[fluid-dev] Using multiple soundfonts from unix command line

[Matt Wallis]

Hi,

Background: I'm generating midi files using Lilypond. I want to create audio that uses instruments from different soundfonts, and I want to control this from the unix command line, so that audio production is in an automated pipeline (I use gnu make).

My first attempt to do this was via a config file:

load /opt/local/share/sounds/sf2/GeneralUser_GS_v1.47.sf2
load /Users/matt/Downloads/Nice-Keys-Ultimate-V2.3.sf2
select 4 2 0 33
select 5 2 0 33
select 6 1 0 33
select 7 1 0 33

... loaded from the command line:

fluidsynth -f config.fluidsynth -F myMusic.wav myMusic.mid

This does not work (the wav is generated, but not using the instruments specified in the select commands) - I guess that the instructions in myMusic.mid override what I had set up in the config file.

Second attempt:

1. Put the select commands into a separate file `selections.fluidsynth`

2. `config.fluidsynth` now contains only the load commands

3. Redirect commands from `selections.fluidsyth` on the command line:

fluidsynth -f config.fluidsynth -F myMusic.wav myMusic.mid < selections.fluidsynth

This doesn't work. I guess -F means that FluidSynth is not reading shell commands from stdin.

What should I do? I'm new to using FluidSynth in anything other than a very basic way. I may well be missing something 'obvious'!

Matt

Using: FluidSynth 1.1.6 on Mac

PS - FluidSynth is brilliant! Thanks to all involved.