Fifty years of progress in speech waveform coding

Over the past 50 years, sustained research in speech coding has made it possible to encode speech with high speech quality at rates as low as 4 kb/s. The technology is now used in many applications, such as digital cellular phones, personal computers, and packet telephony. The early research in speech coding was aimed at reproducing speech spectra using a small number of slowly varying parameters. The focus of research shifted later to accurate reproduction of speech waveforms at low bit rates. The introduction of linear predictive coding (LPC) led to the development of new algorithms, such as adaptive predictive coding, multipulse and code-excited LPC. Code-excited LPC has become the method of choice for low bit rate speech coding and is used in most voice transmission standards. Digital speech communication is rapidly moving away from traditional circuit-switched to packet-switched networks based on IP protocols (VoIP). The focus of speech coding research is now on providing to low cost, reliable, and secure transmission of high-quality speech on IP networks.