-
I am working with pipe organ sound fonts. I work with GrandOrgue and Hauptwerk organs as well as my own eplayOrgan64. All of these can play many different types of virtual organs and accommodate many hundreds of (usually stereo) samples
Recently I have tried using Polyphone 64 to convert some GO .organ samplesets to .sf2 format. This results in HUGE files some of which can cause polyphone to crash. Also some appear to hit some limiting number (65,536) of "objects".
What are the factors which limit the size of .sf2 sound fonts? The limit appears to be around 1 to 2GB. Is this due to some fundamental limit in the .sf2 specification or is it due to polyphone. All the programs I am using are 64 bit and I have 8GB of memory.
csw900 -
the factor that limits the size of .sf2's is common sense.
if you want to play a soundfont smoothly, then dont go over 500 Mb.
first of all; monomize the stereo samples; instruments thet play different sounds for the left and the right ear dont exist.
regards bottrop -
bottrop Thanks for your reply.
However it was not helpful - clearly you do not use or play organs. GO can use .wav sound fonts which extend beyond 16GB and they play perfectly well.
I have made .sf2 sound fonts up to around 2GB and they also play perfectly well - there is no technical reason why they shouldn't. Also your idea of "momomizing" samples is exactly the opposite of what I am trying to achieve - Organs DO play different sounds for the left and right ears - in fact they play perfect stereo as do all other large instruments (even pianos).
The only real problem of large sound fonts is the load time - which can be 10's of seconds for large .sf2 fonts and much longer for compressed .sf3 fonts.
I would still like to know what the maximum size limitations are for .sf2 fonts. Does the specification include features which have a size limitation?
csw900 -
if tou have two ears, you will hear all sounds in stereo, even a piccolo flute playing through a key hole, but that flute produces just mono sound waves. you confuse stereo with panning. it is very easy to pan a mono sound. it is very difficult to pan a stereo sound since the channels are never the same quality.
regards bottrop -
Yes I do have two ears and do hear all sounds in stereo and I agree that panning can produce an artificial form of stereo. However proper reproduction of stereo sounds requires (at least) two channels which are normally provided by making a recording with two microphones.
There is no need to provide artificial panning if you have a proper stereo signal to begin with. The left channel is panned to -50 and the right channel is panned to +50. This produces the best and most realistic reproduction possible. And it does work with .sf2 sound fonts. I have many examples to prove it.
I could go on but this is off the subject and I still need to know the maximum size capabilities (if any) of .sf2 sound fonts. I know that the fluidsynth people are talking about over 4GB sound fonts (read about it in their version 2.2 documentation)
csw900 -
i never said that panning can produce a form of artificial stereo, so i dont know whom you agree with.
panning and stereo are completely different things, the only thing they have in common is the number of speakers they use.
regards bottrop
Sign in or register to take part in discussions.
Polyphone needs you!
Polyphone is free but there are costs associated with its website and development. A small donation will help a lot.
Donate
Learn the basics
Try a tutorial
Scroll to
top
top