Abstracts
of Lectures and Short Contributions:
Click
here to download this file
Age as a Disguise in a Voice Identification
Task
Ruth Huntley Bahr
Dept. of Communication
Sciences and Disorders, University of South, Florida, Tampa, USA
ABSTRACT:
Listeners have been shown
to be quite reliable in identifying speaker age; however, little is known about
how speakers may use age as a voice disguise. Previous research has shown that
fundamental frequency and speech rate may be powerful predictors of speaker
age. Yet, it is not known if talkers manipulate these parameters when
attempting to produce an older voice or if they will rely on voice stereotypes
to guide their productions. The following investigation is designed to study
how manipulation of speaker age might affect speaker identification.
Four young actors (2 males and 2
females) provided voice samples that represented young, middle-aged, and older
voices. Four healthy elders also provided voice samples. In the first
experiment, naïve listeners were asked to estimate talker age of both the young
adult simulations of vocal age and the “real age” samples. In the second task,
listeners were asked to determine if the presented voice pairs were from the
same or different speakers.
Results indicated that talkers were
able to produce voices that were perceived as different from their
chronological age. However, these vocal disguises were not perceived as being
representative of actual older voices. In terms of disguise effectiveness, the
listeners did not confuse the age-disguised voice for another individual, but
these same auditors had much difficulty matching an age-disguised voice to its
owner. These results will be discussed in terms of aging stereotypes and cues
to speaker identity. The influence of these findings on the speaker
identification process will be described.
Virtual talking heads and ambiant face-toface
Communication
Gérard BAILLY, Frédéric ELISEI and
Stephan RAIDT
Institut de la Communication Parlée, 46 av. Félix
Viallet, 38031 Grenoble - France
ABSTRACT:
We describe here
our first effort for developing a virtual talking head able to engage a
situated face-to-face interaction with a human partner. This paper concentrates
on the low-level components of this interaction loop and the cognitive impact
of the implementation of mutual attention and multimodal deixis on the
communication task.
Detection of Faces and Recognition of Facial Expressions
Nikolaos Bourbakis
ITRI, Wright State
University, Dayton, Ohio, USA
Abstract
Face detection
is the foremost task in building vision-based human-computer interaction
systems and in particular in applications such as face recognition, face
identification, face tracking, expression recognition and content based image
retrieval. A robust face detection
system must be able to detect faces irrespective of illuminations, shadows,
cluttered backgrounds, facial pose, orientation and facial expressions. Many approaches for face detection have been
proposed. However, as revealed by FRVT
2002 tests, face detection in outdoor images with uncontrolled illumination and
in images with varied pose (non-frontal profile views) is still a serious
problem. In this talk, we describe a
Local-Global Graph (LGG) based method for detecting faces and for recognizing
facial expressions accurately in real world image capturing conditions both
indoor and outdoor, and with a variety of illuminations (shadows, high-lights,
non-white lights) and in cluttered backgrounds. The LG Graph embeds both the local information (the shape of
facial feature is stored within the local graph at each node) and the global
information (the topology of the face). The LGG approach for detecting faces
with maximum confidence from skin segmented images is described. The LGG approach presented here emulates the
human visual perception for face detection.
In general, humans first extract the most important facial features such
as eyes, nose, mouth, etc. and then inter-relate them for face and facial
expression representations. Facial
expression recognition from the detected face images is obtained by comparing
the LG Expression Graphs with the existing the Expression models present in the
LGG database. The methodology is
accurate for the expression models present in the database.
Image Chromatic
Adaptation for Face Skin Color Detection
Nikolaos Bourbakis and
Praveen Kakumanu
ITRI, Wright State
University, Dayton, Ohio, USA
Abstract
The
goal of image chromatic adaptation is to remove the effect of illumination and
to obtain color data that reflects precisely the physical contents of the
scene. We present in this talk an
approach to image chromatic adaptation using neural networks (NN) with
application for detecting - adapting human skin color. The network is trained on randomly chosen
color images containing human subject under various illuminating conditions,
thereby enabling the model to dynamically adapt to the changing illumination
conditions. The proposed network
predicts directly the illuminant estimate in the image so as to adapt to the
human skin color. The comparison of our
method with Gray World, White Patch and Neural Network on White Patch
algorithms is presented. We also
present our results on detecting skin regions in NN color corrected test
images. The results are promising and
suggest a new approach for adapting human skin color using NN’s. The skin
detect technique presented here is the first part of an integrated
methodology-tool used for detecting human face and facial expressions of
emotion.
Nonverbal Communication
as a Factor in Linguistic and Cultural Miscommunication
Maja Bratanić,
Professor, University of Zagreb, Croatia
ABSTRACT:
The
presentation discusses two major assumptions:
-
that a great deal of human communication is culturally molded
and conditioned
-
that people convey meanings not only through language but
through various aspects of nonverbal communication as well.
Nonverbal
behavior is to a great extent universal but in many ways also marked by
culture-specific patterns. Being less obvious than misunderstandings in verbal
communication, nonverbally induced miscommunication is far more difficult to
detect. Furthermore, the line between verbal and nonverbal components of
communication is often hard to delineate precisely.
Main
categories of nonverbal behavior and its role in communication will be briefly
discussed with the focus on proxemics - the study of the human use of space
within the context of culture. The concept of proxemics and its implications
will be elaborated on examples from American cultural patterns.
Further examples of
culturally-conditioned miscommunication will draw on an aviation-related
context.
The
presentation will be accompanied by a video A
World of Differences: Understanding Cross-Cultural Communication by D.
Archer.
Face recognition from 2D still images
Paola Campadelli
Università
di Milano, Italy, campadelli@dsi.unimi.it
Abstract:
In
the past two decades a lot of research work
has been devoted to the development of automatic methods aimed at recognizing
people from images;
such
systems are attractive since this type of identification does not require any
interaction with the subject. However, the problem is very difficult especially
when very few assumpions are done on the images to be treated.
In
this talk the most interesting methods developed for face recognition from
still images will be presented and compared. Open problems will be dealt with, and
the contribution that 3D information might provide will be discussed.
Technology for Non-Verbal
Speech Processing
Nick
Campbell,
ATR Science Labs, Kyoto,
Japan, nick@atr.jp;
Abstract:
This
pair of lectures will focus on the technological needs for the processing of
non-verbal speech in a dialogue context.
It is based on an analysis of a very large corpus of spoken interactions
captured under extremely natural situations.
The talks will present a model of speech interaction as not only
facilitating the exchange of linguistic or propositional information, but also
facilitating the display of affect and interpersonal or social relationships.
Part I: Speech Synthesis
and Discourse Information
This
talk presents some recent work towards a conversational speech synthesis system
for use in interactive dialogues, such as might take place between a person and
an information system, a robot, or a speech translation device. The talk describes several types of response
utterances that are currently very difficult to implement using traditional
speech synthesis methods, and shows how these non-verbal speech sounds function
to provide feedback and status-updates in an interactive discourse. The lecture will be illustrated with
examples of such practic utterances, including laughter and grunts as well as
common phrases and idiom, showing how their variety can reveal several types of
information about the speaker- (i.e., listener) states. The proposed model of information exchange
through non-verbal speech shows how this feedback from the listener can help
the speaker to deliver content more efficiently, and at the same time to be
reassured of success in information transmission.
Part II: Towards Recognising Speech Gestures in Discourse
This
talk describes how the lowest level of information can be processed in a speech
signal for annotation of discourse progress and speaker participation
status. In a semi-formal round-table
meeting situation there is typically only one main speaker at any given moment,
but several participants may be speaking simultaneously, expressing
(dis-)agreement, chatting, translating, etc., in addition to the main
speaker. We are currently performing
research into technology to process this audio landscape in order to detect the
main speaker and to categorise the competing forms of speech. Several speech gestures such as laughter,
agreement, and feedback-responses can be recognised, isolated, and used to
determine the progress of the meeting and the degrees and types of
participation status among the members present. This talk will describe the current state of the technology and
will present examples of the frequent gestures with descriptions of their
typical usage.
The Amount of Information on Emotional
States Conveyed by the Verbal and Nonverbal Channels: Some Perceptual Data
Anna
Esposito
Dipartimento
di Psicologia, Seconda Università di Napoli, and IIASS Italy
ABSTRACT:
In
a face-to-face interaction, the addressee exploits both the verbal and nonverbal
communication modes to infer the speaker’s emotional state. Is such an
informational content redundant? Is the amount of information conveyed by each
communication mode the same or is it different? How much information about the
speaker’s emotional state is conveyed by each mode and is there a preferential
communication mode for a given emotional state? This work attempts to give an
answer to the above questions evaluating the subjective perception of emotional
states in the single (either visual or auditory channel) and the combined
channels (visual and auditory). Results show that vocal expressions bring the
same amount of information as the combined channels and that the video alone
brings poorer emotional information than the audio and the audio and video
together. Interpretations of these results (that seem to not support the data
reported in the literature proving the dominance of the visual channel in the
emotion’s perception) are given in terms of cognitive load, language expertise
and dynamicity. Also, a mathematical model inspired to the information
processing theory is hypothesized to support the suggested interpretations.
Analyzing
and modelling verbal and non-verbal communication for talking animated
interface agents
David
House and Björn Granström
Royal
Institute of Technology, Sweden
ABSTRACT:
The
use of animated talking agents is a novel feature of many multimodal spoken
dialogue systems. The addition and integration of a virtual talking head has direct
implications for the way in which users approach and interact with such
systems. However, understanding the interactions between visual expressions,
dialogue functions and the acoustics of the corresponding speech presents a
substantial challenge. Some of the visual articulation is for obvious reasons
closely related to the speech acoustics (e.g. movements of the lips and jaw),
while there are other articulatory movements affecting speech acoustics that
are not visible on the outside of the face.
On the other hand, many facial gestures used for communicative purposes
do not affect the acoustics directly, but might nevertheless be connected on a
higher communicative level in which the timing of the gestures could play an
important role. The context of much of our research regarding these questions
is to be able to create an animated talking agent capable of displaying
realistic communicative behaviour and suitable for use in conversational spoken
language systems.
The
focus of these lectures is to look into the communicative function of the
agent, both the capability to increase intelligibility of the spoken
interaction and the possibility to make the flow of the dialogue smoother,
through different kinds of communicative gestures, such as visual prosodic
gestures (e.g. focal accent and emphatic stress) and gestures for different
expressive states, turntaking and negative or positive system feedback. We will
give some examples of recent work, primarily at KTH, involving the collection
and analysis of databases for audiovisual prosody. We will report on methods
for the acquisition and modelling of visual and acoustic data, and provide some
examples of analysis of e.g. head nods and eyebrow settings related to
communicative functions. We will also demonstrate how this analysis can be
implemented to generate useful and realistic expressive audiovisual synthesis
using a combination of data-driven and rule-based methods.
Individual Speech Rhythm Variation Within
the Plosive Structure of Speech
Eric
Keller, IMM, University of Lausanne, Switzerland
Abstract:
The
"perceived humanness" of speech revolves centrally around the issues
of regularity and variation. Within an utterance's temporal structure, we
subjectively experience humans to speak with a certain regularity -- which
creates perceived rhythm within speech -- at the same time as we expect them to
display variation, mostly for emphasis and to satisfy personal preferences.
Synthesized speech that does not exhibit these perceptual qualities is often
classified as "robotic" and "unnatural".
The search for the objective bases of the perceived regularity in speech is old
and has produced less than satisfactory results. In fact in 1977, Ilse Lehiste, in an extensive review of the
issue of isochrony (acoustic evidence
for rhythmicity in speech) came to the conclusion that there were no direct acoustic correlates of
rhythmicity, a view that has formed the consensus for spontaneously produced
speech since then, despite a number of further studies performed on the issue.
However, we have data to show that
regularity may actually be directly dependent on what might be called the
"plosive structure" of the speech chain. If one considers vowel
onsets in terms of the suddenness and the relative strength of voice onset, it
turns out that human speakers exhibit considerable inter-speaker agreement with
respect to the placement of sudden ("strong") vowel onsets, but that
this inter-speaker agreement is
gradually reduced as vowel onsets "weaken". The
"strength" or "weakness"
of the vowel onset can be determined automatically from the acoustic signal, and is thus likely to
correlate with both motor and
perceptual saliency within the utterance. "Strong" vowel
onsets (i.e., those that resemble plosive sounds) appear to set a
"frame" for speaking, and between those onsets, we are much freer to
choose the timing values appropriate to
the specific semantic and personal context.
Current statistical or neural network
models for temporal structuring of
speech may thus well be flawed. Currently, we model all aspects of speech timing with rigid, totally
predictable statistical structures.
Instead, temporal prediction systems should probably provide a set
of main "temporal anchor
points" within the utterance, and introduce "motivated variation" for the remaining aspects of
temporal structure. We will pass in
review the different types of psycholinguistic and pragmatic events that can motivate such variation, and we will
consider a prediction system that can
handle these new requirements.
TWO LECTURES ON
GESTURE
ADAM KENDON
Abstract:
We
begin with the question: what is ‘gesture’?
Can we identify, in a theoretically coherent manner, a domain of human
action to be called ‘gesture’ which is to be distinguished from other forms of
human bodily expression? After
answering this question, we shall proceed to look at the problem of how meaning
is attributed to gestural actions. Then we shall look at gesture in relation to
speech and consider the different ways it may be employed whenever it is used
in conjucntion with speech within an utterance. Examples will be presented that
show that speaker’s appear to employ speech and gesture as partners in a common
enterprise of utterance construction. Gesture, it will be shown, must be seen
to be as much the ‘final product’ of a speaker’s utterance as are the speaker’s
words. Discussion will then turn to a consideration of how gesture and speech
interact semantically as these two modalities create unified expressions. It
will be shown that the ways in which gestures contribute to the overall meaning
of an utterance are quite diverse. No simple generalizations are possible. Here
we shall first consider the contributions gestures may make to the
propositional content of utterances. This will be followed by a discussion of
the ‘meta-discursive’ or pragmatic functions that gestures also frequently can
be shown to have. Throughout these
lectures numerous examples will be presented, drawn from video-recordings of
conversations in many different circumstances in southern Italy (especially in
the provinces of Naples and Salerno) and in central England.
Recommended
reading: Adam Kendon Gesture: Visisble Action as Utterance. Cambridge
University Press, 2004. Especially Chapters 7-13.
A Methodological
and Theoretical Framework for the Study of Psychological and Biometrical
Characteristics in Verbal and Nonverbal Communication
Dominic W. Massaro, Ph.D
Perceptual Science
Laboratory, Department of Psychology, University of California
Santa
Cruz, CA 95060 U.S.A, 1-831-459-2330, FAX 1-831-459-3519, massaro@fuzzy.ucsc.edu, http://mambo.ucsc.edu/psl/dwm/
Abstract:
The
goal of the lectures is to provide an empirical and theoretical overview of a
paradigm for inquiry on the Psychological and Biometrical Characteristics in
Verbal and Nonverbal Communication. A persistent theme of our approach is that
humans are influenced by many different sources of information, including
so-called bottom-up and top-down sources. Understanding spoken language, for
example, is constrained by a variety of auditory, visual, and gestural cues, as
well as lexical, semantic, syntactic, and pragmatic constraints. We will first
present a theoretical framework for language processing, and the methodological
implications of this framework. We will then review experimental evidence in
support of this framework while inconsistent with other frameworks. We will
indicate some limitations in other current studies, and suggest changes for
improvement. Finally, we will extend the analyses to cross-linguistic studies
of speech perception, as well to studies of emotion.
Research
questions for psycholinguists and speech and reading scientists include the
nature of the sources of information; how each source is evaluated and
represented; how the multiple sources are treated; whether or not the sources
are integrated; the nature of the integration process; how decisions are made;
and the time course of processing. Research in a variety of domains and tasks
supports the conclusions (for summary see Massaro, 1998) that a) perceivers
have continuous rather than categorical information from each of these sources;
b) each source is evaluated with respect to the degree of support for each
meaningful alternative; c) each source is treated independently of other
sources; d) the sources are integrated to give an overall degree of support for
each alternative; e) decisions are made with respect to the relative goodness
of match among the viable alternatives; f) evaluation; integration; and
decision are necessarily successive but overlapping stages of processing; and
g) cross-talk among the sources of information is minimal. The fuzzy logical
model of perception (Massaro, 1998; FLMP) will be described and contrasted with
other models of speech and emotion processing. These models will be tested
against speech perception experiments involving cross-linguistic comparisons
and experiments on emotion and gesture perception.
Abstract
of McNeill lectures
David
McNeill University, Chicago, USA, dmcneill@uchicago.edu
1.
Introduction to gesture study – how and why it is done.
2.
Some basic facts of speech-synchronized gestures.
3.
The growth point, context, catchment, imagery-language
dialectic, dynamic and static dimensions and how they relate.
4.
Gesture in social interaction.
5.
Gesture and culture, linked in non-obvious ways.
6.
Gesture and brain, relevance to the origin of language.
7.
Multimodal
expressive ECAs
Catherine Pelachaud,
Universite
de Paris 8, France, c.pelachaud@iut.univ-paris8.fr;
Abstract:
Embodied Conversational Agents (ECAs) are human-like entities capable of
communicating with other ECAs and/or users. They exhibit synchronized verbal
and nonverbal behaviors (facial expression, gesture, body movement and gaze).
During these lectures we will present an ECA system architecture. We will
introduce the taxonomy of communicative
functions developed by
Isabella Poggi. Following this taxonomy each communicative function is
represented as a pair where the first element corresponds to the meaning of the
communicative function while the second element is a description of the signal
that is used to transmit this meaning. A representation language, Affective
Presentation Markup Language, APML has been elaborated. It is used to drive the
animation of the agent and ensures synchrony between the multimodal signals.
Behaviors are defined not only by the signals that composed them but also by
how they are displayed. We will present an expessivity model where six
parameters have been designed to change the gesture and face expressivity.
Intonation, Accent and Personal Traits
Michelina Savino
Dept. of Psychology,
University of Bari, ITALY
Abstract:
Speaking with an
accent reveals the sociolinguistic background of locutors, and it is a widely
shared belief that intonation plays a crucial role in characterising regional
varieties of spoken languages. This is attested by the large amount of
descriptive studies in the literature, and also from the perceptual point of
view a number of experiments have been carried out to test the hypothesis that
language varieties can be identified by pitch information alone (even though it
is currently not clear to what extent segmental information can also play a
role in such identification task).
An exemplar case
is represented by Italian speakers, for whom the spoken language represents a
reliable way for identifying their sociolinguistic traits in verbal
interactions, as they always speak with an accent, in both formal and informal
situations. This is a consequence of the particular status of Italian with
respect to other languages: for historical reasons, in fact, the process of
standardisation has been successfully achieved for the written form but not for
the spoken language, which is presently characterised by quite strong regional
accents. In fact, standard Italian has never been taught in any level of the
Italian education system, and its use has been restricted to a small number of
professional speakers and actors, and therefore in very specific contexts.
In this lecture,
a discussion on the role of intonation in conveying information on the
speakers’ sociolinguistic background as personal traits will be presented,
basing mainly on examples of Italian varieties.
Blind Signal Pre-Processing for MPEG-7
based Multimedia Metadata
applications: an Assisted Living use-case
Giovanni Tummarello, Stefano Squartini, Francesco Piazza
Università
Politecnica delle Marche, Italy
ABSTRACT:
Metadata extracted from Multimedia or live sensoring is set to play a major role in any intelligent and multimodal interactions between humans and computers. Furthermore, it is generally required that such metadata are structured and encoded according to well agreed standards. This is fundamental to enable interoperability and create complex applications as a mesh of heterogeneous services and components. On purpose, the MPEG-7 standard for dealing with multimedia metadata and the tools developed within the Semantic Web initiative are providing today the basic framework. Their application to real world problems, however, is made problematic by the fact that the data are often captured from difficult live conditions. It is therefore of primary importance to enhance the quality of the observable signals before the metadata extraction algorithms are employed. In particular, for the case of audio signals, it is important to perform separation and deconvolution of audio signals captured in real environments and in blind conditions. In this work a full featured real world multimedia metadata assisted living scenario is constructed using a combination of Blind Signal Processing and MPEG-7 based metadata techniques. In such example, an array of microphones captures speech and audio signals and thanks to MPEG-7 technologies the user can select multimedia content to be played.
Visual Phrasems and Pragmaphrasems in
English, Polish and Croatian
Neda Pintaric
University
of Zagreb, Croatia
ABSTRACT:
The author
writes about etimological and semantic meaning of an eye as the organ with the
biggest capacity among all human organs. Eyes, ears and tactil receptors are
main receptors in human communication. The author claims that nonverbal code is
a pracode based on these receptors, therefore it has been developed earlier
than the verbal code.
Nonverbal code is multicode
consisting of kinetic, tacezic, deiktic, proxemic and prosodic signs which we
use consciously and unconsciously. Our unconscious informations couldn't be
hidden and the other person in communication can read it in our eyes.
In various cultures people operate
with visual culturemes, such as a custom of eye-contact which can mean
sincerity (e.g. among Croats) or sexual affection (e.g. among Poles).
The main part of the paper consists
of lexical and phrasematic analyse of pragmatic items called pragmemes and
phrasopragmemes. The author compares different linguistic systems using
examples in English, Polish and Croatian eye-signs.
Videocaptured Verbal and Nonverbal Foreign
Language Teacher Feedback
Leticia
Vicente-Rasoamalala
Aichi Prefectural University,
JAPAN
ABSTRACT:
Videotaping
for research purposes in the field of Second Language Acquisition (SLA)
classroom context is still quite experimental and a minority practice. Apart
from the difficulties for getting the consent of the participants (i.e. the
school authorities, the teachers and the parents of the students) for
videorecording classroom interactions, there are not very defined research
lines and instruments to work with the obtained data. Specifically, the present
poster will highlight different issues concerning the collection, the
identification and the analysis of the videocaptured verbal and the nonverbal
foreign language teacher feedback in classroom context.
The
focus on this area is at identifying from the collected database the teachers’
strategies that appear to be more successful in dealing with L2 learner oral
output containing deviant forms. From the last decade, a number of SLA studies
influenced by Long’s Interaction
Hypothesis (1996) are studying in detail the negative implicit forms of
teacher feedback under the assumption that they might be more beneficial for
acquisition. Such works have adopted diverse qualitative and quantitative
paradigms. For instance, elements of ethnographic research, Conversational
Analysis and the Neo-Vygotskian perspective (Vygotsky, 1968). The ultimate goal
of many studies in this area is finding out the instructional sequences that
might optimize foreign language teaching and learning. Nevertheless, there are
some shortcomings relating to the lack of consensus for the existing schemes
analyzing corpora and the types of verbal and non verbal annotations.
Originally, most approaches in classroom discourse research
have built analytic frameworks almost exclusively dealing with audiotaped linguistic
data designed
to build didactic models. Significantly, most instruments have neglected
the audio-visual data
which might be captured in videorecordings.
Thanks to the
new technologies the notations of teacher gestures and the manipulation of tools are being incorporated in some classroom studies.
Some works have suggested that non verbal elements might enhance verbal
feedback and often regulate classroom exchanges among teachers and learners.
Additionally, those outcomes suggest that using extensively video recordings is
necessary for future comprehensive studies of classroom
discourse.
1.
Block, D (2003) The Social Turn in
Second Language Acquisition. Edinburgh University Press.
2.
Gardner, R. & Wagner, J. (Eds.)
(2004) Second Language Conversations. London: Continuum.
3.
Long, M. H. (1996). The role of the
linguistic environment in second language acquisition. In Bhatia and Richie,
(Eds.), Handbook of second language acquisition, San Diego: Academic Press,
Inc.
4.
Lyster, R. (2001) Negotiation of
form, recasts and explicit correction in relation to error types and learner
repair in immersion classrooms. Language Learning, 51, 265-301.
5.
Vicente-Rasoamalala, L. (2006)
Elementos No Verbales en la Retroalimentación del Docente de L2”. The Journal of the Faculty of Foreign Studies. Aichi Prefectural
University 38, 159-188.
Effectiveness of Short-Term Prosodic
Features for Speaker Verification
Iker Luengo, Eva
Navas, Inmaculada Hernáez
University of
the Basque Country, Spain, ikerl@bips.bi.ehu.es
Abstract:
In this work a traditional MFCC based system is combined with a
prosody based one to determine whether simple short-term prosodic information
is useful for improving current state-of-the-art ASV. Results do not show
significant improvement
A Partially Observable Markov Decision
Process approach to Affective Dialogue Modeling
Bui Huu Trung,
Human Media Interaction, Department
of Computer Scence, Faculty of Electrical Engineering Mathematics and Computer
Science, University of Twente, P.O. Box 217 7500AE Enschede the Netherlands
Abstract:
We
propose a novel approach to developing a dialogue model which is able to take
into account some aspects of the user's emotional state and acts appropriately.
The dialogue model uses a Partially Observable Markov Decision Process approach
with observations composed of the observed user's emotional state and action. A
simple example of route navigation is explained to clarify our approach and
preliminary results & future plans are briefly discussed.
A Systemic Approach to Enhance Writing,
Analysis, and Presentation Skills
Aly N. El-Bahrawy,
Professor
Faculty of Engineering,
Ain Shams University, Cairo, Egypt
Abstract:
The
paper discusses the systemic of technical communication - for researchers and
professionals - which includes writing, analysis and presentation. Each of the three components has sub-components
related to conveying the technical message clearly to the receiver. The first component ‘Technical writing’ is
related to language rules in general, and technical writing guidelines in
particular. The second component ‘Data
analysis’ is related to statistical and database principles in addition to
graphics basics. The third component
‘Professional Presentation’ is related to organization of material,
audio-visual equipment, the body movement of the speaker, which includes voice,
hand gestures, facial expressions, etc.
Another encompassing factor is the use of computers to help the three
components send the respective message clearly. As an example, the Microsoft programs WORD, EXCEL, and PowerPoint
are powerful tools to write, analyze and present the technical message. In WORD, features like formatting, language
tools, and automatic generation of reference tables are very appealing. In EXCEL, data analysis and graphics are
very elaborate. In PowerPoint,
organization and animation tools can be used to enhance the presentation
significantly. The paper presents
examples of the use of such approach to execute successful training courses for
engineers and researchers. Finally, the
paper stresses the importance of the three components to form the technical and
academic character of professionals.
Can
Word prime gestures?
Paolo Bernardis
Università di Bologna, Scuola
superiore di Studi Umanistici. Università di Parma, Dipartimento di
Neuroscienze
Abstract:
Can words prime gestures?
Evidence for language and
action relationships was recently highlighted in both behavioral (Glenberg,
2002) and neurophysiological research (Gentilucci, 2001), thus reinforcing the
already well established link between language and gestures for communicative
purposes (Bellugi,1979; 1987; McNeill,1992; 2000). However, at present there is
no clear evidence of the direct interaction of the two systems.
Aim of this study was
precisely to check for evidence of this interaction with the priming effect,
i.e. to investigate whether words could prime the recognition of a gesture with
the same meaning. The participants were presented with a word and then asked to
recognize a gesture either having the same or a different meaning. The words
chosen as primes were of different kinds: 1) a noun referring to a simple
object with (1a) no specific action required (e.g., ‘clock’) or (1b) a specific
action required (e.g., ‘gun’ -to shoot); 2) a verb referring to (2a) a direct
simple action (e.g., ‘knock’) or (2b) an action to be performed with a tool
(e.g., ‘write’). The gestures presented were video-clips showing the upper
half-body of an actor, which was blurred, performing the gestures with his arms
and hands. Response times and errors were recorded.
Two
main results were obtained. The first was a clear priming effect of meaning
(i.e., a semantic priming effect of word on gesture). The second result showed
a greater priming effect when both words and gestures referred to objects and
actions requiring tools.
The study
of similiarites in learning foreign languages
Daboveanu
Diana-Cristina, Niculae Carmen – Eufrusina
Romania, Bucharest, Sector 3, 37, Mircea-Voda
Boulevard, Block M29, Scara D, 3rd Floor, Flat 113,
Abstract:
Today
student mobility has already become reality. It is supported by numerous
education programmes at national and international levels and in particular
within the framework of EU funded programmes. There is, however, a clear need
for support programmes to assist exchange students to prepare for their
studies. The EUROMOBIL project (72139-CP-2-2000-1-FI-L2) the development of a
multimedia language learning and information programme on CD-ROM for DE, EN, HU
and FI) was started in 1999 with the aim of developing a self-study course,
which would enable exchange students to prepare, both in terms of the host language and knowledge of the culture,
for their visit to universities in DE, UK, HU and FI. The project was expanded
to CZ, FR, PL, PT and RO.
In
a needs analysis at the beginning of the project differences in the
requirements for studies abroad were noted. Starting from this need analysis
and using rank distance as an measure for similarity our goal was to research
how related are the problems that apear when studying a foreign language and to
expand this result to see the differences and similarities between the cultures
and languges that are part of the project.
Tongue
motor cortex excitability is modulated by the observation of the type of grasp
action
Dalla
Volta R¹, Bernardis P¹, Buonocore A¹, Sato M¹,
Palumbo D¹, Gentilucci M¹
¹Dept.
Neuroscience, University of Parma, Italy, gentiluc@unipr.it
Abstract:
Voice spectrum
and lip kinematics during pronunciation of syllables are affected by the
simultaneous execution or observation of transitive actions, like the grasp of
different objects, according to the type of involved hand grip. The study aimed
to verify whether in humans the observation of hand grasping actions onto
objects requiring different types of grip affects tongue motor cortex
excitability. We recorded motor evoked potentials (MEPs) from the tongue of 16
right handed healthy subjects after delivering single pulses by using
transcranial magnetic stimulation (TMS) over the tongue left motor cortex.
While stimulating the participants looked at the PC monitor where video-clips
showing either hand grasping of fruits of different size that required
different types of grip or the same fruits alone were presented. The syllable
DA appeared on the fruits in both cases. Power grip of large fruits was linked to
MEPs significantly greater than those linked to precision grip of small fruits.
In a control experiment we presented either different tools approaching
geometric solids or the same solids alone. Neither when presenting the same
fruits alone nor in the control experiment any MEP modulation was observed. We
conclude that observation of different biological hand actions onto edible
objects specifically modulates tongue motor cortex excitability. These data
support the hypothesis that a motor resonance system is activated by hand
action observation. This resonant circuit sends double motor commands to both
hand and mouth.
Machine
Translation Evaluation: a Case study of Croatian-English and Russian-English MT
Systems
Ivana
Simeon
Department of Linguistics Faculty of Philosophy,
University of Zagreb, Ivana Lučića 3, HR-10000 Zagreb, Croatia,
E-mail: isimeon@ffzg.hr
Abstract:
From the earliest days of machine translation (MT), evaluation has been
an inherent and significant part of efforts invested into machine translation
research. In this paper, an overview of the history of MT evaluation is
presented, with emphasis on one of the most comprehensive MT evaluation
projects, undertaken in the 1960s, namely the Automatic Language Processing
Advisory Committee Report.
Furthermore, strategies and problems pertaining to MT
evaluation are discussed, with emphasis on the distinction between subjective
criteria, such as comprehensibility, and the objective, quantifiable criteria, such
as error quantification and analysis.
Within the
practical part of the paper, the results of testing four MT systems – one for
the language pair Croatian-English, and three for the language pair
Russian-English – are shown. The systems were tested on three textual samples
belonging to general, fictional and scientific genres. The analysis of the
results included a comprehensibility poll which included five informants
(native or proficient target language speakers) for each target language, as
well as quantification of errors and error type assessment across genres and
across individual MT systems. Finally, cumulative results are given for each MT
system and for each language pair.
As a conclusion, recent developments
in the field of MT evaluation are presented, including automatic evaluation
methods, such as IBM’s measures BLEU and NIST.
Using the Wavelet Transform in Real-time Digital Signal
Processing
Jan Vlach, Přinosil Jiří
Department of Telecommunications,
Faculty of Electrical Engineering and Communication, Brno University of
Technology, Purkynova 118, 612 00 Brno, Czech Republic,
Abstract:
The new method
of segmented wavelet transform (SegWT) makes it possible to exactly compute the
discrete-time wavelet transform of a signal segment-by-segment. This means that
the method could be utilized for wavelet-type processing of a signal in
"real time", or in case we need to process a long signal (not
necessarily in real time), but there is insufficient memory capacity for it
(for example in the signal processors). Then it is possible to process the
signal part-by-part with low memory costs by the new method. The method is
suitable for universal utilization in any place where the signal has to be processed
via modification of its wavelet coefficients (e.g. signal denoising,
compression, speech segmentation, music processing, alternative modulation
techniques for xDSL systems). It is also possible to use SegWT in
wavelet-processing (e.g. compression, selective area processing) of large
images. In the paper, the principle of the forward segmented wavelet transform
is described.
The Integrative and Structuring Function of Speech in Face-to-Face
Communication from the Perspective of Human-Centered Linguistics
Krzysztof
Korżyk
Jagiellonian
University Kraków, Poland
Abstract:
This paper illustrates the need for study of the
interdependencies between verbal and nonverbal behavior treated as a unified
form of activity, manifesting itself in face-to-face communication. Invoking
the principles of human-centered linguistics (see Yngve 1996, 2000), the author
treats verbal communication not as something passed on via language, but rather
as something to which language merely contributes. One of the consequences of
such an approach to this issue is a reassignment of focus. Rather than
attention being drawn to linguistic phenomena,
the spotlight is on the communicative
properties of the interlocutors, creatively utilizing various elements
of the interactional ”symbolic space.”
With
reference to the above, this text presents a realistic account of the
interpretational activity taking place between communicating subjects. The
action is perceived as a function of choices correlating verbal, prosodic, and
kinesthetic signs and signals. Concurrently, taking the pragmatic and
interactional aspects of these multimodal choices under consideration, the
author discusses typical situations in which the structuring role of speech is
particularly evident. Light will also be shed on the crucial interconnections
between the above-mentioned systems of signs and signals, as well as on the
advantages stemming from an integrated modeling of communicative phenomena.
1.
Yngve V.H. (1996) From Grammar to Science. New Foundations for
General Linguistics, Amsterdam: John Benjamins.
2.
V.H. Yngve and Z. Wąsik (2000) Exploring the Domain of
Human-Centered Linguistics from a Hard-Science Perspective (Workshop),
Poznań: Motivex.
Research on Speech Synthesis and Speech
Recognition of Croatian language on the Faculty of Humanities and Social
Sciences
Lazic Nikolaj,
Faculty of Humanities and
Social Sciences University of Zagreb- Ivana Lucica 3 10000 Zagreb- Croatia, nlazic@ffzg.hr;
Abstract:
Speech
synthesis and speech recognition are processes in need of multidisciplinary
approach. Faculty of Humanities and Social Sciences in Zagreb has departments
that can aid in the processes of synthesis and recognition, namely Information
sciences, Linguistics, Phonetics, Croatian language.
One
of the problems in Croatian language is accurate word accent, distinguishing
orthographically identical, but differently sounding words. Proper word
accentuation is therefore essential for accurate speech synthesis of Croatian
language. Different word accentuations may be completely wrong or
"dialectally coloured". Accurate word accent recognition in speech
recognition systems is needed for semantics in case of later machine
translation.
Different
approaches to speech synthesis may ease or complicate production of correctly
sounding synthesized speech. Speech synthesis based on concatenation needs all
variants of word accents present in the language repertoire for synthesis, but
makes synthesized speech more natural. Formant synthesis, on the other side,
produces whatever accentuation needed, but it is harder to describe all
acoustically relevant sound segments for speech synthesis.
Intercultural
Differences in Vocal Communication of Emotions: An Experimental Comparison Between
Chinese and Italian Young Adults
Fabrizia
Mantovani, Luigi Anolli, Lei Wang, Alessandro De Toni
CESCOM_Centre
for Studies in Communication Sciences University of Milan-Bicocca P.za Ateneo
Nuovo,1 20123 Milan Italy, mantovani@unimib.it;
ABSTRACT:
The poster presents an experimental study comparing the vocal
communication of emotions between Chinese and Italian young adults. Main goal
of the study is to investigate whether:
(a) the vocal
expression of eight emotions (joy, sadness, anger, fear, contempt, pride,
guilt, shame) is characterized by distinguished patterns of paralinguistic
features;
(b) the vocal
patterns of emotional expressions - produced in reaction to comparable eliciting
situations - differ between members of two different cultures (Chinese and
Italian);
(c) specific
cultural configurations exist in vocal expression of emotion.
Forty-eight undergraduates
(29 Chinese and 19 Italian) were asked to read aloud short stories inducing
different emotions via scenario approach. The short stories had been prepared
and validated in a preliminary phase: in each text a standard sentence was
included in order to carry out subsequently acoustic comparisons. Acoustic
analyses were carried out through the Computerized Speech Lab (CSL) 4300B.
Different acoustic parameters were considered referring to time (total
duration, partial duration, duration of pauses, speech rate and articulation
rate), fundamental frequency (mean, standard deviation, range, minimum and
maximum of F0) and intensity (mean, standard deviation, range, minimum and
maximum).
Results from statistical analyses confirmed the
importance of vocal production in generating distinctive emotional patterns, as
well as the presence of both similarities and differences between the vocal
emotional patterns of Chinese versus Italian participants. The theoretical
implications of these findings will be discussed.
Non-verbal Interaction and Ambient Entertainment
Anton Nijholt, Dennis Reidsma, and others
University of Twente
Department of Computer Science PO Box 217 7500 AE Enschede, The Netherlands, anijholt@cs.utwente.nl;
Abstract:
In future
Ambient Intelligence (AmI) environments we assume intelligence embedded in the
environment, its objects (furniture, mobile robots) and its virtual, sometimes
visualized agents (virtual humans). These environments support the human
inhabitants or visitors of these environments in their activities and
interactions by perceiving them through their sensors (proximity sensors,
cameras, microphones, etc.). Support can be reactive, but also and more
importantly, pro-active, anticipating the needs of the inhabitants and
visitors.
Health,
recreation, sports and playing games are among these needs. Sensors in these
environments can detect and interpret bodily activity and can give multimedia
feedback to invite, stimulate, guide and advise on bodily activity. Rather than
aiming at improving user task efficiency, in the environments we investigate
the aim is to improve physical and mental health (well-being) through exercise
and through play. Exercises can be done in order to improve fitness, to prevent
certain injuries (e.g., RSI), or to recover from an accident (e.g.,
physiotherapy exercises). Other exercises may aim at improving certain
capabilities related to a profession (ballet, etc.), some kind of recreation
(juggling, etc.), or sports (fencing, etc.). Fun, just fun, achieved from
interaction (e.g. dancing or physical gaming) can be another aim of such
environments.
In
this presentation we look at our research on bodily and gestural interaction
with environments equipped with some simple sensors (cameras, microphones,
dance pads), some application-dependent intelligence (allowing reactive and
pro-active activity), and an embodied virtual agent employed in the display of
reactive and pro-active activity. Dance, music, and associated movements in
human and virtual agents are the main modalities that are used in our
environmental installations.
Towards an all-inclusive
cross-media relations framework
Katerina Pastra
Language Technology Applications Department,
Institute for Language and Speech Processing kpastra@ilsp.gr
Abstract:
While
there is a growing demand for developing Intelligent Multimedia Interfaces and
Systems, one still strives to find a descriptive framework of how different
media and modalities interact with one another. The significance of the latter
becomes evident, when one attempts to build multimedia systems or intelligent
agents, where multimedia content integration decisions are to be made. In this
paper, we identify two important parameters in developing such a framework: the
use of multiple and clearly stated criteria for defining interaction relations
across media and the integration of findings from the analyses of the
interaction of as many different media-pairs as possible. In correlating our
own corpus-based work on image-language interaction with existing work on
image-language and gesture-language interaction, we identify three such
criteria and corresponding interaction relations. We further suggest a way of
validating the applicability and expressiveness of these interaction relations,
which involves a set of simple metrics for computing them in a multimedia
corpus. Therefore, we lay the bases for a descriptive framework that will be
closer to an ``all-modalities'' and an ``all-perspectives'' inclusive one.
“Unseen gestures" and the Mind of the
Speaker: An analysis of co-verbal gestures in map-task activities
Nicla Rossini
Dipartimento di Linguistica Teorica e Applicata
Università degli Studi di Pavia, tattvamasi@libero.it;
Abstract:
The
analysis of co-verbal gestures in map-task activities is particularly
interesting for several reasons: on the one hand, the speaker is engaged in a
collaborative task with an interlocutor; on the other hand, the task itself is
designed in order to place a cognitive demand on both the speaker and the
receiver, who are not visible to one another. The cognitive effort in question
implies the activation of different capabilities, such as self-orientation in
space, planning (which can also be considered a self-orientation task
concerning the capability of organising successful communicative strategies for
the solution of a given problem), and communication in "unnatural"
conditions.
The
co-verbal gestures performed during such a task are quantitatively and
qualitatively different from those performed in normal conditions, and can
provide information about the Mind of the Speaker (Poggi & Magno
Caldognetto, 1997). In particular, the recursive pattern of some metaphors
(McNeill, 1992 and following) can be interpreted as a reliable index of the
communicative strategy adopted by the speaker: recurrent metaphors indicating
the adoption of a plan, its abandonment, or its confirmation will be shown and
analysed. Moreover, cases of gestures indicating the opposition between Given
and New (Halliday, 1985), and other basic psycholinguistic phenomena centred on
collaborative speech acts, such as awaiting feedback, frustration,
wrong-footing, etc., will be discussed and compared with the co-verbal
gesticulation of subjects intent on a face-to-face interaction.
On the analysis of fundamental frequency
control characteristics of nonverbal utterances and its application to
communicative prosody generation
Ke
Li, Yoko Greenberg and Yoshinori Sagisaka
Waseda
univ. GITI 29-7 building 1-3-10
Nishi-Waseda Shinjuku-ku Tokyo 169-0051 Japan, yoshinori.sagiska@atr.jp;
Abstract:
Aiming at communicative speech generation, control characteristics of nonverbal
utterances were analyzed. From the analyses using F0 generation model,
utterance specific control characteristics were observed. Their prosodic
characteristics are linked to the multi-dimensional vectors expressing
listener’s subjective impression. A quantitative prosody control scheme is newly proposed to test the
validity of F0 generation and their effectiveness
is conformed by perceptual evaluation tests.
Face recognition using sparse meshes: a promising
approach
Samokhval Vladimir
United Institute of
Informatics Problems National Academy of Sciences of Belarus
Surganova 6, 220012 Minsk,Belarus
Abstract:
The
sparse meshes are considered as the suitable tool for construction of the
classifier in recognition problems of human faces. Their use for face modeling
is based on a number of remarkable properties of such data presentation and
additional opportunities to increase a level of authentic recognition. In
particular, representation of area of interest in the form of a 2,5-dimensional
mesh potentially allows to receive missing foreshortenings of the image of
human face, that essentially increases recognition rate as it is shown with use
of PCA and discriminant analysis methods. For successful work of PCA method is
necessary to receive the frontal image of the face, and it is possible to
rotate a mesh in depth on a certain angle. At the synthesis of discriminant
filters their functioning is directly connected with the volume of training
sample. In this case mesh rotations enable to receive additional views of
facial image and to expand training set. Besides, the degree of mesh sparseness
in itself is the parameter, that influences both on classification results, and
on the volume of calculations, and finally on speed and system functioning. In
this research we consider some aspects of the performance of PCA and
discriminant analysis methods for which input data are 2,5-dimensional models
of the face in the form of sparse meshes, and also we establish a degree of
sparseness of these meshes for optimum performance of recognition system.
Verbal and nonverbal resources in
constructing the topical flow in early interaction in picture book environment
Sari Karjalainen
Department of Speech Sciences, Siltavuorenpenger
20 A/F, P. O. Box 9, 00014 University of Helsinki, Finland , sari.karjalainen@helsinki.fi
Abstract;
The
gestures, particularly pointings, will be analyzed in the process of topical
co-operation between adult and child at preverbal stage and, especially, how
these resources are used in making the shifts within the topic when looking at
picture books. The method for the study is qualitative and data driven CA
(conversation analysis). The data base is composed of videotaped naturalistic
picture book conversations between child (at the age from 1 to 2 years) and
adult. Video data from 6 Finnish families is transcribed and analyzed and the
events are presented with different text-based transcriptions, and also a
computer-aided visualization of these annotations to supplement the analysis.
The micro-analysis is focused on the sequential organization of the
participants’ verbal and nonverbal action (pointing and other gestures, gaze,
vocalizations and adult's speech) and, especially, on the sequences where the
topic is extended from the referent in the picture book to the noticeable or
non-noticeable referents outside the book. The child’s acts, for example, the
pointing at the window, get different meanings in different sequential
contexts. The topical sequences of different kind will be presented focusing on
how both the verbal and nonverbal resources used in referring to picture
referents and related referents outside the picture book reveal the use of
already existing shared knowledge or constructing the shared knowledge between
the participants.
Visual Search, Baggage Screening, and the
Assessment of Mental Workload through the Analysis of Eye-Movements
Michela
Terenzi & Francesco Di Nocera
Cognitive
Ergonomics Laboratory, Department of Psychology, University of Rome “La
Sapienza”
Abstract:
In
light of the events of September 11, 2001, many efforts have been made to
support security officers in identifying potential threats. Most of them are technological
aids used by airports’ personnel when performing security operations such as
baggage screening. However, automation support is known to alleviate some tasks
and, at the same time, to create new forms of workload. For this reason, also
Human Factors / Ergonomics researchers addressed the vast range of technical
challenges that, because of these acts of terrorism, now face society (Hanckock
& Hart, 2002). The most important issues in this field, is the analysis of
the operators’ mental workload, which is a key factor in determining human
error. Therefore, it is crucial to find viable strategies for minimizing the
cognitive load, for optimizing work schedule, and for managing automation
support by workload-matched procedures.
Indeed,
one approach for improving performance could be to support the limited human
information processing capabilities through the use of adaptive aids, triggered
by variations in human physiology and behavior. Previous studies (Di Nocera et
al, 2006a; 2006b) showed a relation between the distribution of eye fixations
and workload, providing a real-time measure of the operator’s load. In the
present study, a typical visual search task was used. Subjects were requested
to find a target among a set of distractors. Eye movements were recorder during
the task. Results showed sensitivity of the proposed index to variations in
mental workload, thus confirming the utility of fixations patterns as triggers
for adaptive automation.
The socio-cultural differences and the personal traits
in Ukraine’s scientific life
Oksana Udovyk and Oleg Udovyk
National University “Kyiv Mohyla
Academy”and National Institute for Strategic Studies Kyiv, Ukraine, E-mail: xenna_2003@ukr.net and oleg_udovyk@hotmail.com
Abstract:
Ukraine suffers from an identity crisis that is
inhibiting its scientific, as well as its economic and political, development.
The 47 million inhabitants of the former Soviet republic are deeply divided
between pro-European and pro-Russian factions. The celebrated 'orange
revolution' of November 2004 did less to bridge this divide than is commonly
thought.
The nation's
research system broadly reflects this wider societal divide. On the one hand,
there are many young, well-educated and highly motivated researchers and a
network of increasingly independent universities. On the other, there's the
National Academy of Sciences of Ukraine, a leviathan of militant senility that
retains just enough power to control critical aspects of Ukraine's scientific
life.
The academy
employs 47,000 permanent staff in a network of largely unproductive research
establishments. Given the advanced age of its senior management, time alone
will eventually resolve the issue. But that won't happen soon enough for those
young Ukrainians currently in search of a productive scientific career.
Integrating the
Ukraine into the Framework research programme of the European Union (EU) would
allow this generation far greater interaction with its peers abroad. The
European Commission supports the idea, which could also help open the way to
future EU membership for Ukraine. But the leadership of the academy, deeply
rooted in Soviet traditions, seems to be thwarting such integration through a
mixture of contrariness and lack of interest.
A high-level
EU–Ukrainian steering committee on scientific cooperation, for example, was
established on paper four years ago but has yet to actually meet. When it does,
the academy's leaders are expected to obstruct collaborative steps that might
bring an infusion of foreign influences into the country — including respect
for the value of independent peer review.
Ukrainian science
has potential in several spheres, including materials sciences, radioastronomy,
theoretical physics and agricultural research. The nation badly needs to focus
its scarce resources in those areas where its scientists can compete, and
dispose of some of its anachronistic scientific heritage. That will require a
rigorous external evaluation of the performance of hundreds of the academy's
institutes.
The government
needs to identify these reforms as a priority and then act with determination
to overcome the academy's likely resistance to them. The oligarchy that has
controlled Ukrainian science since Soviet times may then lose out. But the
nation's economic potential and its prospects for integration into the EU, as
well as science itself, can only benefit. Reform of Ukraine's archaic research
system is needed sooner rather than later.
Recognizing the effects of voluntary facial
activations using heart rate patterns
Toni Vanhala and Veikko
Surakka
Research Group for
Emotions, Sociality, and Computing Tampere Unit for Human-Computer Interaction,
Dept. of Computer Sciences, FI-33014 University of Tampere, Finland, Email:
Toni.Vanhala@cs.uta.fi
Abstract:
Continuously
measured physiological signals have the potential to act as non-invasive, real
time indicators of human psycho-physiological phenomena. Recently, several
non-intrusive, wireless, and discrete measurement devices have been developed.
For these reasons, there has been growing interest for using physiological
signals for estimating emotions and other psychological processes during
human-computer interaction, as well as for person identification [e.g. 1]. Due
to the interaction of the human physiological and psychological systems there
are several unique challenges for analyzing these signals. In the current work,
we present the first steps towards constructing an online system that
automatically identifies heart rate responses and estimates subjective
experiences during voluntary facial activations. The preliminary results of our
study showed that voluntarily produced facial expressions had an effect on
subjective emotional experiences and physiological processes. Further, our
results suggest that heart rate responses to facial activations can be detected
in order to develop face detection systems for more accurate, online
person-identification and emotion recognition.
1.
Poulos, M. Rangoussi,
M., Chrissikopoulos, V., Evangelou, A. (1999) Parametric person identification
from the EEG using computational geometry. In Proceedings of ICECS '99, 1005-1008.
On the analysis of disfluencies in large
spontaneous speech corpora: the case of autonomous fillers
Ioana Vasilescu
Limsi-Cnrs Spoken Language
Processing Group Bat.508 Bp 133 F-91403 Orsay Cedex France, ioana@limsi.fr
Abstract:
The
hesitation or “edition” phenomena, such as filled pauses, silent pauses, word
lengthening etc. are widely encountered in world’s languages. They are to be
distinguished from the lexical level. Consequently, they have been for decades
considered as “speech disfluencies”, i.e. articulatory events without a role in
building the verbal message. Recently, the research of cognitivists such as
Clark and Fox Tree, brought into light a new decoding of the presence of those
phenomena in speech [1]. The authors focused more particularly on the
autonomous fillers in English (“uh”, “um”), which are defined as long and
stable vocalic segments, potentially inserted at any moment within spontaneous
speech. According to the authors, those items play a role in communication,
i.e. “to announce the initiation of what is expected to be a […] delay in
speaking”.
My
work focuses on the analysis and modeling of the speech disfluencies in the
framework of automatic speech processing in a multilingual context. In this
purpose I analyzed fillers from a multilingual corpus of broadcast news in
Arabic, Mandarin Chinese, French, German, Italian, European Portuguese,
American English and Latin American Spanish. I addressed so far the question of
the specificity of acoustic fillers models: generic across languages or
language-dependent. I will present some acoustic and perceptual findings
supporting the second hypothesis. I will also mention the effects of number of
external factors which influence the acoustic and prosodic patterns of fillers
in different types of speech corpora (prepared, conversational etc.). Among
those factors language, gender, speaking style and language proficiency
engender significant variation needing to be taken into account in order to accurately
model the phenomenon.
1.
H.H., Fox Tree J.E., Clark, Using uh and um in spontaneous
speaking, Cognition 84, 73-111, 2002.
Prosodic Cues for Automatic Phrase Boundary
Detection in ASR
Klara VICSI - Gyorgy SZASZAK
Budapest University for
Technology and Economics, Dept. for Telecommunications and Mediainformatics,
Budapest, Hungary.
ABSTRACT:
This
article presents a cross-lingual study for Hungarian and Finnish about the
segmentation of continuous speech on word and phrasal level based on prosodic
features. A word level segmentationer has been developed which can indicate the
word boundaries with acceptable accuracy for both languages. The ultimate aim
is to increase the robustness of Automatic Speech Recognizers (ASR) by
detection of word and phrase boundaries, and thus significantly decrease the
searching space during the decoding process, very time-consuming in case of
agglutinative languages, like Hungarian and Finnish. They are however fixed
stressed languages, so by stress detection, word beginnings can be marked with
reliable accuracy. Algorithms based on data-driven (HMM) approach were
developed and evaluated. The best results were obtained by time series of
fundamental frequency and energy together. Syllable length was found to be much
less effective, hence was discarded. By use of supra-segmental features, word
boundaries can be marked with high correctness ratio, if we not going to find
all of them. The method we evaluated is easily adaptable to other fixed-stress
languages. To investigate this we adapted our data-driven method to the Finnish
language and obtained similar results.
Experiments in Assessing the Validity and
Reliability of Item N-gram Distributions in Texts as Proxy for Textual
Fingerprints
Carl Vogel
Computational
Linguistics Group, Trinity College, University of Dublin, Dublin 2, Ireland.
Abstract:
My
research in this area is associated with text rather than speech or other
visual cues, and is driven by hypotheses within forensic linguistics that there
are valid and reliable methods which can be used for authorship attribution
tasks. Behind these hypotheses is the
claim that individuals essentially unconsciously fingerprint themselves in the
texts that they produce. There are many
proposals in the literature and includes analysis of consciously manipulated
author style, and other aspects of text that are extremely difficult for an
author to manipulate. Orthography is
one such aspect of texts.
While
a great deal of deliberation may go into selection of lexical items from open-class
categories, somewhat less reflection tends to be involved in closed class
categories, and therefore it is interesting to explore distributions of words
in closed class categories used by an author across texts and genre, and in
comparison with other authors as a complement to any analysis of lexical
richness or quirkiness. However, orthographic analysis crosses both categories,
and is basically driven by the fact that while one can choose one's words, one
does not generally choose how words are spelled.
Along
these lines I and my research group have been experimenting with character
n-gram analyses of texts and using similarity of character n-gram distributions
to guide the assessment of similarity among texts. The research involves comparative analysis, rating the character
approach with various values of n with other methods of tokenizing texts --
e.g. word n-grams, n-grams of part of speech tags, accounting for stop-words,
etc. Some of the experiments have been
on closed systems of texts in which authorship is actually known, others
involve partly open systems of texts in which the claim of single authorship is
disputed (e.g. whether one author is responsible for the entire Shakespeare
corpus, including the apocrypha), and still more open ended explorations in
sentiment analysis -- using essentially bag-of-character analysis to estimate
similarity among non-governmental political party positions on the basis of
election manifestos.
The
research relates to the scope of the meeting in that the communication of
information about authorship is assumed to be unintended by the author
(although it is acknowledged that some authors do directly manipulate
orthography for effect --- lipograms provide a relevant example), yet which can
potentially be used to identify the author.
I intended to present and obtain feedback on the experiments reported,
as well as directions of ongoing research which include tracking language
change in individuals over time (an issue perhaps of interest in research on
aging and early detection of neuro-degenerative disorders which impinge on
language production).
A
couple of relevant papers:
1.
Van Gijsel, Sofie and Carl Vogel (2003) Inducing a Cline from
Corpora of Political Manifestos, International Symposium on Information and Communication
Technologies, edited by Markus Aleksy, et al.,
pp 304 – 310.
2. O'Brien, Cormac
and Carl Vogel (2003) Spam Filters: Bayes vs. Chi-Squared; Letters vs. Words,
International Symposium on Information and Communication Technologies, edited
by Markus Aleksy, et al., pp 298 - 303.
Empty pauses detection in a noisy speech
conditions
Vojtich
Stejskala, Zdenek Smékala,
Anna Espositob
aDep. of Telecommunications, Brno University of
Technology, Purkynova 118, 612 00, Brno CZ, stejskal@kn.vutbr.cz;
bDep. of Psychology, Second University of Naples
and IIASS, Italy
Abstract:
Nowadays,
the most important role in a process of speech recognition plays successful pause
detection. There is need of robust detection algorithm, if we consider that
most of speech recordings are taken under very adverse conditions. This paper
presents a comparison of several algorithms for empty pauses detection on a
spontaneous speech records gained in noisy environments. Input signal is
transformed into log spectral energy and divided into specific frequency bands.
Each band is smoothed and tracked by dynamically adjusted threshold based on
pause (noise) energy estimation. Then the post processing edges correction
follows. All proposed algorithms are capable to process a real time input.
Images-Signes and Cognitive Scenes to
Anchor Comprehension in Language Learning: An Experiment on the Relationship between
Verbal and Non Verbal Communication through Italian Film.
Rosa
Volpe
148
rue Saint Honoré, 75001 Paris,
France, rvolpe@univ-orleans.fr;
Abstract:
This
study has as a central focus a foreign language classroom environment that
makes daily use of film discourse in the target language in order to provide
« in-context » and « situational training » as well as
« anchored » target language input. More specifically this study explores
(1) whether first-semester, first-year college students of Italian can and will
understand film narrative (2) if and how, the comprehension of the film
narrative will affect, if any, their written production. The first experimental
probe consists in a comprehension task. The results suggest that the Anchored
Learning Group performed better than the Basic-Skills Learning Group in the
comprehension of the two film segments. The second experiment probe consists in
a production task and shows that compared to the Basic-Skills Learning Group,
the written production of the Anchored Learning Group reflects far better the
structure of the narrative discourse they have been exposed to. In the occasion
of this production task, students were required to write their essay using at
least 10 verbs they were familiar with. The performance of the Anchored Group
was different from the performance of the Basic-Skills Group both in quality
and quantity. The anchored Group wrote more correct verbs, resulting in the
production of more correct sentences, and the structure of their discourse was
closer to the structure of the discourse of film narrative.
Language and Communication. An
Eco-Anthropological Point of View
GALATCHI Liviu-Daniel
Ovidius University of
Constanta, Str. Dezrobirii, nr. 114, bl. IS7, sc. A,
et. 4, apt. 16,
RO
900241, Constanta – 4, Romania, galatchi@univ-ovidius.ro or liviugalatchi@yahoo.com
Abstract:
Although
wild primates have only call systems, chimps and gorillas can understand and
manipulate non-verbal symbols based on language. Primates emit calls only in
the presence of particular environmental stimuli. Calls cannot be combined when
different stimuli are present simultaneously. At some point in human evolution,
our ancestors became capable of displaced speech. Other contrasts between
language and call systems include productivity and cultural transmission. Over
time, our ancestral call systems developed into true language. Call systems
grew too complicated for genetic transmission and began to rely on learning.
Language is the main system humans use to communicate, although we also use
nonverbal communications, gestures, and body stances and movements.
No
language includes all the sounds that the human vocal apparatus can make.
There
are culturally distinctive as well as universal relationships between language
and mental processes. The lexicons and grammars of particular languages can
lead speakers to perceive and think in particular ways. Speakers of different
languages categorize their experience differently. However, language does not
tightly restrict thought, because cultural changes can produce changes in
thought and in language.
People
vary their speech on different occasions, shifting styles, dialects, and
languages. As linguistic systems, all languages and dialects, are equally
complex, rule-governed, and effective for communication. However, speech is
used, is evaluated, and changes in the context of political, economic and
social forces. The linguistic traits of a low-status group are negatively
evaluated (often by the members of the group) not because of their linguistic
features but because they are associated with and symbolize low social status.
One dialect, supported by the dominant institutions of the state, exercises
symbolic domination over the others.
Cultural
similarities and differences often correlate with linguistic ones. Linguistic
clues can suggest past contacts between cultures. Related languages descend
from an original protolanguage. Relationships between languages don’t
necessarily mean that there are biological ties between their speakers, because
people can learn new languages.
Face Recognition Experiments on AR Database
Marco Grassi1, Marcos Faundez2
1Ingegneria
Eletronica dell'Universitá degli Studi di Ancona, margra75@hotmail.com
2Escola Universitaria Politecnica de Mataro, Spain
Abstract:
Biometric recognition and authentication based on face
recognitions can actually be used in many real-time applications such as:
surveillance, security systems, access control, and much more. For these
purposes the system has to grant a fast computation speed but also robustness
to illumination and facial expressions variations. The main objective of this
paper is to implement a face recognition system using the DCT (Discrete Cosine
Transform) method for characteristics extraction. Appling
the DCT to the image results possible to concentrate the information reducing
the dimensionality of the problem. For the classification has been used nearest
neighbour classifiers using MAD (Mean Absolute Difference) and MSE (Mean Square
Error), that grant a very fast computation, and a RBF (Radial Basis Function)
Neural Network, that presents a faster training than a classic Neural Network.
Simulation results, over the AR face Database, show
that the proposed system have very good performances with very fast computation
speed, high recognition rate and good robustness.
Overcomplete Blind Separation of Speech
Sources in the Post Non Linear case through Extended Gaussianization
S.Squartini, S.Cecchi,
E.Moretti, F.Piazza
A3Lab-DEIT Università
Politecnica delle Marche
Via Brecce Bianche 31,
60131, Ancona, Italy
Abstract - This
work deals with the blind separation problem in presence of more sources than
sensors and Post-Nonlinear (PNL) mixing. The addressed method is made of three separate
steps: compensation of nonlinearity, mixing matrix recovery and final unknown
source estimation. It has been recently proposed and successfully evaluated in
the case of synthetic mixtures of real world data (like speech signals). Here, the Extended Gaussianization approach
is employed to perform the first step instead of the common Gaussianization one
in order to reduce the approximation error on the linearized mixture pdfs.
Computer simulations allowed to achieve a significant improvement of separation
performances over the previous approach.
The relationships between gestures and
prosody: A preliminary investigation on Italian
A. Esposito1, D.
Esposito1, M. Refice2, M. Savino3, S.
Shattuck-Hufnagel4
1Department of
Psychology, Second University of Naples
(SUN), Italy
2Department
of Elettrotecnica, Politecnico di Bari, Italy
3Department of
Psychology, Università di Bari, Italy
4MIT, Research
Laboratory of Electronics, Cambridge, MA, USA
Abstract.
This
work investigates on the relationships between gestures and prosody, exploiting
a class of gestural movements named HITS and defined by Yasinnik, Renwick,
Shattuck-Hufnagel (2004) as: “An abrupt stop or pause in movement, which breaks
the flow of the gesture during which it occurs” Our analysis show that, as in
American English, also in Italian, these gestural entities are correlated with
high level prosodic units.
[1] Y. Yasinnik, M. Renwick, S. Shattuck-Hufnagel: The Timing of Speech-Accompanying Gestures with Respect to Prosody. Proceedings of the International Conference From Sound to Sense, C97-C102, MIT, Cambridge, June 10-13, 2004.
