SRLA

aka Soleil Rising Lossless Audio codec

How to build

Requirement

• CMake >= 3.15

Build SRLA Codec

git clone https://github.com/aikiriao/SRLA.git
cd SRLA/tools/srla_codec
cmake -B build
cmake --build build

Usage

SRLA Codec

Encode

./srla -e INPUT.wav OUTPUT.srl

Mode -m

You can change compression mode by -m option. Following example encoding in maximum compression (but slow) option.

./srla -e -m 4 INPUT.wav OUTPUT.srl

Max block size -B

The'- B' option can change the maximum block size. The following example encodes with a maximum block size of 4096 samples.

./srla -e -B 4096 INPUT.wav OUTPUT.srl

Number of divisions in block -V

You can change the number of divisions using the -V option. The number of divisions specifies the search depth of optimal block division. The following example encodes with a number of divisions in the block to be $2^{2} = 4$.

./srla -e -V 2 INPUT.wav OUTPUT.srl

Lookahead sample factor -L

You can change the multiplying factor for lookahead samples in variable block division. The following example is encoding with a lookahead factor of 8.

./srla -e -L 8 INPUT.wav OUTPUT.srl

Long-term prediction order -P

The -P option specifies long-term (pitch) prediction order, which improves the compression ratio, especially for vocal/pure tonal signals. Following example encoding with long-term prediction order 3.

./srla -e -P 3 INPUT.wav OUTPUT.srl

Decode

./srla -d INPUT.srl OUTPUT.wav

Performance

We use RWC music dataset for comparison.

Decoding speed vs compression ratio

Encoding speed vs compression ratio

License

MIT