adaptive transform acoustic coding (ATRAC)
Adaptive Transform Acoustic Coding (ATRAC) is an audio codec developed by Sony for the MiniDisc. Introduced in 1991, it was followed by ATRAC3 and ATRAC3plus in 2000.
The audio compression is generally based on psychoacoustics. ATRAC also filters out irrelevantdata from imperceptible sounds. Furthermore, the human hearing characteristic is implemented into the procedure. This is mainly done to reduce the quantization noise caused by the quantization error. ATRAC uses the psychoacoustic effect that humans perceive certain frequency ranges better, others worse.
The highest hearing sensitivity is at frequencies around 4 kHz, higher, but also lower tones are perceived as quieter at the same sound pressure.
Furthermore, a loud sound can mask a softer sound, this is called masking. This masking effect is more pronounced the different the sound pressure of the two sounds is and the closer they are to each other in frequency. ATRAC takes advantage of these effects by evaluating the tone ranges during compression and deriving the quantization from them. Thus, frequency ranges where the ear is very sensitive are quantized higher than those where hearing is less sensitive. This measure keeps the quantization noise low and superimposed by the sound mixture.
ATRAC, which has undergone continuous development, constantly adjusts the volume and frequency of the sound mixture to the hearing characteristics. The process is suitable for high quality audio, it operates at a sampling rate of 44.1 kHz, has a quantization of up to 24 bits and a data compression ratio of 5:1. The resulting data stream is 292 kbps.
ATRAC operates with three frequency bands from 0 to 5.5 kHz, from 5.5 kHz to 11 kHz, and from 11 kHz to 22 kHz, each subdivided into 128 bands using Modified Discrete Cosine Transformation( MDCT). There is a block mode of 11.6 ms and a mode consisting of three short blocks of 2.9 ms, 2.9 ms and 1.45 ms.