slmdev / sac

state-of-the-art lossless audio compression
44 stars 4 forks source link

Sac

Sac is a state-of-the-art lossless audio compression model

Lossless audio compression is a complex problem, because PCM data is highly non-stationary and uses high sample resolution (typically >=16bit). That's why classic context modelling suffers from context dilution problems. Sac employs a simple OLS-NLMS predictor per frame including bias correction. Prediction residuals are encoded using a sophisticated bitplane coder including SSE and various forms of probability estimations. Meta-parameters of the predictor are optimized with DDS on by-frame basis. This results in a highly asymmetric codec design.

This program wouldn't exist without the help from the following people (in no particular order):

Matt Mahoney, Dmitry Shkarin, Eugene D. Shelwien, Florin Ghido, Grzegorz Ulacha

Technical features

Technical limitations

Sac uses fp64 for many internal calculations. The change of compiler options or (cpu-)platform might effect the output. Use at your own risk and for testing purposes only.

Benchmarks

Sac v0.7.8

16 files (51.014.742 bytes) parallel on i7-13700H.

Asymmetric encoding profiles - bits per sample (bps) is mean bps over all files Profile Size Enc-time Dec-time bps
--normal 26.812.987 00:00:41 00:00:30 8.441
--high 26.708.465 00:04:55 00:00:52 8.407
--veryhigh 26.621.497 00:32:04 00:01:04 8.379
--best 26.554.323 05:17:19 00:01:05 8.358

 

Comparison with other lossless audio codecs Program Parameters Source
Sac v0.7.8 --best open
OFR v5.100 --preset max closed
paq8px_v208fix1 -6 open
MAC v10.44 --insane open
Numbers are bits per sample (bps) Name Sac OFR paq8px MAC
ATrain 7.026 7.156 7.353 7,269
BeautySlept 7.418 7.790 7.826 8,464
chanchan 9.692 9.778 9.723 9,951
death2 5.068 5.465 5.215 6,213
experiencia 10.866 10.915 10.963 11,005
female_speech 4.372 4.498 4.708 5,190
FloorEssence 9.081 9.409 9.488 9,537
ItCouldBeSweet 8.200 8.310 8.330 8,531
Layla 9.490 9.571 9.725 9,783
LifeShatters 10.768 10.808 10.868 10,838
macabre 8.986 9.026 9.249 9,172
male_speech 4.276 4.256 4.509 5,255
SinceAlways 10.344 10.409 10.455 10,522
thear1 11.379 11.400 11.474 11,451
TomsDiner 6.996 7.108 7.057 7,432
velvet 9.762 9.990 10.030 10,461
Mean 8.358 8.493 8.561 8,817