ABSTRACT
Dutch listeners rated the perceived prominence of the second Fo peak in a number of artificial two-peak intonation contours. When the two peaks were contained within a single intonation phrase, as signalled by the shape of the contour between the two peaks, and the pitch range of the contour was neutral, the perceived prominence of the second peak appeared to be determined by the Fo of the first. However, when the two peaks occurred in separate intonation phrases, or when the pitch range was wide, no such effect was found. The results suggest that the prominence of peaks is judged as a function of the prominence of the contour of its intonation phrase, at least in contours with normal pitch range. These findigs confirm the effect of pitch range found in the literature for English, and extend them by their demonstration that the effect is chiefly to be attributed to the phonological shape of the contour, and only secondarily to its pitch range.
ABSTRACT
The Grammar of Dutch Intonation (`GDI', [1]), the model I adopted to describe intonational phenomena, provides an inventory of accent-lending and boundary-marking pitch configurations for Dutch, but little is known about the factors influencing the choice within these two categories. The present study aims to provide some insight into this issue, by way of experimentally testing abstract linguistic propositions regarding the meaning of a number of accent-lending intonation patterns in Dutch. Two perception experiments have been carried out to test four form-meaning hypotheses and the results confirm the basic correctness of the semantic proposals.
ABSTRACT
In general, most of the developed prosody and intonation models were obtained from a statistical analysis of F0 curves and resynthesis by TTS. But there is yet another chance improving quality and naturalness: effective results can also be obtained by analysing the listeners' common sense about natural intonational behavior. Therefore, we use a digital process that generates signals representing only the melody of the original speech signal. Comprehensive listening experiments become possible to analyse and compare the perception of natural and synthetic intonation. Based on the results of some listening experiments a statistical analysis of the F0 curves was carried out, regarding that a speaker-individual intonation model needs more quantitative F0 information than traditional descriptions. The aim is an prosodical speaker-dependent model for synthetic speech and dialog systems. Furthermore, this flexible approach should not be limited to speaker-individual intonation.
ABSTRACT
In previous works, we proposed, on the basis of acoustic analysis and synthesis, that intonation can be cognitively and linguistically described by a lexicon of prosodic contours. The aim of the present work is to validate on the French language such an approach in perception processing. Two experiments are described hereafter. The first one consists in evaluating on complete utterances the identification of the six attitudes currently implemented in the ICP TTS system. The second one is a gating experiment which aims at showing a stable and early identification of attitudes. Perception results are commented and compared to the acoustic forms to analyse the relation between the perceptive and prosodic points of unicity if any.
ABSTRACT
Two experiments were designed to investigate the perceptual strength of an F0-rise relative to other possible local and global cues to focus in Swedish. The contribution of F0 relative to other possible local cues was investigated by manipulating the F0-contour in naturally produced Swedish sentences. Manipulations involved a gradual reduction of the F0-rise in focused words and a gradual addition of an F0-rise in non-focused words. The possible influence of global cues was explored by varying the amount of global information. In the first experiment, subjects were presented with complete sentences. In the second experiment, they heard sentences with the last part excluded. The results indicate that the F0-rise is neither necessary nor sufficient to perceive focus; other cues, both local and global, appear also to play a role.
ABSTRACT
This paper presents a theoretical framework for a model of perceived accentuation categories. This framework is based on the combined results of a series of experiments on accentuation boundaries in Dutch, French and Swedish and on theoretical work on tonal perception in speech. We propose a model in which several different language-dependent categories of accentuation are represented as the falling or rising pitch movement is advanced through the syllable. The perceived category depends upon whether an onset of the pitch movement or a pitch jump is perceived, whether or not the particular category is represented in the language in question, and whether or not the movement also serves as a cue for phrasing. Dutch and Swedish display similarities in accentuation categories while French differs from these two languages. These differences are explained by the conflict between cues for accentuation and phrasing and the differing intonational structure of French. The proposed perceptual categories have general implications for the understanding and description of accentuation.