In the past, the nonlinear effects of quantization in differential speech coders either were modeled as additive white Gaussian noise or were ignored in analysis and quantified experimentally. Describing functions are used to model the nonlinear quantizer effects of coarsely quantized difference signals found primarily in adaptive predictive coding (APC) systems. The analysis predicts marginal instabilities in APC encoders. The marginal instabilities result in ringing distortions in the reconstructed signal which, in the past, may have been attributed to pitch and spectral predictor mismatches. Adaptive order prediction is introduced, analyzed, and offered as a method for increasing the robustness of APC encoders.

The use of Kalman filtering (KF) techniques in speech coding is investigated. The authors show that the common linear predictor (LP) is a special case of the KF based on an all-pole signal model. They also show that the KF algorithm provides fixed-lag smoothing at no additional complexity. Simulation results reveal that KF based speech coding has significant advantage over the equivalent LP based systems, particularly when used with coarsely quantized measurements.>

A sufficient condition for the existence of tandem distortion accumulation is described. The distortion for a sample time at which the sufficient condition is satisfied is studied. It is shown that the transcoding distortion is greater than the adaptive differential pulse code modulation (ADPCM) distortion when synchronous code correction is not used and less than the ADPCM distortion when synchronous code correction is used. These results are combined to show that when synchronous code correction is used, the upper bound of the single transcoding distortion is determined by the single transcoding distortion when synchronous code correction is not used. This bound cannot be generalized for multiple tandem cases.>

Introduces the concept of multiply occupied cells which occur in the concatenation of codecs. The existence of a single multiply occupied cell of cardinality two is shown to imply that an alternative codec exists that results in lower end-to-end distortion and that eliminates the multiply occupied cell. The results are extended to the general case of multiple multiply occupied cells of any cardinality.>

It has been shown by example that the often used sign-sign variant of the LMS can be unstable and produce unbounded parameter estimates. In this paper a modified sufficient excitation condition on the regressor vector for the sign-sign variant is introduced. Satisfaction of the excitation condition guarantees that there exists a sufficiently small step size to produce parameter estimates which converge exponentially to within a region of the correct parameters. The counter example to stability of sign-sign algorithms is reexamined in light of the new excitation condition and it is shown that the excitation condition correctly predicts when instability will occur in this example.

This paper presents a real time implementation of a speech coder at 16 kbs. The compression algorithm is a combination of two previously reported techniques: Time Domain Harmonic Scaling and Adaptive Residual Coding. Several modifications are made to reduce the hardware complexity with minimal loss in speech quality, even in the presence of channel errors at a BER of 10-3. A single-board, real-time, full-duplex implementation is achieved using an NEC7720 signal processing chip and an 8085 microcomputer for transmission over an RS-423 digital interface.