Adaptive Spectral EntropyCoding (ASPEC) is an audio compression for signals with bit rates of up to 64 kbit/s. In the ASPEC process, broadband signals are divided into sub-bands. In contrast to the Musicam method, these are formed by a modified cosine transformation (MDCT). In ASPEC, the transformation into the frequency domain
is performed with a maximum of 512 coefficients. Therefore, one works with 512 sub-bands and thus has an exact frequency division. In ASPEC compression, blocks are cut out of the input signal and then transformed. The length of each block is varied depending on the signal, with the size controlled by pulse detection. After the blocks are transformed into the frequency domain using the Modified Discrete Cosine Transform
(MDCT), the monitoring thresholds are calculated. ASPEC's quantization is controlled by psychoacoustic evaluations.
For coding, ASPEC uses Huffman coding with variable length code. Frequently used elements are assigned shorter code words, while less frequently used elements are assigned longer code words.