Lexicon MP3
Informal term for a psychoacoustic, lossy data reduction method for digital audio data. The more detailed designation is MPEG-1 Audio Layer 3.
The goal of MP3 and other similar procedures is to encode the audio data with as little audible loss of quality as possible so that as little data as possible is needed. This is to save storage space on storage media or transmission capacity for data transmissions. For coding, these methods use the findings of psychoacoustics, which deals with the characteristics of the human auditory system. In simple terms, errors are allowed during coding that the auditory system would not notice. Because of these errors, the original signal cannot be reconstructed 100%, so the process is called lossy. The errors can be detected very easily by measurement, but are difficult to hear.
With this method it is possible to reduce the data volume of an audio CD to 1/8 to 1/12, for example, without any noticeable deterioration in sound quality. With MP3 you can select the desired data rate and thus the strength of the data reduction within certain limits. The lower the data rate, the more audible the errors. This way you can make individual compromises between data quantity and quality.
The format of the data and the function of the decoder is defined in the ISO/IEC 11172-3 standard. The different decoders should therefore provide matching results. However, encoders are not standardized in all details, and there can be significant differences in sound quality. Good encoders are capable of delivering a result that is generally indistinguishable from the CD at a data rate of 128 kBit/s or more.
The development of this process took place to a large extent between 1985 and 1992 at the Fraunhofer Institute for Integrated Circuits in Erlangen. Data reduction methods were also developed elsewhere during this period. Another such method (MUSICAM) found its way into the same standard as MPEG-1 Audio Layer 1 and Layer 2.
The MPEG-1 standard (ISO/IEC 11172), adopted in 1992, has as its purpose the data-reduced coding of video films at a data rate equivalent to that of an audio CD. For this purpose, both the picture information and the sound information had to be data reduced. Part 2 of the standard applies to the data reduction of the picture information, part 3 deals with the audio information. Part 1 contains general specifications. For the audio information, it was agreed to standardize 3 different methods as "layers", which offer different levels of data reduction with different computational requirements. Layer 1 is the simplest method. It is practically nowhere used, because it offers only low quality. Layer 2 is the method formerly known as MUSICAM. It is widely used in digital broadcasting (e.g. DAB). Layer 3 is MP3, the process of the group from the Fraunhofer Institute, and is widely used on the Internet, computers and mobile audio devices. It is the most computationally intensive of the three, and provides the best quality for a given data rate.
Development has not stopped since it was standardized in 1992, and there are now a number of other coding schemes that work on similar principles. These include Atrac, AAC, AC-3, Ogg-Vorbis, and many others. MP3, however, has the widest distribution of all such methods, and can still keep up in terms of quality.
There are also lossless data reduction methods for audio that allow the original data to be reconstructed bit by bit (e.g. FLAC). However, the achievable data reduction is much lower. Common values are a reduction to about half. Such methods are most comparable to the archive formats that are widely used on computers, such as ZIP. Lossless methods allow to encode and decode the data as often as desired one after the other without any loss of quality.
In contrast, the quality deteriorates with each encoding process in a lossy process. Repeated compression and decompression should therefore be avoided with such methods. For this reason, these methods are not a suitable means of processing audio data. They are useful for distributing finished material, e.g. over the Internet, via digital broadcasting, or as a data format for playback devices.