Articulatory Modelling 2
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
Acoustic-Articulatory Evaluation of the Upper Vowel-Formant Region and its Presumed Speaker-Specific Potency
Authors:
Frantz Clermont, University of Tsukuba (Japan)
Parham Mokhtari, LORIA-Campus Scientifique (France)
Page (NA) Paper number 87
Abstract:
We present some evidence indicating that phonetic distinctiveness and speaker individuality, are indeed manifested in vowels' vocal-tract (VT) sha pes estimated from the lower and upper formants, respectively. The methodology developed to demonstrate this dichotomy, implicates Schroeder's (1967) acoustic-articulatory model which can be coerced to yield area-function approximations to VT-shapes of differing formant components. Using ten steady-state vowels recorded in /hVd/-context, five times at random, by four adult-male speakers of Australian English, VT-shape variability was then measured on an intra- and an inter-speaker basis. Gross shapes estimated from the lower formants, caused the largest spread amongst the vowels of individual speakers. By contrast, more detailed sha pes estimated from certain higher formants of front and back vowels, caused the largest spread amongst speakers. These results contribute a quasi-articulatory substantiation of a long-standing view on the speaker-specific potency of the upper vowel-formant region, together with some useful implications for speech and speaker recognition.
SL980087.PDF (From Author)
SL980087.PDF (Rasterized)
TOP
Control of Larynx Height in Vowel Production
Authors:
Philip Hoole, Institut fuer Phonetik, Munich University (Germany)
Christian Kroos, Institut fuer Phonetik, Munich University (Germany)
Page (NA) Paper number 1097
Abstract:
Digital video filming of the thyroid prominence was used to measure larynx height in German vowels, with focus on contrasts involving front unrounded, front rounded and back rounded vowels. The study aimed to provide a foundation for interpreting the acoustic consequences of articulatory maneuvres not only at the larynx but also elsewhere in the vocal tract. Results showed the expected pattern of lower larynx position for the rounded vowels. However no clear preference emerged for the same, more, or less larynx lowering on front rounded versus back rounded vowels. Coarticulatory effects of the flanking consonants were weak. The most striking result was that the magnitude of the differences between vowels varied substantially over speakers. This reinforces the contention that interpretation of vertical laryngeal gestures must be embedded in speaker-specific analysis of downstream articulatory maneuvres. Work in this direction is currently in progress.
SL981097.PDF (From Author)
SL981097.PDF (Rasterized)
TOP
Analyzing the Effect of Secondary Excitations of the Vocal Tract on Vocal Intensity in Different Loudness Conditions
Authors:
Paavo Alku, University of Turku (Finland)
Juha Vintturi, Helsinki Univ. Central Hospital (Finland)
Erkki Vilkman, University of Oulu (Finland)
Page (NA) Paper number 67
Abstract:
For voiced speech the main excitation of the vocal tract occurs at the end of the glottal closing phase when the rate of change of the flow reaches its absolute maximum. This study presents a straightforward method that yields a numerical value to characterize the effect of the main excitation on vocal intensity. The method, Energy Ratio by Modified Excitation (ERME), takes advantage of the glottal flow and the model of the vocal tract transfer function given by inverse filtering and it synthesizes two signals based on the source-filter theory. The first synthesized sound is produced using the glottal flow waveform given by inverse filtering per se. The second signal is synthesized by removing the main excitation from the differentiated glottal flow. ERME is defined as the ratio between the energy of the first synthesized signal and the energy of the second one. It is shown that when the loudness of speech increases, the value of ERME first rises but in the case of loud voices it starts to decrease. This behavior of ERME shows that effects of secondary excitations of the vocal tract that occur during glottal opening become important in the production of loud voices.
SL980067.PDF (From Author)
SL980067.PDF (Rasterized)
TOP
An Analysis of Modal Coupling Effects During the Glottal Cycle: Formant Synthesizers from Time-Domain Finite-Difference Simulations
Authors:
Gordon Ramsay, ICP-INPG (France)
Page (NA) Paper number 670
Abstract:
Speech is typically modelled using time-domain or frequency-domain simulations of the acoustic field in the vocal tract. Using a biorthogonal modal decomposition, it is shown that time-domain finite-difference simulations can be transformed algebraically into equivalent formant synthesizers, the parameters of which vary in time and are calculated directly from the laws of physics. Examining the structure of the equivalent formant synthesizer, it is observed that formant excitation is largely due to internal modal coupling effects, induced by rapid perturbation of the acoustic eigenmodes caused by vibration of the glottis, and does not rely precisely on external sources provided by boundary conditions. This leads to a novel interpretation and justification of traditional models of the glottal source.
SL980670.PDF (From Author)
SL980670.PDF (Rasterized)
TOP
Laryngoscopic Analysis of Pharyngeal Articulations and Larynx-Height Voice Quality Settings
Authors:
John H. Esling, University of Victoria (Canada)
Page (NA) Paper number 617
Abstract:
Using fibreoptic laryngoscopy to observe pharyngeal articulations, the aryepiglottic sphincter mechanism is shown to be responsible for the production of speech sounds in the phonetic category "pharyngeal." Major differences in auditory/acoustic quality are also produced when the larynx as a whole is raised or lowered during the production of pharyngeals. The voiceless pharyngeal fricative and voiced pharyngeal approximant are the result of increased sphincteric constriction of the laryngeal "tube" in a continuum that begins with normal glottal stop and ventricular fold closure. A pharyngeal stop is produced when the aryepiglottic sphincter mechanism achieves complete closure, and trilling accompanying friction is evident at the pharyngeal place of articulation in both voiceless and voiced modes. It is suggested that all five sounds share a common, pharyngeal place of articulation, but differ in manner of articulation. Raised larynx is the default setting for these articulations, but they may be produced with lowered larynx.
SL980617.PDF (From Author)
SL980617.PDF (Rasterized)
TOP
Effects of Shapes of Radiational Aperture on Radiation Characteristics
Authors:
Hiroki Matsuzaki, Hokkai-Gakuen University (Japan)
Kunitoshi Motoki, Hokkai-Gakuen University (Japan)
Nobuhiro Miki, Hokkaido University (Japan)
Page (NA) Paper number 656
Abstract:
The acoustic characteristics of acoustic tubes with protrusions at the radiation end are computed by FEM simulation. In the first experiment, two different shapes of the protrusions, a symmetrical and an asymmetrical shape with respect to the vertical, are investigated. Frequency characteristics of the radiation impedance are computed from simulation results. The simulation results show that the results of FEM simulation are in good agreement with our measurement results. The proposed 3-D radiational model is useful for analysis of the acoustic characteristics of human speech. In the 2nd experiment, the protrusion is attached to our 3-D vocal tract model. The vocal tract shape corresponds to the Japanese vowel /a/. The cross sections of the tubes are eliptic in shape. The simulation results show that the vocal tract transfer function of the FEM results is different from our previous FEM results and 1-D analytical solution.