Hirst, D., Di Cristo, A., Espesser, R. Levels of representation and levels of analysis for the description of intonation systems. In Horne, M. (Ed.), Prosody: Theory and Experiment, Dordrecht: Kluwer Academic Publishers. 2000 96  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... that can fairly easily be studied with the aid of a good neural network simulation package, such as MATLAB or the Stuttgart Neural Network Simulator (SNNS) The last problem might be solved by using the neural network methodology to predict values for an underlying intonation model (e.g. MOMEL [24]) which already o#ers a solution to the microprosody problem. Appendix A DATABASE LABELING CRITERIA AND SOME STATISTICAL DESCRIPTIONS OF THE DATA This appendix contains a summary of the labeling criteria used in the databases listed in Chapter 4. Some statistical analyses of the larger ....

Daniel Hirst, Albert Di Christo, and Robert Espesser. Levels of representation and levels of analysis for the description of intonation systems. In Merle Horne, editor, Prosody: Theory and Experiment -- Studies Presented to Gosta Bruce, pages 37--88. Kluwer Academic Publishers, 2000.

....the underlying model of Hirst s method is particularly simple, and the representation by means of target points very economical. 3.2. Algorithm An efficient algorithm for computing target points (MOMEL, standing for MOdlisation de MELodie) was proposed by Hirst and Espesser ( 9] see also [8]) The MOMEL algorithm consists in four stages (we summarize the description given by [9] 1. Aberrant values due to F 0 detection errors are eliminated. 2. A quadratic regression technique is applied in a moving analysis window, which computes a candidate target for all F 0 values in that ....

....eliminate the majority of them [3] 4. SYMBOLIC CODING Various systems for symbolic coding of intonation have been proposed. They can be categorized in two types: linguistic systems, such as ToBI [13] which encode events of a linguistic nature, and phonetic systems, such as HLCB [15] or INTSINT [8], which aim only at providing a purely configurational description of the macroprosodic curve without interpretation. Systems of the first category are probably more desirable, but (1) are not straightforward to adapt to new languages; 2) pose difficult problems for automatic labeling (see [1] ....

Hirst, D.J., Di Cristo, A. and Espesser, R. "Levels of representation and levels of analysis for the description of intonation systems." In Horne, M. (Ed.), Prosody: Theory and Experiment, Kluwer Academic Publishers, Dordrecht, forthcoming.

....surprising differences across language and sex. This paper is only intended to present raw results, and will not attempt at theoretical explanations of the observed phenomena. 2. STYLISATION METHOD The stylization method used in this study was proposed by Hirst and Espesser [7] see also [6]) Their algorithm (MOMEL, for MOdlisation MELodique) reduces the F 0 of the signal to a series of pitch target points that represent the relevant macromelodic movements of the utterance (Figure 1) Once interpolated by a quadratic spline curve, these points generate an F 0 contour which (apart ....

Hirst, D.J., Di Cristo, A. and Espesser, R. "Levels of representation and levels of analysis for the description of intonation systems." In Horne, M. (Ed.), Prosody: Theory and Experiment, Kluwer Academic Publishers, Dordrecht, forthcoming.

....of re generating an F 0 curve perceptually identical to the original from the extracted prosodic labels. Prosodic coding systems can be categorized in two types: linguistic systems, such as ToBI, which encode events of a linguistic nature, and phonetic systems, such as HLCB [8] or INTSINT [5], which aim only at providing a purely configurational description of the macroprosodic curve without interpretation. It is obvious that the first category poses greater problems for the automatisation, although work is underway in that area ( 1] 7] The second category, easier to implement, is ....

....be trivially reached by increasing the number of categories, but usable systems must obviously aim at keeping this number minimal. We use a two stage model for coding: 1. First, the F 0 curve is stylized by means of target points which capture the macroprosodic information of the utterance [5]. 2. Target points are then converted to a sequence of labels constituting a categorization of the pitch movements. The pitch target points can be extracted automatically from the signal ( 6] see also [5] Once interpolated by a spline curve they produce an F 0 contour undistinguishable from ....

[Article contains additional citation context not shown here]

Hirst, D.J., Di Cristo, A. and Espesser, R. "Levels of representation and levels of analysis for the description of intonation systems." In Horne, M. (Ed.), Prosody: Theory and Experiment, Kluwer Academic Publishers, Dordrecht, forthcoming.

....attempts have been made to do this (Wightman and Ostendorf 1992; Ostendorf and Ross, 1997) Finally, the regeneration of an F 0 curve from the ToBI coding is far from obvious. In our study, we use a symbolic coding system INTSINT (INternational Transcription System for INTonation) proposed by Hirst and Di Cristo. 1998) and Hirst et al. forthcoming) and which has been used for the manual transcription of intonation patterns in a number of languages (see different chapters in Hirst and Di Cristo, 1998) Unlike ToBI, which encodes events of a linguistic nature, INTSINT aims to provide a purely formal encoding of ....

....study, we use a symbolic coding system INTSINT (INternational Transcription System for INTonation) proposed by Hirst and Di Cristo. 1998) and Hirst et al. forthcoming) and which has been used for the manual transcription of intonation patterns in a number of languages (see different chapters in Hirst and Di Cristo, 1998). Unlike ToBI, which encodes events of a linguistic nature, INTSINT aims to provide a purely formal encoding of the macroprosodic curve. Each target point of the stylised curve is coded by a symbol either as an absolute tone, defined globally with respect to the speakers pitch range, or as a ....

Hirst, D.J., Di Cristo, A., Espesser, R. 1998. Levels of representation and levels of analysis for the description of intonation systems. In Horne, M. (Ed.), Prosody: Theory and Experiment, Dordrecht: Kluwer Academic Publishers (forthcoming).

.... rules: R1) division of the utterance into Intonation Units (IUs) R2) assignment of a prominence to the right edge of each IU; R3) assignment of prominence to the left and the right edge of each accentuable word; R4) reorganisation of prominences to take into account their relative proximity [11]. The construction of the metrical grid also makes use of a dominance principle (stress subordination) which iteratively assigns a higher level of prominence to a final stress than to an initial stress, the highest level being assigned to the nuclear stress. This hierarchical strategy accounts ....

Hirst, D., Di Cristo, A., Espesser, R. (forthcoming). Levels of representation and levels of analysis for the description of intonation systems. In Horne, M. (Ed.), Prosody: Theory and Experiment, Dordrecht: Kluwer Academic Publishers.

....coding which allows the generation of an F 0 curve which is very close to the original curve. The system has been applied successfully to French and to Italian. 2. STYLISATION The method of stylisation used in this study: MOMEL (MOd lisation de MELodie) was originally proposed by [7] see also [8]) Contrary to most other methods of stylisation which use a sequence of straight line segments, MOMEL uses a quadratic spline function (sequence of parabolic segments) resulting in a continuous, smooth curve, without the angles which occur when using straight lines. Unvoiced segments are ....

....to carry out automatically, although attempts have been made to do this [1] Finally, the regeneration of an F 0 curve from the ToBI coding is far from obvious. In our study, we used the symbolic coding system INTSINT (INternational Transcription System for INTonation) described by [9] and [8]. Unlike ToBI which encodes events of a linguistic nature, INTSINT aims to provide a purely formal encoding of the macroprosodic curve. Each target point of the stylised curve is coded by a symbol either as an absolute tone, defined globally with respect to the speakers pitch range or as a ....

Hirst, D., Di Cristo, A., Espesser, R. (forthcoming). Levels of representation and levels of analysis for the description of intonation systems. In Horne, M. (Ed.), Prosody: Theory and Experiment, Dordrecht: Kluwer Academic Publishers.

No context found.

Hirst, D., Di Cristo, A., Espesser, R. Levels of representation and levels of analysis for the description of intonation systems. In Horne, M. (Ed.), Prosody: Theory and Experiment, Dordrecht: Kluwer Academic Publishers. 2000 96

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC