A5047637263- Speech Recognition and Synthesis
- Phonetics and Phonology Research
- Speech and Audio Processing
- 3D Shape Modeling and Analysis
- Dupuytren's Contracture and Treatments
- Natural Language Processing Techniques
- Voice and Speech Disorders
- Medical Image Segmentation Techniques
- Music and Audio Processing
- Medical and Health Sciences Research
- Image and Signal Denoising Methods
- Computer Graphics and Visualization Techniques
- Algorithms and Data Compression
- Clinical practice guidelines implementation
- Human Pose and Action Recognition
- Image Processing and 3D Reconstruction
- Advanced MRI Techniques and Applications
- 3D Surveying and Cultural Heritage
- Optical Imaging and Spectroscopy Techniques
- Cerebrovascular and Carotid Artery Diseases
- Stroke Rehabilitation and Recovery
- Underwater Acoustics Research
- Advanced Neuroimaging Techniques and Applications
- Face and Expression Recognition
- Parkinson's Disease and Spinal Disorders
Institute for Language and Speech Processing
2023
Imagerie Adaptative Diagnostique et Interventionnelle
2019-2022
Université de Lorraine
2019-2021
Laboratoire Lorrain de Recherche en Informatique et ses Applications
2019-2021
Centre National de la Recherche Scientifique
2019-2021
Inserm
2019-2021
National Technical University of Athens
2018
Heat Transfer Research (United States)
2013
University College London
1999-2003
The work is concerned with the signal-to-symbol problem of building skinned, segmented, landmarked and labeled 3D models whole human body from range data. A fully automated model based process presented that takes raw data, cleans skins it, then locates "interesting" features, to enrich surface symbolic information for specific applications. method validated via volumetrics in medicine anthropometry.
A bstract : Measurement of body volume (BV) can be used to estimate composition using two‐ or multicomponent models. The traditional approach, underwater weighing (UWW), is awkward and unsuitable for many subjects. newer alternative, whole air displacement plethysmography (ADP), less demanding but still young children, who may not remain during the measurement. We have, therefore, considered whether a novel three‐dimensional photonic scanning, viable alternative. Duplicate measurements were...
Abstract The study of articulatory gestures has a wide spectrum applications, notably in speech production and recognition. Sets phonemes, as well their articulation, are language-specific; however, existing MRI databases mostly include English speakers. In our present work, we introduce dataset acquired with from 10 healthy native French A corpus consisting synthetic sentences was used to ensure good coverage the phonetic context. real-time technology temporal resolution 20 ms acquire vocal...
There are an increasing number of applications that require the construction computerized human body models. Recently, Hamamatsu Photonics have developed accurate and fast scanner based on position-sensitive photon detectors, capable providing in a few seconds dense representation body. The work presented here is designed to exploit such scanner's capabilities. An algorithm introduced deals with surface-from-curves problem can be combined existing curves-from-points solve surface-from-points...
There are an increasing number of applications that require the construction computerised human body models. The work presented here is a follow-up on previously surface reconstruction algorithm has been greatly improved by following local approach. advantages method along with explanation why its results better compared to previous algorithm. result compound, multi-segment, and yet entirely smooth watertight surface. Such strong potential being used for mainly medical nature, such as...
In this paper, we investigate the use of articulatory information, and more specifically real time Magnetic Resonance Imaging (rtMRI) data vocal tract, to improve speech recognition performance. For purpose our experiments, from rtMRI-TIMIT database. Firstly, Scale Invariant Feature Transform (SIFT) features are extracted for each video frame. Afterwards, SIFT descriptors frame transformed a single histogram per picture, by using Bag Visual Words methodology. Since kind information is...
We present an algorithm for augmenting the shape of vocal tract using 3D static and 2D dynamic speech MRI data.While images have better resolution provide spatial information, capture transitions.The aim this work is to combine strong points these two types data obtain image quality extend domain.To produce a consonant-vowel (CV) sequence, our takes as input CV transition targets C V. To enhanced sequence images, first step find transformation between mid-sagittal slice acoustically...
The new Air Quality Directive (2008/50/EC) encourages the introduction of modelling as a necessary tool for air quality assessment and management. Towards this aim, Management System (AQMS) has been developed installed in Department Labour Inspection Republic Cyprus. core system handles compilation an emissions inventory that includes data from all major activity sectors functions on basis continuous update database. Emission are then fed into advanced simulates concentration fields...
We evaluate velocity of the tongue tip with magnetic resonance imaging (MRI) using two independent approaches. The first one consists in acquisition a real-time technique mid-sagittal plane. Tracking manually and computer vision method allows its trajectory to be found calculated as derivative coordinate. also propose use another approach—phase contrast MRI—which enables velocities moving tissues measured directly. recorded sound simultaneously MR which enabled us make conclusions regarding...
Pretreatment ischemic location may be an important determinant for functional outcome prediction in acute stroke. In total, 143 anterior circulation stroke patients the THRACE study were included. Ischemic lesions semi-automatically segmented on pretreatment diffusion-weighted imaging and registered brain atlases. The percentage of tissue each atlas-segmented region was calculated. Statistical models with logistic regression support vector machine built to analyze predictors outcome....
In this work we present an algorithm for synthesising pseudo rtMRI data of the vocal tract. on midsagittal plane were used to synthesise target consonant-vowel (CV) using only a silence frame speaker. For purpose, several single speaker models created. The input is both train and CV. An image transformation computed from each CV next one, creating set transformations that describe dynamics production. Another adapt previously adapted applied to synthesise his/her data. Synthesised images...
In this paper we propose an algorithm for estimating vocal tract para sagittal slices in order to have a better overview of the behaviour articulators during speech production. The first step is align consonant-vowel (CV) data plains between them train speaker. Sets transformations that connect midsagittal frames with neighbouring ones acquired Another set calculated which transforms speaker corresponding test and used adapt domain previously computed sets transformations. newly adapted are...
In this work, we address the problem of creating a 3D dynamic atlas vocal tract that captures dynamics articulators in all three dimensions order to create global speaker model independent speaker-specific characteristics. The core steps proposed method are temporal alignment real-time MR images acquired several sagittal planes and their combination with adaptive kernel regression. As preprocessing step, reference space was created be used remove anatomical information speakers keep only...
Chara Tsoukala, Kosmas Kritsis, Ioannis Douros, Athanasios Katsamanis, Nikolaos Kokkas, Vasileios Arampatzakis, Sevetlidis, Stella Markantonatou, George Pavlidis. Proceedings of the Second Workshop on NLP Applications to Field Linguistics. 2023.