settings for wav/sf2 files (from 17.00.68)
thread for delta/plzma4:a0 settings discussion, moved from:
since on version specific thread it will be pushed back on thread list quite fast.
@spwolf compression of sounds is much more than delta on .wav. Wav can have different bit depths and number of channels which affect delta number - delta:1 is good for 8-bit mono, delta:2 is good for 8-bit stereo, delta:4 is good for 16-bit stereo…
Sound modules (.dmf, .it, .med, .mod, .mptm, .okt, .rns, .s3m, .umx, .xm) and soundfonts (.sf2): most of them support 8-bit mono and stereo, some also contain 16-bit mono and stereo. If you are using 7z delta filter for files like this, I would recommend delta:2 + LZMA for all sound modules except .s3m which are compressed better using delta:1 + LZMA.
delta:3 + PPMd:mem256m:o3 would help on all non-compressed images in 24-bit such as .PPM, .PNM and also .PAM; with LZMA2 instead of PPMd delta:3 would help on all camera raw in Mamiya .mef format while delta:4 would help on all camera raw in Leaf .mos format and on uncompressed Sony .arw.
Soundfonts are compressed best by closed-source sfArk (nothing practical comes close atm), but libbsc -b1024rcf -m0 -H28 is good here. A parser with Monkeys Audio would be superior to both.
@spwolf worth to mention are also those audio formats: .xi (instruments), in some rare cases there are .au and .aif/.aiff, .voc and .snd
@Stephan we have optimized switches for 16 bit stereo. We assume that is the most commonly used format, hence the delta:4 and lp2, pb2. With higher chunk size and dictionary in next version, it is not only faster than ppmd_sh but also compresses better on our “real” life samples. That goes the same for sf2 files (also tested au but did not work there).
For bmp, our results were too varied and we could not pick a single setting to use on our samples. So that will have to wait for detector to be built first.
MA is a lock, likely for the next release, but we will still need fast settings with delta for faster modes, so detector for wavs will also be beneficial in the future.
Can you send me those sound modules files for testing, if we can replicate results on larger amount of samples, we would also include it for next release.
Thanks a lot for the feedback!
Let me know if you need further files.
.70 adds larger dictionary and chunk sizes in stronger options for wav mode, also added sf2 to that mode.