In this study, a duration{based measure is formulated for assigning con dence scores to phonetic time{alignments produced by an automatic speech segmentation system. For speech corrupted by additive noise or telephone channel environments, the proposed con dence measure is shown to provide a reliable means by which gross segmentation errors can be automatically detected and marked for human han...

It is argued that standard impulse response analysis based on vector autoregressive models has a number of shortcomings Although the impulse responses are estimated quantities measures for sampling variability such as con dence intervals are often not provided If con dence intervals are given they are often based on bootstrap methods with poor theoretical properties These problems are illustrat...

Explicit computation of the score vector and the observed information matrix in hidden Markov models is described. With the help of the information matrix Wald's con dence intervals can be formed for the model parameters. Finite sample properties of the maximum-likelihood estimator and its standard error are investigated by means of simulation studies. We compare the con dence levels of interva...

In this paper, the speech understanding problem in the context of a spoken dialog system is formalized in a maximum likelihood framework. Word and dialog-state n-grams are used for building categorical understanding and dialog models, respectively. Acoustic con dence scores are incorporated in the understanding formulation. Problems due to data sparseness and out-of-vocabulary words are discuss...

It is argued that standard impulse response analysis based on vector autoregressive models has a number of shortcomings Although the impulse responses are estimated quantities measures for sampling variability such as con dence intervals are often not provided If con dence intervals are given they are often based on bootstrap methods with poor theoretical properties These problems are illustrat...

For many practical applications of speech recognition systems, it is desirable to have an estimate of con dence for each hypothesized word, i.e. to have an estimate which words of the speech recognizer's output are likely to be correct and which are not reliable. Many of today's speech recognition systems use word lattices as a compact representation of a set of alternative hypothesis. We explo...

This paper investigates the empirical relation between consumer and expert expectations about macroeconomic conditions in Germany. Using data from the EU Consumer Con…dence Survey and the ZEW business expectations, I estimate static models and error-correction models explaining consumer expectations as functions of expert expectations, consumer retrospections, and lagged consumer expectations. ...

In response to the strawman document [9] we propose that trust be treated as synonymous with integrity rather than synonymous with con dence. We also propose that mandatory controls be taken to mean controls based on properties of the object and/or the subject. Label-based mandatory controls are then a special case of this more general notion. The TCSEC [11] presents criteria for establishing p...

Monte Carlo and quasi Monte Carlo methods are simulation techniques that have been de signed to e ciently estimate integrals for instance Quasi Monte Carlo asymptotically outperforms Monte Carlo but the error can hardly be estimated We propose here to recall how hybrid Monte Carlo Quasi Monte Carlo have been developed to easily get error estimations with a special em phasis on the so called ran...

We consider the problem of deriving one-sided tolerance intervals in the one-way random model with balanced as well as unbalanced data, under the usual normality assumptions. The problems investigated deal with the computation of such intervals for the observable random variable, as well as the unobservable random effect in the one-way random model. The tolerance limits are derived using the co...