The Motion Picture Experts Group (MPEG ) audio compression algorithm is an International Organisation for Standardisation (ISO) standard for high-fidelity audio compression. MPEG audio is one of a three-part compression standard for synchronised video and audio.

Like many compression algorithms, MPEG audio compression is lossy, meaning that the output signal is not identical to the input signal, having lost some data discarded during the compression process. This lost data is not perceptible to the human ear due to the use of auditory masking -

" a perceptual weakness of the ear that occurs whenever the presence of a strong audio signal makes a spectral neighbourhood of weaker audio signals imperceptible" [26]

The compression is therefore achieved by removing parts of the audio signal which are considered irrelevant to the human ear (such as those frequencies beyond our range of hearing, known as signal redundancy ), resulting in a slight distortion of the sound which is imperceptible to the human ear and therefore almost identical to the original input (uncompressed) sound. Most audio encoding techniques undergo stringent listening tests and psychoacoustic experiments to guarantee the effectiveness of the auditory masking.

As well as using a psychoacoustic model to mask the encoded signal, an MPEG encoder also feeds the input signal through a time-to-frequency mapping device that constructs sets of spectrum components for subsequent encoding. A certain number of data bits is used by the encoder to form a template for the entire length of the input signal, and subsequent information is stored in relation to the initial data, rather than as data in its own right. This concept is similar to the idea of storing ten letter As as A(x10) or 10A rather than as AAAAAAAAAA - the nine subsequent letters are stored in relation to the first to take up less storage space.

MPEG Audio Layer III (more commonly known as MP3 due to its .mp3 file extension) has become a popular format for digital music delivery because of the smaller amounts of storage space required, and because of its excellent sound quality which even survives multiple coding (signals are often coded several times to serve different tasks). The four most common areas of usage of MPEG audio in the professional and consumer markets are

• broadcasting - cable radio, digital audio broadcasting, cable and satellite TV, internet radio
• storage - consumer recording (Digital Compact Cassette), disk-based editing etc
• multimedia - CD -i, CD -Video, Digital Versatile Disc, computer-based multimedia
• telecommunications - stand-alone electronic information systems, distribution links etc. [27]

Comparing the compression figures and audio download times in the table above with the one below, we see how compressing audio data can significantly decrease download times. With a 28.8 modem a full CD of uncompressed sound would take 36 hours to download whereas the compressed CD requires but 3½ hours. Using compression companies can now reduce the download time by a factor of 10 or more while still maintaining sound quality.

Table: Music download times for compressed and uncompressed music(Source: a2b music) [28]

	28.8 kbps	ISDN 56.6 kbps	ISDN 2B+D 112 kbps	Satellite 400 kbps
Full CD (477 Mb)	36 hours	18 hours	9 hours	2.6 hours
Compressed CD (32.4 Mb)	2.5 hours	1.3 hours	38 mins	11 mins
5 minute song  (53 Mb)	4 hours	2.1 hours	1.05 hours	18 mins
Compressed 5 min song (3.6 Mb)	16 mins	8.5 mins	4.25 mins	1.2 mins

 

The CD -audio data rate is 1.4 Mbps. compressed audio at 128Kbps (11:1) provides CD -quality, at 64 Kbps (22:1) the quality is "CD -like." Using similar compression techniques, a single CD can hold 22 hours of high-quality compressed music rather than just 74 mins" uncompressed music.[28]
 

Back: Digital audio compression | Contents | Next: What is streaming media?