Sample numbering in SFZ for LinuxSampler

Hello everybody,

I am currently writing a tool to emulate the Kawai K1 in the LinuxSampler engine (by generating the SFZ files automatically).
But I am facing a “stupid” problem, and I can’t find the answer in any existing SFZ file specification.
So maybe one of the members know the answer…

The question is simply : “how are samples numbered in SFZ files?”
The SFZ file specification is quite strange on this point. For loop_end operator, the doc says that value shall be from 0 to 4294967296, which is not possible (this does not follow the mathematic rule of 2^32-1 maximum value)
Either samples are numbered from 0 to 4294967295, or they are numbered from 1 to 4294967296, but it can’t be a mix between the two :thinking:
Simply said : if I have a sample which is 32K long, shall I declare this in the SFZ file
loop_start=0 loop_end=32767
or shall I declare this :
loop_start=1 loop_end=32768

Any idea ?