Skip to main content
Aller à la page d’accueil de la Commission européenne (s’ouvre dans une nouvelle fenêtre)
français fr
CORDIS - Résultats de la recherche de l’UE
CORDIS
Contenu archivé le 2024-04-16

High-Resolution Speech Recognition : Auditory Connectionist Technologies for Speech

Objectif

The key research questions explored were:
-How can auditory processing performed in the cochlea be simulated?
-How can the transient information concerning the onsets and offsets of parts of speech be extracted from an early stage of the auditory processing chain?
-How can existing connectionist recognisers be developed to perform vowel recognition using the cochlea simulation as a front-end ?
-How can training techniques for connectionist recognisers be improved?
-How do humans perform on a phoneme recognition task in the absence of lexical and contextual constraints?
The computer modelling of auditory processing and connectionist systems for speech recognition was addressed. Based on research into peripheral auditory processing and cognitive research on speech perception and memory, the primary goal was to develop a recognition system that benefits from the processing embodied in current models of the human auditory system.

At the present time, developed software and published research on the project include:
an efficient computer simulation of cochlea processing, documentation of the software for users in speech recognition and hearing research and published papers describing experiments on octave perception and phoneme recognition;
efficient implementation of a connectionist recognizer producing performance comparable to the best fast Fourier transform (FET) filterbank and improved training and pattern recognition algorithms;
experimental research on the structure of human phonological memory.
APPROACH AND METHODS
The goal was to construct a recognition system involving a high-resolution front-end processor based on current models of the human auditory system, and a word-recognition system based on abstract phonological representations and a phonological buffer store.
The three parallel research streams aimed to:
-Develop a detailed but efficient functional model of human hearing in five stages: spectral analysis, neural transduction, phase allignment, pitch extraction and timbre stabilisation. Feature extraction mechanisms used in the model were to be developedin line with psychological research.
-Compare connectionist recogniser performance using auditory preprocessing with conventional spectrographic preprocessing.
-Develop new recogniser architectures and training algorithms in line with psycholinguistic research.
Work was also directed towards developing user documentation for the cochlea simulation software, establishing base-line performance for phoneme recognition using both traditional and auditory front-ends, feature extraction software, and an efficient parallel implementation of a connectionist recogniser. Phonological memory research, experimentation concerning human vowel recognition and human psycho-acoustical research are continuing parts of the Action.
PROGRESS AND RESULTS
At the present time, developed software and published research on the Action include:
-An efficient computer simulation of cochlea processing; documentation of the software for users in speech recognition and hearing research; published papers describing experiments on octave perception and phoneme recognition.
-Efficient implementation of a connectionist recogniser producing performance comparable to the best HMM recognisers; comparisons of the auditory model front-end with an FFT filterbank; improved training and pattern-recognition algorithms.
-Experimental research on the structure of human phonological memory.
POTENTIAL
This Action is laying the foundations for the construction of balanced and efficient auditory connectionist speech recognition systems that are founded on cognitive research in the areas of speech perception and memory.

Champ scientifique (EuroSciVoc)

CORDIS classe les projets avec EuroSciVoc, une taxonomie multilingue des domaines scientifiques, grâce à un processus semi-automatique basé sur des techniques TLN. Voir: Le vocabulaire scientifique européen.

Vous devez vous identifier ou vous inscrire pour utiliser cette fonction

Programme(s)

Programmes de financement pluriannuels qui définissent les priorités de l’UE en matière de recherche et d’innovation.

Thème(s)

Les appels à propositions sont divisés en thèmes. Un thème définit un sujet ou un domaine spécifique dans le cadre duquel les candidats peuvent soumettre des propositions. La description d’un thème comprend sa portée spécifique et l’impact attendu du projet financé.

Données non disponibles

Appel à propositions

Procédure par laquelle les candidats sont invités à soumettre des propositions de projet en vue de bénéficier d’un financement de l’UE.

Données non disponibles

Régime de financement

Régime de financement (ou «type d’action») à l’intérieur d’un programme présentant des caractéristiques communes. Le régime de financement précise le champ d’application de ce qui est financé, le taux de remboursement, les critères d’évaluation spécifiques pour bénéficier du financement et les formes simplifiées de couverture des coûts, telles que les montants forfaitaires.

Données non disponibles

Coordinateur

MRC Applied Psychology Unit
Contribution de l’UE
Aucune donnée
Coût total

Les coûts totaux encourus par l’organisation concernée pour participer au projet, y compris les coûts directs et indirects. Ce montant est un sous-ensemble du budget global du projet.

Aucune donnée

Participants (4)

Mon livret 0 0