Discusses the problem of robust parametric model estimation and classification in noisy acoustic environments. Characterization and modeling of the external noise sources in these environments is in itself an important issue in noise compensation. The techniques described provide a mechanism for integrating parametric models of the acoustic background with the signal model so that noise compensation is tightly coupled with signal model training and classification. Prior information about the acoustic background process is provided using a maximum likelihood parameter estimation procedure that integrates an a priori model of the acoustic background with the signal model. An experimental study is presented on the application of this approach to text-independent speaker identification in noisy acoustic environments. Considerable improvement in speaker classification performance was obtained for classifying unlabeled sections of conversational speech utterances from a 16-speaker population under cross-environment training and testing conditions.>

The Lincoln hidden Markov model (HMM)-based word spotting system has demonstrated good performance in spotting keywords in completely unconstrained continuous speech utterances (see R.C. Rose and D.B. Paul, 1990). The word spotter has been evaluated under a number of scenarios, and has been integrated into a system that performs the higher level of classifying conversational speech messages according to topic. In all of these scenarios, anywhere from 25 to 78 examplars per keyword have been used to train the subword acoustic HMMs that are used in the word spotter. In most word spotting applications it is simply not possible to collect such a large number of spoken utterances for all the keywords in the vocabulary every time the system is to be reconfigured for a given task. Therefore, it is essential that techniques be developed to reduce the amount of task-specific speech data required for training HMM-based work spotters.>

The use of probabilistic mixture densities for text-independent speaker identification in a noisy telephone channel environment is investigated. Two techniques for noise compensation are considered. In the first approach, a background noise model is integrated directly into the model for speech. In the second approach, noise preprocessing techniques are used to compensate noisy observations before passing them along to the speaker identification on conversational utterances collected over long-distance telephone channels from ten speakers. The integrated noise model was noted for having a noise suppression characteristic that arises naturally from the statistical model, which is important if one wishes to avoid the ad hoc fine tuning of thresholds required in the noise processing approach implemented.>

Methods are investigated for short-time analysis and synthesis of signals from a class of second-order difference equations with a cubic nonlinearity. In analysis, two methods are explored for estimating equation coefficients: (1) prediction error minimization (a linear estimation problem) and (2) waveform error minimization (a nonlinear estimation problem). In the latter case, which improves on the prediction error solution, an iterative analysis-by-synthesis method is derived which allows as free variables initial conditions, as well as equation coefficients. Parameter estimates from these techniques are used in sequential short-time synthesis procedures. Possible application to modeling quasi-periodic behavior in speech waveforms is discussed.>

A packetized speech multiplexer differs from a circuitswitched TASI system in that the presence of a packet buffer allows a tradeoff where the TASI advantage can be increased at a cost in packet delay. This tradeoff is investigated via a simulation. Results are presented to show the relations between TASI advantage and delay, for both an average delay criterion and a maximum delay criterion. It is shown that, particularly for the case where small numbers of talkers are multiplexed, the packetized system offers significant improvements in TASI advantage over the conventional circuit-switched multiplexer, at modest costs in packet delay.

The Schwarz inequality is used to derive the Barankin lower bounds on the covariance matrix of unbiased estimates of a vector parameter. The bound is applied to communications and radar problems in which the unknown parameter is embedded in a signal of known form and observed in the presence of additive white Gaussian noise. Within this context it is shown that the Barankin bound reduces to the Cramér-Rao bound when the signal-to-noise ratio (SNR) is large. However, as the SNR is reduced beyond a critical value, the Barankin bound deviates radically from the Cramér-Rao bound, exhibiting the so-called threshold effect. The bounds were applied to the linear FM waveform, and within the resulting class of bounds it was possible to select one that led to a closed-form expression for the lower bound on the variance of an unbiased range estimate. This expression clearly demonstrates the threshold behavior one must expect when using a nonlinear modulation system. Tighter bounds were easily obtained, but these had to be evaluated numerically. The sidelobe structure of the linear FM compressed pulse leads to a significant increase in the variance of the estimate. For a practical linear FM pulse of 1-\mu sduration and 40-MHz bandwidth, the radar must operate at an SNR greater than 10 dB if meaningful unbiased range estimates are to be obtained.

In a recent paper on radar detection in clutter, the authors stated that the signal-to-interference ratio\rho_{cf} (u)obtained with the signaluand its optimum filter can have stationary points which are not stationary points of\rho_{mf} (u), the signal-to-interference ratio obtained withuand its matched filter, even when the clutter distribution in range and Doppler is symmetric. This letter strengthens this statement and exhibits an example to support it. The stronger statement is the following: Even for symmetric clutter distributions, the global maximum of\rho_{cf}can exceed the global maximum of\rho_{mf}.

The problem of designing radar signals and receivers that are optimum for detecting a point target masked by thermal noise and clutter returns is considered. In a previous paper, the authors described an iterative algorithm for designing such signals subject to an energy constraint. This paper describes an algorithm for the case where the energy constraint is replaced by a dynamic-range constraint, that is, the ratio of maximum to minimum signal amplitude is limited. The signal to which the algorithm converges satisfies the Kuhn-Tucker conditions, which are necessary for optimality. Examples of the application of both algorithms to radar signal design problems are given.

This paper considers the problem of detecting and estimating the unknown angular location of a radar target that is observed simultaneously by a number of antennas. The amplitude of the signal received by a particular antenna is assumed to depend on the angular location of the target, but the time of arrival of the signal is assumed to be the same at all of the antennas. The generalized likelihood ratio test is used to derive the detection and parameter-estimation strategy for the radar receiver. Explicit expressions for the detector and the angle estimates are derived in a number of important special cases, and the performance of the detector is evaluated. Accuracy formulas for the angle estimates, valid for high signal-to-noise ratios, are derived for the general problem and used to compare the performance of an optimum four-beam monopulse system with a type of monopulse system commonly in use.

The problem to be considered in this paper is that of designing radar signals and receivers that are optimum for detecting a point target masked by a background of clutter returns and thermal noise. The problem of choosing an optimum signal when no constraints are placed on the type of signals allowed is discussed briefly, but the remainder of the paper is restricted to signals and receiver impulse responses that are uniformly spaced, phase and amplitude-tapered pulse trains. Expressions for the signal-to-interference ratio obtained when a signal is used with its matched filter (pm/) and with the optimum filter or clutter filter(\rho_{cf})are then derived together with an explicit expression for the clutter filter. An iterative technique for maximizing\rho_{cf}is devised. This scheme has the useful property that it generates a sequence of signals whose\rho_{cf}'s form a monotone, nondecreasing sequence. This is followed by an application of the calculus of variations to derive the Euler equations for the stationary points of\rho_{cf}and\rho_{mf}. The form of the Euler equations suggests iterative techniques for their solution; in fact, the technique suggested for the solution of the Euler equation associated with\rho_{cf}is essentially the iterative technique that was described above.

The problem of finding maximum-likelihood estimates of the partially or completely unknown autocorrelation function of a zero-mean Gaussian stochastic signal corrupted by additive, white Gaussian noise is analyzed. It is shown that these estimates can be found by maximizing the output of a Wiener estimator-correlator receiver biased by a smoothed version of the output noise-to-signal ratio of the Wiener estimator over the class of admissible autocorrelation functions. For the case where the autocorrelation function is known except for an amplitude scale parameter, an illuminating expression for the Cramer-Rao minimum estimation variance is derived. Detailed study of this expression yields, among other results, an interpretation of the maximum-likelihood estimator as an adaptive processor.

Augmenting Woodward's total-volume invariance, new information on ambiguity-function behavior is obtained and illustrated by example. The results are of two types, the first dealing with the time4requency volume distribution and providing lower bounds to local volume, and upper bounds to attainable "clear area"; these follow from Parseval's theorem as applied to Siebert's self-transform property. The second category of necessary conditions consists of a set of upper bounds on the norms of the ambiguity function, which places restrictions on the height distribution of the ambiguity surface. As a corollary, it is demonstrated that not all nonnegative functions that are serf-transformable can be ambiguity functions.

A necessary and sufficient condition is derived for a function to be the autocorrelation function of a time-limited function of specified length. A constructive procedure for determining all possible time-limited functions having the same autocorrelation function as a given time4imited function is presented. The procedure consists of passing the given time function through one out of a collection of specially tailored all-pass filters. The class of functions so generated may be either finite or infinite depending on the original time function. Specific examples of the construction procedure are given.