This paper presents an analysis of the performance of a digital speech interpolation (DSI) system as a function of the traffic offered to the system. We assume that the DSI overload degradation occurs as clipping or freezeout. This is identical to TASI freezeout. Unlike conventional analysis, which assumes all of the trunks entering the DSI system are busy all the time, we calculate the freezeout fraction as a function of the traffic offered. In almost all practical situations the freezeout fraction, computed based on traffic assumptions, is considerably less than that computed assuming all trunks are busy. When a DSI system is used to compressNtrunks down intocchannels, there is always a tradeoff between the blocking probability in the trunk group and the degradation introduced by the DSI system. This tradeoff depends on the traffic offered and is explored in this paper.

A theoretical method of evaluating degradations of variable rate coders in a multichannel digital speech interpolation (DSI) system is developed. Each of the coder outputs has a variable rate based on the algorithm. The DSI system multiplexes the outputs of these variable rate coders into a fixed bit rate channel. During periods of high activity all active users are served, but at a reduced rate depending on the demand. The degradation due to high activity is shared by all active users. This system avoids speech clipping and "freeze-out" distortion. Theoretical expressions of the system overload probability and the probability of degradation to a particular user in the DSI system are derived. Two types of variable rate coders, namely, a constant quality subband coder and a constant noise subband coder, are chosen and used as examples. Comparisons of the degradations are made between the theoretical results and computer simulated results for the two types of variable rate coders, and close agreement is observed. The theory is applicable to other variable rate coding algorithms as well. In this study, all of the simulations are made at 40 percent speech activity and the average rate of the variable rate coders is close to 16 kbits/s. Objective quality measures indicate that in a system with a trunk size larger than 40, the variable rate coder DSI system can achieve a 2:1 compression with a degradation of less than 1 dB compared to non-DSI variable rate coders. This corresponds to a total gain of 8:1 when compared to 64 kbit/s PCM.

Permutation codes for the Laplacian source are developed. The performance of these codes is evaluated and compared with other quantizers and the rate-distortion function. It is shown that there is a bit-rate region in which the permutation codes outperform certain single-sample quantizers.

This paper describes the design of a digital speech interpolation (DSI) system called ADPCM/TASI for adaptive differential PCM with time assignment speech interpolation. This system is designed to compress the output of two T1 24-channel PCM carrier terminals into a 1.544 Mbit/s signal that can be transmitted over a single T1 carrier line. The design is based on a bit slice microprocessor structure. Alternative designs are also described.

Computer simulation was used to evaluate the performance of eleven coder/decoders (codecs) with phase shift keying (PSK) voiceband data signals. The codecs were PCM, differential PCM and delta modulation systems designed for speech and operating at bit rates from 16 to 64 kb/s. The voiceband data signals processed by these codecs were demodulated to determine the phase error caused by the codec. Three performance metrics were used to evaluate the performance of these codecs-signal to quantizing noise ratio, variance of the phase error and maximum value of the phase error.

Computer simulation was used to evaluate the performance of eleven coder/decoders (CODEC's) with phase shift keying (PSK) and differential PSK(DPSK) voiceband data signals. The CODEC's were PCM, differential PCM and delta modulation systems designed for speech and operating at bit rates from 16 to 64 kbits/s. The voiceband data signals processed by these CODEC's were demodulated to determine the phase error caused by the CODEC. The phase error introduced by the CODEC's is a function of the phase of the CODEC sampling clock relative to the data modem bit clock. Some of the statistics of the phase error are presented. Three performance metrics were used to evaluate the performance of these CODEC's-signal to quantizing noise ratio, variance of the phase error and maximum value of the phase error.

This paper presents a three-dimensional non-linear prediction technique for use in differential pulse code modulation systems used to encode television signals. The prediction is a weighted sum of three estimates called representatives. Each representative is a nonlinear estimate based on previous pels in one of three planes containing the pel to be predicted. The paper contains an explanation of the technique, results of computer simulations of the algorithm and a comparison of the algorithm with other prediction procedures. The predictor is relatively insensitive to image statistics. Errors in the predictor input decay out with time.

Rate-distortion functions for 2-dimensional homogeneous isotropic images are compared with the performance of five source encoders designed for such images. Both unweighted and frequency weighted mean-square error distortion measures are considered. The coders considered are a) differential pulse code modulation (DPCM) using six previous samples or picture elements (pels) in the prediction--herein called 6-pel DPCM, b) simple DPCM using single-sample prediction, c) 6-pel DPCM followed by entropy coding, d)8 \times 8discrete cosine transform coding, and e)4 \times 4Hadamard transform coding. Other transform coders were studied and found to have about the same performance as the two transform coders above. With the mean-square error distortion measure, 6-pel DPCM with entropy coding performed best. Next best was the8 \times 8discrete cosine transform coder and the 6-pel DPCM--these two had approximately the same distortion. Next were the4 \times 4Hadamard and simple DPCM, in that order. The relative performance of the coders changed slightly when the distortion measure was frequency weighted mean-square error. FromR = 1to 3 bits/pel, which was the range studied here, the performances of all the coders were separated by only about 4 dB.

A formula for signal-to-quantizing noise ratio (SNR) in delta modulation systems is obtained by modifying an existing result for DPCM systems. Computer simulation is used to examine the behavior of the terms in the modified SNR formula.

As the transmission rateRgets large, differential pulse-code modulation (PCM) when followed by entropy coding forms a source encoding system which performs within 1.53 dB of Shannon's rate distortion function which bounds the performance of any encoding system with a minimum mean-square error (mmse) fidelity criterion. This is true for any ergodic signal source. Furthermore, this source encoder introduces the same amount of uncertainty as the mmse encoder. The 1.53 dB difference between this encoder and the mmse encoder is perceptually so small that it would probably not be noticed by a human user of a high quality (signal-to-noise ratio(S/N) \geq 30dB) speech or television source encoding system.

This paper describes experiments with a television (TV) source encoder which consists of a differential PCM encoder followed by entropy coding. This encoder converts analog television signals into a digital bit stream for digital transmission or storage. When optimized, this type of system is known to perform very close to the rate distortion bound. The differential PCM encoder has a 16-level quantizer during low entropy areas of the picture (quiet areas) but switches to a 6-level quantizer in high entropy (busy) areas of the picture which tend to fill up the buffer. This strategy avoids buffer overflow and has the desirable property that it produces low noise in quiet areas of the picture and higher noise in busy areas of the picture.

This paper is an analytical study to determine the performance of single-integration delta-modulation(\DeltaM)encoders with inputs which are various voice-band data signals. Signal-to-quantizing noise ratiosS/Nare calculated for\DeltaMbit rates from 16 to 96 K bits/s. The input data signals studied are phase modulation at 1200 and 2400 bits/s, partial response at 4800 and 9600 bits/s, and single sideband at 4800 bits/s. Predictions of the performance of these modems, when transmitted over the\DeltaMsystems, are based on the calculatedS/Nratios.

A study of combining two ways of reducing the redundancy in the digital representation of speech signals is presented. Differential pulse-code modulation (DPCM) encodes the signal into digital form and reduces the redundancy due to correlation in adjacent sample values of the signal. Following this DPCM operation, entropy coding is used to reduce redundancy due to the unequal probabilities of the DPCM quantizer levels to be transmitted. Theoretical studies agree with Computer simalation results with real speech signals. The concepts of sliding entropy and sliding signal to quantizing noise(S/N)ratio are developed to measure the way in which the entropy andS/Nratio vary with time during a speech utterance. Plots of these quantities versus time for four different utterances are shown. Both adaptive and nonadaptive quantizers are studied. And both uniform and minimum mean-square error quantizing rules are included. Buffer length requirements are calculated for the entropy coders.

This paper contains an examination of the performance of low bit rate differential PCM systems when used to encode monochrome National Television System Commission (NTSC) television pictures. 12 different encoders with various sampling rates and numbers of quantizing levels operating at bit rates from 9 to 24 Mb were considered. The differential PCM systems were implemented by using an 8-b PCM A/D converter followed by digital logic that performed the differential operation in a simple previous sample feedback loop. The subjective performance of these encoders was determined using an anchored seven-point quality rating scale and equipreference contours were plotted in a plane whose axes are sampling rate and quantizing bits. Entropy measurements were made on the output of the differential PCM encoders. These entropy measures were compared with the subjective performance of encoding systems operating at the same transmission rate. The encoding systems studied produced television pictures of medium quality in which quantizing noise and bandwidth limitations were apparent.

The performance of differential PCM encoders used for both speech and data signals is investigated. Primarily concerned with data signals, this paper shows how differential PCM systems designed for speech perform with differential-phase-modulation, single sideband and partial response data signals operating at data rates of 1200-9600 bits/s. The seven signals considered in this study are shown in Table I. Also discussed are compromise differential PCM systems, which operate well for either speech or data but which are not optimum for either. These results show the advantages and limitations of using differential PCM on switched telecommunication networks carrying speech and data. This paper contains results generated by theoretical studies, computer simulation, and experiments with hardware encoders and modems.

Quantizing noise is present whenever analog information is encoded into digital form suitable as an input to any digital system such as a computer or digital transmission line. The subjective impairment caused by this noise is frequently measured by the ratio of signal power to frequency-weighted quantizing noise powerS/N_Y. An upper bound onS/N_Yis found such that source encoding systems will always have values ofS/N_Yless than this bound. The bound has the form (in decibels)S/N_Y \leq T_B + T_P + T_S, whereT_Bis a constant that depends on the hit rate of the signal,T_Pdepends on the redundancy (or predictability) of the signal, andT_Sdepends on subjective considerations (as embodied in a subjectively determined frequency-weighting function). The bound is applied to source encoding systems for speech and television signals. By using the frequency-weighting function, bounds on commonly used measures of subjective impairments are possible.

Much of the redundancy in a speech or television signal is eliminated when it is encoded into digital form by a differential pulse-code-modulation (DPCM) encoder. Additional coding of the DPCM output using entropy coding techniques (Huffman or Shannon-Fano coding) can result in a further increase in the signal-to-quantizing-noise ratio of 5.6 dB without increasing the transmission rate.

A modification of the conventional mutual-information figure of merit for feature selection in pattern recognition is described. A weighting function is combined with the mutual-information function that results in greater recognition accuracy for a linear classifier.