Prosody and Emotion 6
HomeFull List of Titles
1: ICSLP'98 Proceedings
Keynote SpeechesText-To-Speech Synthesis 1Spoken Language Models and Dialog 1Prosody and Emotion 1Hidden Markov Model Techniques 1Speaker and Language Recognition 1Multimodal Spoken Language Processing 1Isolated Word RecognitionRobust Speech Processing in Adverse Environments 1Spoken Language Models and Dialog 2Articulatory Modelling 1Talking to Infants, Pets and LoversRobust Speech Processing in Adverse Environments 2Spoken Language Models and Dialog 3Speech Coding 1Articulatory Modelling 2Prosody and Emotion 2Neural Networks, Fuzzy and Evolutionary Methods 1Utterance Verification and Word Spotting 1 / Speaker Adaptation 1Text-To-Speech Synthesis 2Spoken Language Models and Dialog 4Human Speech Perception 1Robust Speech Processing in Adverse Environments 3Speech and Hearing Disorders 1Prosody and Emotion 3Spoken Language Understanding Systems 1Signal Processing and Speech Analysis 1Spoken Language Generation and Translation 1Spoken Language Models and Dialog 5Segmentation, Labelling and Speech Corpora 1Multimodal Spoken Language Processing 2Prosody and Emotion 4Neural Networks, Fuzzy and Evolutionary Methods 2Large Vocabulary Continuous Speech Recognition 1Speaker and Language Recognition 2Signal Processing and Speech Analysis 2Prosody and Emotion 5Robust Speech Processing in Adverse Environments 4Segmentation, Labelling and Speech Corpora 2Speech Technology Applications and Human-Machine Interface 1Large Vocabulary Continuous Speech Recognition 2Text-To-Speech Synthesis 3Language Acquisition 1Acoustic Phonetics 1Speaker Adaptation 2Speech Coding 2Hidden Markov Model Techniques 2Multilingual Perception and Recognition 1Large Vocabulary Continuous Speech Recognition 3Articulatory Modelling 3Language Acquisition 2Speaker and Language Recognition 3Text-To-Speech Synthesis 4Spoken Language Understanding Systems 4Human Speech Perception 2Large Vocabulary Continuous Speech Recognition 4Spoken Language Understanding Systems 2Signal Processing and Speech Analysis 3Human Speech Perception 3Speaker Adaptation 3Spoken Language Understanding Systems 3Multimodal Spoken Language Processing 3Acoustic Phonetics 2Large Vocabulary Continuous Speech Recognition 5Speech Coding 3Language Acquisition 3 / Multilingual Perception and Recognition 2Segmentation, Labelling and Speech Corpora 3Text-To-Speech Synthesis 5Spoken Language Generation and Translation 2Human Speech Perception 4Robust Speech Processing in Adverse Environments 5Text-To-Speech Synthesis 6Speech Technology Applications and Human-Machine Interface 2Prosody and Emotion 6Hidden Markov Model Techniques 3Speech and Hearing Disorders 2 / Speech Processing for the Speech and Hearing Impaired 1Human Speech ProductionSegmentation, Labelling and Speech Corpora 4Speaker and Language Recognition 4Speech Technology Applications and Human-Machine Interface 3Utterance Verification and Word Spotting 2Large Vocabulary Continuous Speech Recognition 6Neural Networks, Fuzzy and Evolutionary Methods 3Speech Processing for the Speech-Impaired and Hearing-Impaired 2Prosody and Emotion 72: SST Student Day
SST Student Day - Poster Session 1SST Student Day - Poster Session 2Author Index
A B C D E F G H I
J K L M N O P Q R
S T U V W X Y Z
Multimedia Files
A Contrastive Study of Lexical Stress Placement in Singapore English and British English
Authors:
Ee Ling Low, Nanyang Technological University (Singapore)
Esther Grabe, University of Cambridge (U.K.)
Page (NA) Paper number 98
Abstract:
Singapore English and British English have been claimed to differ in lexical stress placement. Examples cited in the literature involve polysyllabic words such as 'hopelessly' and compounds such as 'blackboard'. Such words are stressed word-initially in BE, but are said to be stressed word-finally in SE. Two observations lead us to explore the claim that SE and BE differ in lexical stress placement. Firstly, observations about stress differences between SE and BE are based solely on auditory impressions by British English listeners. Acoustic evidence is not available. Secondly, the auditory evidence comes from realisations of test words in citation form, i.e. in nuclear, phrase-final position. If Singapore English has more phrase-final lengthening than British English, then this may account for the suggested differences in lexical stress placement. In the present paper, we investigate the acoustic evidence for the suggested cross-varietal difference.
SL980098.PDF (From Author)
SL980098.PDF (Rasterized)
TOP
Integrated Recognition of Words and Phrase Boundaries
Authors:
Florian Gallwitz, University of Erlangen-Nuremberg (Germany)
Anton Batliner, University of Erlangen-Nuremberg (Germany)
Jan Buckow, University of Erlangen-Nuremberg (Germany)
Richard Huber, University of Erlangen-Nuremberg (Germany)
Heinrich Niemann, University of Erlangen-Nuremberg (Germany)
Elmar Nöth, University of Erlangen-Nuremberg (Germany)
Page (NA) Paper number 328
Abstract:
In this paper we present an integrated approach for recognizing both the word sequence and the syntactic-prosodic structure of a spontaneous utterance. We take into account the fact that a spontaneous utterance is not merely an unstructured sequence of words by incorporating phrase boundary information into the language model and by providing HMMs to model boundaries. This allows for a distinction between word transitions across phrase boundaries and transitions within a phrase. During recognition, the syntactic-prosodic structure of the utterance is determined implicitly. Without any increase in computational effort, this leads to a 4% reduction of word error rate, and, at the same time, syntactic-prosodic boundary labels are provided for subsequent processing. The boundaries are recognized with a precision and recall rate of about 75% each. They can be used to reduce drastically the computational effort for parsing spontaneous utterances. We also present a system architecture to incorporate additional prosodic information.
SL980328.PDF (From Author)
SL980328.PDF (Rasterized)
TOP
Phrase Accents Revisited: Comparative Evidence From Standard and Cypriot Greek
Authors:
Amalia Arvaniti, University of Cyprus (Cyprus)
Page (NA) Paper number 550
Abstract:
Phrase accents, one of the three tonal categories assumed by much recent research on intonation, are expected to associate with a prosodic boundary (e.g. the end of the utterance) but not to phonetically align with a specific tone-bearing unit (TBU), such as a stressed syllable. This paper presents experimental evidence on the intonation of Cypriot Greek polar questions suggesting that phrase accents prefer to associate with specific TBUs. Concretely, it is shown that in Cypriot Greek polar question intonation, autosegmentally described as L* H L%, the H phrase accent does not align with the final stressed vowel as in Standard Greek, but instead it aligns approximately 30ms from the onset of either the penultimate or the final vowel of the utterance. The data provide evidence that phrase accents, like other tonal categories, exhibit stable phonetic alignment and support Ladd, Arvaniti and Mennen's (1997) typology of stress-seeking and non-stress-seeking phrase accents.
SL980550.PDF (From Author)
SL980550.PDF (Rasterized)
TOP
Phonetic Invariance and Phonological Stability: Lithuanian Pitch Accents
Authors:
Grzegorz Dogil, University of Stuttgart (Germany)
Gregor Möhler, University of Stuttgart (Germany)
Page (NA) Paper number 206
Abstract:
We argue that phonetically invariant realizations of phonological categories imply the synchronic and diachronic imperviousness of such categories to phonological rules and sound laws. We claim that phonetic invariance is the foundation of phonological stability. The category we discuss in this contribution is the pitch-accent. We provide a parametric phonetic description of this phonological category. By means of a parametrization technique we apply this description to the contrastive pitch-accents of Lithuanian. The statistic differences between acute and circumflex pitch-accents derived by the parametrization provide a basis for the discussion of synchronic and diachronic behavior of the phonetically nonbalanced phonological contrasts.
SL980206.PDF (From Author)
SL980206.PDF (Rasterized)
TOP
A HMM-Based Recognition System for Perceptive Relevant Pitch Movements of Spontaneous German Speech
Authors:
Christel Brindöpke, University Bielefeld (Germany)
Gernot A. Fink, University Bielefeld (Germany)
Franz Kummert, University Bielefeld (Germany)
Gerhard Sagerer, University Bielefeld (Germany)
Page (NA) Paper number 503
Abstract:
This paper presents an HMM-based recognition system for perceptive relevant pitch movements of spontaneous German speech. The pitch movements are defined according to the perceptively and phonetically motivated IPO-approach to intonation. For recognition we use a hybrid approach combining polynomial classification with Hidden Markov Modelling. The recognition is based only on the speech signal, its fundamental frequency and eleven derived features. We evaluate the system on a speaker independent recognition task.
SL980503.PDF (From Author)
SL980503.PDF (Rasterized)
TOP
Towards a Reversible Symbolic Coding of Intonation
Authors:
Jean Véronis, Université de Provence (France)
Estelle Campione, Université de Provence (France)
Page (NA) Paper number 846
Abstract:
This paper presents a two-step model for the symbolic coding and generation of intonation. First, the F0 curve is reduced to a series of pitch target points that capture the macroprosodic information of the utterance. Target points are then converted into a sequence of labels. Generation is achieved through the reverse steps. The model is language independent and requires no prior training on the data. We discuss the influence of the number of categories on the precision of fit, and show, by an evaluation on a large multilingual corpus (4 hours 20 minutes of speech, 50 speakers, 5 languages) that a model composed of three ascending and three descending categories, plus a category for small or null movements enables a regeneration of ca. 99% of points at less than 2 ST than the original. Given that the model is capable of various improvements, it seems a good candidate for practical applications.