Objective
The objectives of the HARP project are to prove that technology already developed under ESPRIT project SPELL for foreign language learning can be modified for use in teaching pronunciation in the speech rehabilitation of hearing impaired people.
HARP is a two year project. The work of HARP uses as its point of departure, the results from ESPRIT project 7153 (SPELL) which has developed courseware for teaching pronunciation to foreign language learners. This courseware uses an IBM PC-compatible computer with a commercial speech processing card and a well-proven user interface based on Microsoft Windows. Direct extension of this software to address the needs of hearing impaired people represents the key activity of the HARP project. The project will focus on the introduction of the HARP system into schools and clinics during the first year such that extensive trials can be undertaken to prove the system and make it ready for general market release by the end of the second year. A near-market product will be ready at the end of the second year.
The innovative aspects of HARP derive from the use of hidden Markov speech models which are tailored to the speech distortions of hearing impaired talkers; from the use of intonation, rhythm and speech quality models for autonomous teaching of nearly normal pronunciation to hearing impaired people; and from the creation of special graphical user interfaces for use by the hearing impaired such that the system can be widely introduced into schools and homes. One of the hallmarks of the HARP project will be the involvement of hearing impaired users form an early stage. These users will be made available by Sponsoring Partners who can give access to subjects.
Technical Approach : The project objective is to introduce computer-based systems to schools, clinics and peoples homes which will analyse the speech characteristics of speakers with hearing impairment and improve their spoken language skills.
A total of four cohorts of patients will be addressed. These groups comprise
(a) young children with pre-lingual hearing impairment;
(b) young persons who have post-lingual hearing impairment and remnants of speech skills;
(c) persons who have undergone surgery for cochlear implant;
(d) elderly persons who have developed serious hearing impairments.
The project will develop PC-based software and courseware which will be inexpensive and suitable for use in clinics and in the homes of hearing impaired people. In this project, research topics will include phonetic analysis, phonetic distance metrics, multilingual systems, linguistics and computer aided instruction all with specific reference to the detailed needs and requirements of the four cohorts of hearing impaired subjects as prescribed by the Sponsoring Partners. The languages to be studied on the project are English and French, to be analysed in order to extract their phonetic characteristics. The methods developed on the project will allow easy extension to other languages and the needs of porting the system to other languages will be assessed throughout the work. The final demonstrator will be a system able to process the voice of hearing impaired speakers in order to automatically identify and give advice on pronunciation errors thereby providing rehabilitation assistance.
Impact and Expected Results : The technological objectives of the HARP project are to develop methods for analysing the characteristics of the speech of hearing impaired speakers, to develop metrics for identifying differences between the users pronunciation and the optimal one and to provide feedback to the user so that they can improve their speaking skills. An improvement with respect to the state of the art will be necessary especially in the area of speech parameter extraction for speech from the hearing impaired.
Programme(s)
Call for proposal
Data not availableFunding Scheme
CSC - Cost-sharing contractsCoordinator
38640 Claix
France