ATRAC3 uses a data compression algorithm that adapts its analysis depending on the frequency components and characteristics of the sound signal.  This enables it to compress CD data to approximately 1/10 of its original size with almost no loss in sound quality.

1. Analyzes sound characteristics in four bands

ATRAC3 realizes high-efficiency compression by dividing the audio signal into four bands and analyzing the sound characteristics of each band extremely precisely.

2. Divides sound into two types and efficiently compresses each type

Sound signals carrying pure sounds, like those made by a violin or a triangle, when analyzed, are found to be composed of certain sounds that have a higher sound level in frequency analysis.  ATRAC3 therefore classifies the frequency data into these two different kinds of sounds, to increase compression efficiency.

3. Replaces frequently recurring digital patterns with shorter digital patterns

Digital signals are expressed as 0s and 1s.  ATRAC analyzes the pattern sequences of 0s and 1s and replaces combinations that recur frequently with shorter combinations and with those that do not occur often with longer combinations.  This rule enables the overall data size to be reduced.