MP3 is an MPEG compression based on MPEG-1-Audio Layer-3 for the compression of audio signals. The process, developed by the Fraunhofer Institute in Erlangen, is characterized by high playback quality with a compression factor of over 10.
According to the process, inaudible signal components are filtered out in MP3. This masking uses bandpass filters of different bandwidths that cover the human hearing range. This sub-band encoding is done with 32 sub-bands, which in turn are divided into 18 sub-ranges, so that a total of 576 bandpasses are used for filtering. The filtering is accompanied by a DCT transform or a Fourier transform, which transfers the time-related audio signal to the frequency domain. The filtering is followed by an evaluation of the information content of the individual filter components.
This evaluation determines the quantization. Here, each subband has a masking threshold that depends on the soundpressure of the audio frequency. For example, if there is a sound in the subband around 1 kHz with a loudness of 65 dB, then the encoder uses the psychoacoustic model to determine that the masking threshold for this subband is 30 dB, so the sound must be quantized at 35 dB, which can be done with 6-bit digitization. If the sound has a volume of 80 dB with the same masking, it must be digitized with 7 bits to achieve the dynamic range of 50 dB. MP3 supports three main modes: monophonic, stereophonic and joint stereo. MP3 files are played using MP3 players such as Winamp, Macamp or mpeg123. The extension for MP3 files is *.mp3.
In addition to the standard MP3 version, there is an extended version: Enhanced MP3. In this version, more extensive additional information can be integrated into the MP3 file. The further development of MP3 has led to mp3Pro. MP3 files are made available on the Internet for download or podcast.