Communications Research Centre Canada
Symbol of the Government of Canada

Audio Source Coding

A stereo audio signal of compact disk quality requires 1.4 megabits (about 1/8 the capacity of a 3.5" floppy diskette) to transmit or store one second of audio data. Using audio source coding techniques, this amount of data can be reduced by a factor of 10 to 15 without apparent loss in audio quality. To achieve such high compression ratios, audio source coding systems rely on sophisticated signal processing techniques to remove redundancies for the audio signal and on a model of the human auditory system to remove components of the audio signal which are deemed inaudible. Audio source coding has enabled the introduction of new audio services such as digital audio broadcasting as well as Internet radio and the distribution of music over the Internet.

Over the past few years many laboratories have been involved in the design and development of new audio coding systems. Some of these systems have become international or continental standards. The MPEG Layer 2, developped jointly by the CCETT, the IRT and Philips, the MPEG Layer 3 (or MP3) from FhG, Dolby Labs'AC-3 and the MPEG AAC (from AT&T, FhG, Dolby Labs, Sony, University of Hanover) are good examples. Other codecs such as Lucent's ePAC or QDMC from the Canadian QDesign company are used in some multimedia or Internet applications. Most of these well known coding systems are available commercially and decoders can operate in real time on a Pentium personal computer. There has been a rapid improvement in the efficiency of the encoder part of these codecs which operate over data rates ranging from 8 to 640 kbit/s per stereo pair. The best codecs can compress the rate a stereo pair down to around 100 kbit/s without affecting the audio quality significantly. Many of the coding systems available support multichannel audio.

Research is being conducted in our group in order to improve the psychoacoustical models used in audio codecs. Our group is also conducting subjective evaluation of audio codecs in collaboration with standards organizations or under contract with private companies.