From the flac.sourceforge.net FAQ:
"the encoder is looking for functions that approximate the signal. Higher settings make the encoder search more to find better approximations."
What is stored in the FLAC file looks like some mathematical function approximating a given audio waveform, together with some 'corrections' that, when applied to the function, result in the original waveform. The less information there is in a sample the less 'correcting' needs to be done and the more the math function alone will be able to re-create the original sound. Silence or nice sines result in excellent compression. Noise can hardly be compressed at all. Regular music is somewhere in between.
Compressing a WAV file into FLAC and back into WAV results in exactly the same digital waveform.
Slashdot Mirror
From the flac.sourceforge.net FAQ: "the encoder is looking for functions that approximate the signal. Higher settings make the encoder search more to find better approximations." What is stored in the FLAC file looks like some mathematical function approximating a given audio waveform, together with some 'corrections' that, when applied to the function, result in the original waveform. The less information there is in a sample the less 'correcting' needs to be done and the more the math function alone will be able to re-create the original sound. Silence or nice sines result in excellent compression. Noise can hardly be compressed at all. Regular music is somewhere in between. Compressing a WAV file into FLAC and back into WAV results in exactly the same digital waveform.