378 research outputs found
PEC904510 Supplemental Material - Supplemental material for Contextualising Smiles: Is Perception of Smile Genuineness Influenced by Situation and Culture?
No full textSupplemental material, PEC904510 Supplemental Material for Contextualising Smiles: Is Perception of Smile Genuineness Influenced by Situation and Culture? by Phoebe H. C. Mui, Yangfan Gan, Martijn B. Goudbeek and Marc G. J. Swerts in Perception</p
Contrastive accent in dialogue : towards a presuppositional analysis
No full textEmiel Krahmer and Marc Swerts IPO, Center for Research on User-System Interaction, Eindhoven University of Technology fe.j.krahmer/[email protected] 1 PROLOGUE What is the meaning of contrastive accents in dialogue? To answer this question, a number of hurdles have to be taken. To begin with, there is still no consensus in the prosodic literature whether a separately identifiable contrastive intonation exists in the first place. So, before we can even begin to answer the initial question, we have to try and answer a different question: do dialogue participants distinguish contrastive accents from other kinds of accents? While there are various answers to the first question (e.g., Rooth 1992, Hendriks and Dekker 1995, van Deemter 1999, Piwek 1998), this paper is to the best of our knowledge the first where both questions are addressed in combination
Backchannels and Personality Perception
No full textThis dataset contains all data and scripts used for Blomsma, P., Skantze, G., & Swerts, M. (2022). Backchannel behavior influences the perceived personality of human and artificial communication partners. Frontiers in Artificial Intelligence, 42.
This includes:
1. The avatar stimuli (videos)
2. Experiment results
3. Scripts for analysis of experiment results
Tangram O-Cam Experiment Annotations and Analyses
No full textThis dataset contains (1) annotations (encodings) of the video recordings of 14 participants that participated in an o-cam experiment (and the stimulus), as reported in:
Brugel, M. (2014).Het effect van de eye gaze en lach van de spreker op het uitlokken van feedback bij deontvanger. Master’s thesis, Tilburg University. The encodings include Backchannel Opportunity Point annotations (BOPs), head movement and facial expression annotations (facereader encodings), and vocal annotations (ELAN encodings).
and (2) the scripts to get the annotations into the R environment and analyse the annotations as reported in
Blomsma, P., Vaitonyte, J., Skantze, G., and Swerts, M. (2022). Variability between and within addressees in how they produce audiovisual backchannels. [Manuscript submitted for publication
An Extended Pose-Invariant Lipreading System
No full textIn recent work, we have concentrated on the problem of lipreading\ud
from non-frontal views (poses). In particular, we have focused on\ud
the use of profile views, and proposed two approaches for\ud
lipreading on basis of visual features extracted from such views:\ud
(a) Direct statistical modeling of the features, namely use of\ud
view-dependent statistical models; and (b) Normalization of such\ud
features by their projection onto the ``space'' of frontal-view\ud
visual features, which allows employing one set of statistical\ud
models for all available views. The latter approach has been\ud
considered for two only poses (frontal and profile views), and\ud
for visual features of a specific dimensionality. In this paper,\ud
we further extend this work, by investigating its applicability\ud
to the case where data from three views are available (frontal,\ud
left- and right-profile). In addition, we examine the effect of\ud
visual feature dimensionality on the pose-normalization approach.\ud
Our experiments demonstrate that results generalize well to three\ud
views, but also that feature dimensionality is crucial to the\ud
effectiveness of the approach. In particular, feature\ud
dimensionality larger than 30\ud
is detrimental to multi-pose visual speech recognition performance
The communicative import of gestures: evidence from a comparative analysis of human-human and human-machine interactions
No full textBackchannels and Personality Perception
No full textThis dataset contains all data and scripts used for Blomsma, P., Skantze, G., & Swerts, M. (2022). Backchannel behavior influences the perceived personality of human and artificial communication partners. Frontiers in Artificial Intelligence, 42. This includes: 1. The avatar stimuli (videos) 2. Experiment results 3. Scripts for analysis of experiment results
Factors affecting schwa-insertion in final consonant clusters in Standard Dutch Marc Swerts, Hanne Kloots, Steven Gillis, Georges De Schutter
No full textThe current paper reports on a study which fits in a larger investigation on variability in the pronunciation of Standard Dutch, as spoken in the Netherlands and the Flemish part of Belgium. In particular, it deals with the factors that determine the possible insertion of a schwa in specific consonant clusters at the end of words. Our study reveals that the absence or occurrence of such an extra vowel is dependent on regional, social and phonotactic determinants. We discuss how this finding, in combination with other results on speech variability, is important for speech technological applications, both for synthesis and recognition
Prosodic features at discourse boundaries of different strength
No full textThis paper presents the design and the evaluation of a method to study prosodic features of discourse structure in unrestricted spontaneous speech. Past work has indicated that one of the major difficulties that discourse prosody analysts have to overcome is finding an independent specification of hierarchical discourse structure as to avoid circularity. Previous studies have tried to solve this problem by constraining the discourse or by basing segmentations on a specific discourse theory. The current investigation first explores the possibility of experimentally determining discourse boundaries in unrestricted speech. In a next stage, it is investigated to what extent boundaries obtained in this way correlate with specific prosodic variables: the features pause, pitch range, and type of boundary tone are studied as a function of discourse structure
- …
