Trends in audio and speech compression for storage and real-time communication

Summary form only given. Advances in speech compression are generally driven by the need to conserve transmission rate or bandwidth, while maintaining an ability to reconstruct the speech at the receiver with good fidelity. For real-time communication a constant transmission rate and low signal delay for processing are additional requirements. For speech storage applications, variable rate compression techniques which match the short-time compression rate to the nature of the current signal are of significant interest. High-fidelity audio storage systems, such as compact discs require storage rates of roughly 700 kb/s without compression, 1400 kb/s for 2 channel stereo. Reduction to 128 kb/s and even 64 kb/s can be achieved without noticeable degradation by taking advantage of features of the auditory system that permit shaping the quantization noise in time and frequency so that it is best masked by the audio signal. The principal focus of speech compression research is in coding for wireless communication where capacity is limited by the scarce spectrum space available. In particular, a new generation of digital mobile transmission systems is being developed employing 8 kb/s speech coding