> This is still vaporware but my plan is that EVO should have this feature,
> because it will ensure a high degree of modulation flexibility.
> (emulatiog SF modulation , translates into generating a set of simple forumlas
> (like the reverb/chorus addition rules in SF patches etc).
>
That is one thing that slightly pains me in working with Smurf. Is that a much more
flexible sound generating model could be done in software outside of the sound font
standard. So what would be the purpose of using sound fonts then?? I guess it is
one more standard format out there that can work between hardware, but it does have
its limitations. Sound Fonts have a pretty good modulation standard though, that
allows the modulation of pretty much all effects with pitch benders, note #s, note
velocity, and custom midi controllers. I have yet to see this being possible with
the OSS AWE or ALSA driver though, its pretty close though. I'm still not sure
whether this should be done in a driver or a client software program (ALSA MIDI
filter?). This was another topic that I received no response on. I guess there is
just too much work being done for the basics of ALSA and not enough time for the
developers to think about other things. I would like to help with this effort
though.
There are currently two software based sound font emulators that I know of: csound
and Timidity++. I have not been able to get either one to work, but have not really
tried either. I have heard good reports from a user who was using Timidity++ and
Smurf as they did not have an AWE card. But I don't think must users would go
through what he has to.
>
> And no one stops us from importing GUS patches too.
> (what kind of modulation/ sample playback capabilities does the GUS offer?)
>
The GUS seems fairly limited to me. Or at least the driver for it in OSS. Its not
polyphonic and doesn't allow layered instruments. Please correct me if I'm wrong,
somebody.. I was only able to impliment the loading of a single sample in Smurf
with the OSS GUS and SoftOSS driver (equivelant API but done in software, not very
good either IMO).
>
> But the most interesting formats will be AKAI and a format whose name
> starts and ends in G. :-)
>
Hmm.. I haven't heard of this yet. What is it?
Cheers,
Josh Green
This archive was generated by hypermail 2b28 : Sun Sep 17 2000 - 02:52:41 EEST