Mixed pseudo analogue-digital speech and audio coding

Current speech, audio, and video coding and transmission systems are either analogue or digital, with a strong shift from analogue systems to digital systems during the last decades for the benefit of exploiting digital channel coding for error correction. Combining both, digital and analogue schemes results in the benefit of saving transmission bandwidth, complexity, and of improving the achievable quality at any given signal-to-noise ratio on the channel within the range of interest. The combination was achieved by transmitting pseudo analogue samples of the unquantized residual signal of a linear predictive digital filter. This principle, called Mixed Pseudo Analogue-Digital (MAD) transmission, is applied to both, narrowband, and wideband speech, as well as to audio signals. After introduction of the MAD transmission principle, this contribution examines the performance of the novel scheme for speech and audio transmission over a channel modelled as fading Additive White Gaussian Noise (AWGN with flat fading) with Rayleigh fading. An implementation of MAD transmission is compared to the GSM Adaptive Multi-rate speech codec mode 12.2 kbit/s (Enhanced Fullrate Codec, EFR), which uses a comparable transmission bandwidth if channel coding is included. The simulative results are backed by a thorough information theoretical analysis of the principles used in MAD transmission, pointing out that the increased performance mainly stems from the combination of digitally transmitting the spectral envelope of the signal while at the same time the Gaussian residual signal is the optimum input for the AWGN channel. Modulation schemes using the Archimedes Spiral for mapping the pseudo analogue residual to a 2-dimensional signal space are theoretically motivated and developed to enhance the quality of the basic system. Finally, possible applications like MAD microphones and headsets are suggested and further prospects like channel adaptive MAD are briefly given