sfz sound font problems

not enough memory…

Sometimes as I use sfz and I load a soundfont an error display that I’ve not enough memory, if I load a bass soundfont e.g it doesn’t works but if I load before a drum soundfont and then in another track the bass it works…sometimes it works sometimes no…
another question: some soundfonts (e.g. bass or drum) doesn’t work, I load it but no sound… maybe I should configure better sfz (what are pr16, sf16…and so on?)
thank you

OK, there are 5 modes for sfz.

SF16 - load the whole Soundfont as 16 bit samples
SF32 - load the whole soundfont as 32 bit samples
PR16 - load the required patch/program as 16 bit samples
PR32 - load the required patch/program as 32 bit samples
DFD - Stream the samples in direct from disk

If you attempt to use one of the 32bit modes and don’t have sufficient memory, it will then attempt to load as 16bit. If that also fails it will give the ‘out of memory’ error.

Whether you use SF or PR depends on how much you change program in the same soundfont. If, for instance, you have an SF2 of a violin with several articulations in different programs which you switch between alot, then SF will be better; since PR unloads all the samples before loading up the new patch.
However, if you have a bass SF2 with 4 instruments, but you only use the fingerbass, then SF is going to fill your memory unnecessarily, so PR would be better.

If memory is low, use the 16bit modes, but use the 32bit mode for rendering.

DFD uses no memory, and requests the data from disk. This puts more pressure on your HD spindle, but can allow for massive SF2 usage. However there are a number of cases where the left samples arrive before the irght samples in stereo fonts. Rene has battled with this. My old P3 laptop handles it all very well, as does the P4HT here at work. I have tried it in an internet cafe, and had appalling results with DFD.

Generally I use DFD myself, since most of my writing is on my laptop with 256Mb, and it struggles with more than a couple in PR16 mode.


Duncan, there is no difference in the sound quality between 16-bit and 32-bit modes. This only affects the way samples are stored in memory. Rendering/playing is a bit less CPU-intensive in 32-bit mode, that’s all. At the speed of today’s CPUs, this isn’t very significant.

Right on target with all the other facts.

you’re all very kind as always :slight_smile:
thank you all