Audio compression (audio compression) is the type of data compression, encoding used to reduce the volume of audio files or to reduce bandwidth for audio transmission. Sound file compression algorithms are implemented in computer programs called audio codecs. The invention of special compression algorithms for audio data is motivated by the fact that general compression algorithms are inefficient for working with sound and make it impossible to work in real time.
As in the general case, lossless sound compression is differentiated, making it possible to restore the original data without distortion and lossy compression, where such recovery is impossible. Lossy compression algorithms provide a high degree of compression, for example, an audio CD cannot contain more than one hour of “uncompressed” music, with lossless compression, the CD can store almost 2 hours of music and with compression with an average bit rate of 7 to 10 hours.
Lossless compression
The complexity of lossless compression is that recordings are extremely complex in structure. One of the compression methods is to search for samples and their repeats, but this method is not effective for more chaotic data, such as digitized sound or photographs. Interestingly, if computer generated graphics are much easier to compress without loss, synthesized sound will have no advantage in this regard. This is because even computer generated sound generally has a very complex shape, which is a difficult task to invent an algorithm.
Another complication is that the sound generally changes very quickly and this is also the reason why ordered byte sequences appear very rarely.
The most common lossless compression formats are:
Lossless Audio Codec (FLAC), Apple Lossless, MPEG-4 ALS, Monkey’s Audio and TTA.
Lossy compression
Lossy compression is extremely widespread. In addition to computer programs, lossy compression is used in the transmission of audio to DVD, television and digital radio and in the transmission of media on the Internet.
An innovation in this compression method was the use of psychoacoustics to detect sound components that are not perceived by the human ear. An example is the high frequencies, which are perceived only when the energy is sufficient, or the silent sounds that occur simultaneously or immediately after the loud sounds and, therefore, are masked by them; these sound components can be transmitted with less precision, or not at all.
For masking, the time sequence signal of the amplitude samples is converted into a sound spectrum sequence in which each component of the spectrum is encoded separately. To implement this conversion, fast Fourier transformation methods, MDCT, quadrature mirror filters or others are used. The total amount of information during this recoding remains unchanged. Compression in a given frequency domain may consist of the fact that masked or zero components are not stored or encoded at a lower resolution. For example, components with frequencies up to 200 Hz and over 14 kHz can be encoded with 4-bit resolution, while components in the mid-range can be encoded with 16 bits. The result of this operation will be encoded with an average bit depth of 8 bits, but the result will be much better than when the entire frequency range is encoded with an 8 bit depth. However, it is obvious that the low resolution transcoded fragments of the spectrum can no longer be restored exactly and are therefore lost forever.
The main parameter of lossy compression is the bit rate, which determines the degree of compression of the file and, consequently, the quality. Distinguish compression with a constant bit rate (Eng. Constant Bit Rate – CBR), Variable Bit Rate (Eng. Variable Bit Rate – VBR) and Average Bit Rate (Esp. Average Bit Rate – ABR).
The most common lossy compression formats are: AAC, ADPCM, ATRAC, Dolby AC-3, MP2, MP3, Musepack Ogg Vorbis, WMA and others.
