Prosody and Emotion 7
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
On the Amount and Domain of Focal Lengthening in Swedish
Authors:
Eva Strangert, Department of Phonetics, Umeaa University (Sweden)
Mattias Heldner, Department of Phonetics, Umeaa University (Sweden)
Page (NA) Paper number 462
Abstract:
Temporal effects of focus in Swedish were studied in short sentences with systematic variation of the length and prosodic pattern of target words in different syntactic positions. Generally, focus caused an average increase in word duration of about 25%. Variations of word length, stress, or word accent pattern did not produce any systematic effects on the amount of lengthening of the target word, but the lengthening varied considerably between speakers and different positions in the sentence. The most extensive lengthening occurred in combination with the insertion of a boundary after the word in focus in those cases when the word preceded a strong syntactic boundary. Within words, stressed syllables were lengthened most, and lengthening of a primary and following secondary stressed syllable was equal to that of a single primary stressed syllable. As unstressed syllables were also affected, the domain of focal lengthening is assumed to be the word.
SL980462.PDF (From Author)
SL980462.PDF (Rasterized)
TOP
Differential Lengthening Of Syllabic Constituents In French: The Effect Of Accent Type And Speaking Style
Authors:
Daniel Hirst, CNRS, Université de Provence (France)
Corine Astésano, CNRS, Université de Provence (France)
Albert Di Cristo, CNRS, Université de Provence (France)
Page (NA) Paper number 733
Abstract:
This paper presents results from the analysis of segmental duration in French in three different speaking styles, (a) 'reading'; (b) 'news broadcast' ; (c) 'spontaneous interview'. A ten minute corpus was hand segmented and labelled using six accent categories (i) unstressed (ii) word-initial (iii) emphatic word-initial (iv) word-final (v) intonation unit final before non-terminal boundary (vi) intonation unit final before terminal boundary. Results show that when different accent types and different speaking styles are taken into account there is no uniform lengthening of prosodic constituents in French, whether the syllable or some other suprasegmental unit is taken as the domain for lengthening. Instead differential lengthening is observed consistently with essentially greater lengthening towards the beginning of the syllable for initial prominence and greater lengthening towards the end of the syllable for final prominence. The degree of the different types of lengthening was moreover significantly different across the speaking styles.
SL980733.PDF (From Author)
SL980733.PDF (Rasterized)
TOP
Prosodic Analysis of Fillers and Self-Repair in Japanese Speech
Authors:
Felix C.M. Quimbo, Kyoto University (Japan)
Tatsuya Kawahara, Kyoto University (Japan)
Shuji Doshita, Kyoto University (Japan)
Page (NA) Paper number 762
Abstract:
The prosodic features of filled pauses (fillers) and self-repair are investigated with a view towards the detection of disfluencies. First, we compare the prosodic features of typical fillers and their fluent homonyms using read sentences of identical phoneme sequences. It is confirmed that the fillers (1) have at least 2 times longer duration than their non-disfluent counterparts, (2) tend to be followed by definitely longer pauses, and (3) have much smaller movement in their pitch contours. Then, the spontaneous fillers segmented out from a dialogue corpus are also analyzed. The same tendency is confirmed, but some samples lie halfway between the read fillers and their fluent homonyms. The abruptly cut-off endings in the self-repair are also analyzed by comparing with the ordinary endings of words. It is found that a short phoneme ending coupled with a relatively short succeeding pause indicates the abrupt cutoff.
SL980762.PDF (From Author)
SL980762.PDF (Rasterized)
TOP
A Synthesis-Oriented Model Of Phrasal Pitch Movements In Standard Chinese
Authors:
Jinfu Ni, University of Tokyo (Japan)
Goh Kawai, University of Tokyo (Japan)
Keikichi Hirose, University of Tokyo (Japan)
Page (NA) Paper number 750
Abstract:
This paper proposes a computable, tone-driven method to model phrasal pitch movements in standard Chinese by (1) formulating physical constraints for phonetic control mechanisms, (2) defining four phrasal tones quantified by model parameters for generating phrasal tunes, and (3) forming phrasal pitch movements by mapping lexical tones onto phrasal tunes. Several experiments confirm the method's validity. The proposed method is an effective component technology for analyzing, synthesizing and understanding spoken Chinese intonation.