MP3/WMA Playback: Notes

NOTE: The information provided in the following is of a general nature and not specific to BMW.

MP3 stands for MPEG Layer 3 and is a compression method for audio data which achieves very high compression ratios. MP3 makes use of a phenomenon that is associated with psychoacoustics: Human beings do not hear everything that theoretically could be heard - in any case not consciously. Based on the characteristics of human hearing, the MP3 inventors at the German Fraunhofer Institute have developed perception-adapted audio coding whereby part of the digital sound information normally found on the CD is simply omitted. These are sounds that most people do not consciously perceive, e.g. very high or very low sounds. In addition, components of the sound signals are omitted that are of no significance with regard to the sound impression but enabling the very high compression values.

MPEG is the abbreviation for Moving Picture Experts Group. The experts group is an organization that has created standards for the compression of film data. The standards for audio compression were later added.

MPEG layer 1 and layer 2 are the predecessors of MP3. They achieve compression rates which, however, do not approach those of Layer 3. Layer 1, for example, is used in connection with digital music cassettes and Layer 2 for DAB (Digital Audio Broadcast). All three compression methods are downward compatible, i.e. a Layer-3 decoder will also play files that are coded in accordance with Layer 1 or Layer 2.

The computing power for MPEG Layer 1 to 3 is to a high degree asymmetric. This means that the computing power for MP3 encoding is substantially higher than for MP3 decoding.

Since all the values achieved during the coding procedure are stored in the MP3 file, the work of the decoder is restricted to interpreting these values. All three layers support the following sample rates: 32 kHz, 44.1 kHz or 48 kHz. As part of MPEG audio coding, a filter array converts the audio signal to be encoded from the time range to the frequency range. During this procedure it is initially subdivided into 32 frequency ranges (sub-bands). Prior to MPEG Layer 2, all 32 sub-bands had the same width of 625 Hz. Only with the advent of Layer 3 have the sub-bands been adapted to the characteristics of the human ear (psychoacoustic model). The audio signal subdivided into frequency ranges provides the algorithm with distinctly more working area or depth for data reduction than the uniform audio signal.

The MP3 encoder examines each sub-band individually and removes frequencies that are deemed dispensable based on the fundamental psychoacoustic model and compresses the remaining audio data.