Showing papers on "Naturalness published in 1999"

Color reproduction and the naturalness constraint

Abstract: This article aims to determine the process underlying the subjective impression about the fidelity of reproduced object colors. To this end, we present the concept of the naturalness constraint and a framework for specification of naturalness judgments. We consider several research questions that are essential for this framework and discuss plausible answers supported by experiments. In general, naturalness assessment of reproduced object colors can be (1) defined as similarity to prototypical object colors, and (2) characterized by a probability density function (e.g., Gaussian). Experiments show that (3) there is a considerable amount of consistency in naturalness judgments of locally and globally processed images (although observers are slightly more tolerant of global image processing), and (4) naturalness judgments vary for different object categories; e.g., subjects are more consistent in naturalness judgments of skin, grass, and sky reproductions than shirt reproduction. We suggest that (5) naturalness of a whole picture is determined by the naturalness of the most critical object in that picture. Finally, we introduce a naturalness index predicting perceived naturalness of color reproduction of real-life scenes. © 1999 John Wiley & Sons, Inc. Col Res Appl, 24, 52–67, 1999

Naturalness of supersymmetric models

Abstract: After presenting a simple procedure for testing naturalness (similar to Bayesian inference and not more subjective than it) we show that LEP2 experiments pose a naturalness problem for `conventional' supersymmetric models. About 95% of the parameter space of minimal supergravity MSSM is excluded by LEP2 experiments. Moreover in this model electroweak baryogenesis, or detectable supersymmetric corrections to mixing of K and B mesons, are possible only in very small corners of the parameter space. The naturalness problem is stronger in gauge mediation models, expecially with light messengers. We recall some possible explanations (different from an improbable numerical accident) of why supersymmetry has not (yet?) been found.

Computational image quality

Abstract: In this paper we will discuss image quality in the context of the visuocognitive system as an information-processing system. To this end, we subdivide the information-processing as performed by the visuo-cognitive system into three distinct processes: (1) the construction of a internal representation of the image; (2) the interpretation of this representation by means of a confrontation with memory; and (3) task-directed semantic processing of the interpreted scene in order to formulate a proper response. A successful completion of these processes can only be ensured when two main requirements are satisfied: (1) the internal representation of the image should be sufficiently precise; and (2) the degree of correspondence between the internal representation and “knowledge of reality” as stored in memory should be high. We then relate these requirements to the attributes “usefulness” and “naturalness” of the image, and give a functional description of image quality in terms of naturalness and usefulness. To conclude, experimental results supporting this description of image quality will be discussed. 0This chapter is a slightly modified version of Janssen & Blommaert (1997). 7 2. Image quality semantics

Exploring the naturalness of several German high-quality-text-to-speech systems.

Abstract: The synthesis of near-to-natural F0 contours is an important issue in text-to-speech and crucial to the naturalness and intelligibility of synthetic speech. In earlier studies of the first author a model of German intonation was developed that is based on the quantitative Fujisaki-model. The current paper addresses a perception experiment comparing a TTS-system incorporating this new approach with several German TTS-systems with high segmental quality. Natural speech samples and a synthesis version with natural segment durations were used as references. Results show, that the natural speech samples unanimously received 10 points on a 0 to 10 point scale. The best TTS-systems cluster around a mean value of 5.0, whereas the variant with natural durations reached a mean score of 6.6 points, indicating the importance of closely modeling natural segment durations.

A new model of rank-frequency distribution

Abstract: We assume that if there is a homogeneous class of musical or linguistic elements set up on the basis of a property that is relevant for the use of music or language, then the exploitation of these elements abides by its own laws in the sense that their frequency of occurrence displays a regularity which can be modelled by a monotone decreasing rank-frequency distribution. This assumption has three consequences: (i) The frequencies of occurrence of members of any linguistic or musical (jointly “textual”) class are not equal, i.e., they do not follow the discrete uniform distribution, (ii) it is a consequence of some generating mechanism, i.e., it is lawlike (cf. Bunge, 1967, p. 16: “Everything abides by laws”), and it must have developed historically in order to fulfil some human needs (cf. Kohler, 1991), (iii) if one succeeds in finding some laws capturing this phenomenon (and there are a number of successful models) then these laws are in turn the best criteria of “correctness”, “naturalness”, “prolificn...

Superheavy Supersymmetry from Scalar Mass--A Parameter Fixed Points

Abstract: In supersymmetric models, the well-known tension between naturalness and experimental constraints is relieved if the squarks and sleptons of the first two generations are superheavy, with masses of order 10 TeV, and all other superpartners are light, with masses of order 1 TeV. We show that even if all scalar masses and trilinear A parameters are of order 10 TeV at some high scale, a mass-squared hierarchy of order 400 may be generated dynamically through renormalization group evolution. The required high energy relations are consistent with grand unification, or, alternatively, may be realized in moduli-dominated supersymmetry-breaking scenarios.

Spatial audio technology for Telepresence

Abstract: A key requirement to enhance the usability of a range of telepresence applications is the exploitation of all the sensory modalities, thus maximizing naturalness. Spatial audio has a vital role to play in enhancing speech communication performance and naturalness in synthetic spaces by giving directionality to individual sound sources and providing feedback about the virtual space in which the user is located. When groups of people meet in synthetic spaces either for business or recreation it is possible to provide services and features which are not available in a face-to-face meeting. In particular, interaction with intelligent agents, manipulation of virtual 3-D objects, data sharing, and automatic note taking may all seem quite natural in the enhanced meeting spaces of the future.

Designing for Naturalness in Automated Dialogues

Abstract: Many of the automated telephone dialogue systems currently in commercial operation are intended to be used by the general public. This means that the typical human callers who use the system/s are essentially novice users who do not build up expertise over a period of time To cater for such a user group, designers strive to make their dialogues as ‘natural’ as possible. In this paper we discuss and illustrate what ‘naturalness’ means in the context of the kind of automated dialogues which are at present in operation. We present some examples from automated dialogues, focusing particularly on openings, closings, confirmations and repairs, and compare these with the way similar operations are carried out in human-human dialogues. We summarise a set of experiments carried out recently to investigate how callers respond to supposed ‘naturalness’ in automated dialogue systems, and we conclude with some comments which challenge current ideas about what really is ‘natural’ for an automated dialogue.

Naturally Heavy Scalars in Supersymmetric Grand Unified Theories

Abstract: The supersymmetric flavor, CP and Polonyi problems are hints that the fundamental scale of the soft supersymmetry breaking parameters may be above a TeV, in apparent conflict with naturalness. We consider the possibility that multi-TeV scalar masses are generated by Planck- or unification-scale physics, and find the conditions under which the masses of scalars with large Yukawa couplings are driven, radiatively and asymptotically, to the weak scale through renormalization group evolution. Light third generation scalars then satisfy naturalness, while first and second generation scalars remain heavy to satisfy experimental constraints. We find that this mechanism is beautifully realized in the context of grand unified theories. In particular, the existence of right-handed neutrinos plays an important role in allowing remarkably simple scenarios. For example, for SO(10) boundary conditions with the squared masses of Higgs scalars double those of sleptons and squarks, we find that the entire scalar mass scale may be increased to 4 TeV at the unification scale without sacrificing naturalness.

The Naturalness Theory of Laws

Abstract: If I were to drop an apple, then it would fall. It is not possible that it would fly upwards. It is necessary that it would fall. I think these statements are true, but at the same time I believe that it is not logically necessary that the apple would fall. I believe that there is in nature a kind of necessity weaker than logical necessity: natural necessity.

Comparing the naturalness of several approaches for generating F0 contours in German text‐to‐speech systems

Abstract: Generating near‐to‐natural F0 contours is an important issue in text‐to‐speech synthesis and contributes vastly to the quality of synthetic speech. In earlier studies by the authors, a model of German intonation was developed that is based on the quantitative Fujisaki model. A typical F0 contour is described as a sequence of major rises and falls, which are modeled by onsets and offsets of accent commands connected to accented syllables. The current paper addresses perception experiments comparing the intonational naturalness of a Fujisaki‐model‐based TTS and four other German TTS systems with comparably high segmental quality. Natural speech samples were used as a reference. Three of the TTS systems had PSOLA, and one LPC segmentals. Two types of experiments were conducted with 20 subjects: (1) a pair comparison of 15 isolated sentences, (2) a ranking test based on a news passage of about 15 sec produced with each of the systems. Preliminary results from experiment (1) show, that on a naturalness scale from 0 to 5, the natural speech samples reach a maximum score of 4.5, with values of 2.8 for the best synthesis, the LPC‐based one. The system with Fujisaki‐model‐based intonation leads the group of PSOLA systems, which is closely clustered at a mean of 2.4.