Search

The relation between subjective and objective speech intelligibility measurements is researched. For a large series of speech degradations, noise, linear and nonlinear distortions (speech codecs), intelligibility tests were carried out using short CVC words. In the subjective domain the percentage correctly identified words is taken as the intelligibility score for a certain type of degradation. In the objective domain ITU-T Recommendation P.862 is used as the starting point to develop a perceptual model that allows predicting the perceived intelligibility of a speech fragment.

Nearly all types of military speech communication involve the use of so-called (narrow band) voice coders or vocoders. Usually the Speech Transmission Index (STI) uses artificial test signals, which can not be reproduced by vocoders with the usual fidelity. Therefore the STI is not able to evaluate vocoders at this time. Although it is theoretically feasible to measure the Speech Transmission Index with natural speech instead of the usual artificial test signals, each of the various speech-based STI measurement methods proposed in the literature has its own shortcomings and inaccuracies. A new procedure is proposed for estimating a speech based modulation transfer function (MTF), on which the STI is based, that approaches the accuracy of conventional STI implementations based on artificial signals. The new method enables evaluation of vocoders by means of the STI. Applying the method to a voice coder database shows promising results, giving an average squared correlation coefficient R2 of 0.87 between the subjective CVC scores and the calculated STI for male speech.

A coherent description of architectures provides insight, enables communication among different stakeholders and guides complicated (business and ICT) change processes. Unfortunately, so far no architecture description language exists that fully enables integrated enterprise modeling. In this paper we focus on the requirements and design of such a language. This language defines generic, organization-independent concepts that can be specialized or composed to obtain more specific concepts to be used within a particular organisation. It is not our intention to re-invent the wheel for each architectural domain: wherever possible we conform to existing languages or standards such as UML. We complement them with missing concepts, focusing on concepts to model the relationships among architectural domains. The concepts should also make it possible to define links between models in other languages. The relationship between architecture descriptions at the business layer and at the application layer (business-IT alignment) plays a central role.

The NATO N4 corpus contains speech collected at naval training schools within several NATO countries. The speech utterances comprising the corpus are short, tactical transmissions typical of NATO naval communications. In this paper, we report the results of some preliminary speaker recognition experiments on the N4 corpus. We compare the performance of three speaker recognition systems developed at TNO Human Factors, the US Air Force Research Laboratory, Information Directorate and MIT Lincoln Laboratory on the segment of N4 data collected in the Netherlands. Performance is reported as a function of both training and test data duration. We also investigate the impact of cross-language training and testing.