Samplerbox uses the the WAV file format, specifications explained here.
The samples themselves have to be standard WAV files, stereo or mono, 16 bits or 24 bits, commonly at a sampling rate of 44.1 Khz.
So yes: you can also create your own - this was the first requirement I had when starting to work with it.
Processing of these files is controlled with a definition.txt, which can be made as complex as you wish
- also as simple as you wish, you can even work without one.
But simple or not, these samples sets need to be made or downloaded somewhere.
As SamplerBox is not mainstream, you can't find much of them ready to use.
Basically, a line in the definition.txt (perhaps defaulted by the %%parameters) describing the way the wav's should be processed is a soundfont.
Packages can use soundfonts with other formats like the SFZ format used by the popular SF2 samplesets
or proprietary formats like used by the GrandOrgue sample sets (usually high quality and easily extractable)
and commercial sample sets in the protected Native Instruments Kontakt nki/nkm format.
Despite the soundfont differences, most of the sample sets have the needed wav's stored in them in the directory structure,
so it's a matter of getting the right wav's out and write the definition statements.
Sometimes this is not possible (e.g. gig/nki/nkm), tough, write your congressman.
Usually they already have the loop markers - count your blessings as this can be a tedious job.
Some may even have release samples defined, however it's a challenge to get the %xfade.. parameters properly tuned.
Bottom line: you can find lots of good wav collections - or sound font sets where you can just copy the wav's from the sample set or have to perform a small conversion (SF2)
... or need the utilities with your commercial product to get them out. I collected some links on the bottom of this page.
If you want to extract WAV files from an SF2, here is a way to start:
File > Import Soundfont, then choose your SF2 file
File > Save As, then save it as SFZ. Choose Save the samples> Yes.
You'll have the WAV files in the output folder!
But why not just support the other more popular formats ?
I know it's very annoying to not support existing formats (such as .sfz and .sf2) and start and (n+1)th format ;)
But really I prefer to start something fresh (based on more than 10 years of experience with all these formats), and short...
An example: here is the definition.txt file in SamplerBox format, for a banjo sample-set:
As I really want something fast (a sample-set should be done in 2 minutes, not more), I don't want to support these very verbose formats,
for which you need a third-party software to write definition files.
Editing / correcting samples
Samples may have a silence in the beginning,
which might make you think SamplerBox suffers from latency.
Some packages take care of that in real time or give the possibility to
define the actual start in the soundfont definition, thus hiding this recording deficiency.
SamplerBox does nothing in this area, so you have repair that - which can be quite tedious.
For this I wrote a simple python script to trim all samples in a folder
(trim = both sides, which is useful for memory usage and polyphony).
When they don't have loop markers, I use loopauditioneer
and wavosaur for creation of these.
Wavosaur also has a batch feature, enabling easy normalizing, resampling and other conversions.
If you have better (free) alternatives for the tools I mention above, please share them on !
Where to find samples
In the you'll find a samples directory with a few sample sets.
But a search engine is your best friend......really... And please do interesting findings.
For a start some general/index sites.
https://www.samplerbox.org/instruments Important note: samplerbox version here(homspace) has been optimized with regards to velocity levels making samplesets using the %velocity parameter incompatible. Result is no sound at all ("empty sample set").
Quick fix is adding %%velolevs=127 to the configuration.txt; background details in the velocity levels paragraph on the "More info" page.
For the grand piano I recommend the version on this site as this gives audible advantages while load time is lower compared to original.
Many more exist (I didn't want to mention all the single wav repositories like freesound containing tons of stuff that may fit your particular need).
In the paragraphs above you'll find hints how to deal with the various results of your search results..