Hi,
there are large collections of SF2 files that sometimes sound really great,
sometimes do not. In most cases simply looking at the file size already helps
to guess the quality of sounds, although it can be misleading if there are a
lot of sounds stored in the SF2 file.
But in some cases otherwise poor sounding collections contain real gems which
one would not have expected.
If I really want to get the sounds I like with the best quality I would have
to dig through all collections I have and listen to all the sounds in them.
Unfortunately, this takes a lot of time which is wasted in many cases as
there are so many equal sounding sounds which were taken from one SF2 file
and copied to another. Sometimes with very few if not no at all parameter
adjustments. It is also very time consuming to handle all files manually,
copy presets, listen, rate, re-arrange presets etc.
Now I wonder if this could be done automatically. Firstly there would have to
be a tool that extracts all sound presets (including drum kits) of a SF2 file
and stores them to harddisc, then runs a kind of file compare tool on each
sound preset.
There are some tools that can create a graphical representation of what is in
e.g. a MP3-file (moodbar and the like), so it means there could be something
that creates a graphical representation of how "similar" sounds are. And if
they are, label them as doublettes that need closer examination. If the
presets are identical, doublettes could be deleted.
If we can compare sounds by an algorithm then we can also do the following:
"My sax sounds much too bright, I need a sax that is much softer sounding. I
know I have it somewhere, but where is one stored and in which SF2 file is
it?"
We could have citeria for sound comparisons by labeling a particular sound as
a reference sound that we rate "100% fuzzy" and another that we rate "not
at
all fuzzy", and when we let the tool examine all presets then we find all
sounds that are "slightlier less fuzzy" or "similar" to our reference
sound
etc.
These are things we do all the time with our ears and our brain: rate sounds,
compare them.
Do You know what kind of development is going on there? What could I do to
make life simpler regarding comparison of SF2 sounds?
Regards,
Crypto.