Résumé des résultats (Abstract)
(Anglais)
|
INSPIRE will combine leading-edge technology & experience gained from previous projects to provide natural spoken dialogue interaction to infotainment (information & entertainment) devices installed in a smart home environment. It will use wireless hardware for I/O, robust speech understanding, room acoustics signal processing, speaker recognition and advanced personalization features. The prototype will be able to handle sophisticated mixed initiative dialogue, it will support multilinguality, it will facilitate communication with external information sources and it will be highly user-friendly. A strong user-oriented acceptability and assessment component will ensure its efficient use especially by elderly, disabled and not-technically inclined persons, in home, work or on the move.
Objectives:
INSPIRE has the following objectives:
a) to combine leading-edge speech technology in order to provide user-friendly access to information and entertainment devices installed in a smart home environment;
b) offer a portal to complex devices to not technically-inclined or disabled users, in home, workplaces etc.;
c) to set up a prototype voice-activated assistant which achieves user friendliness through robust, unconstraint natural dialogue, and which is flexible for other application scenarios, technology advancement, and multi-language adaptation;
d) to make use of human-factor oriented design and evaluation procedures which will feed back on the technology development and scientific know-how;
e) to provide a significant lead for industry partners in order to rapidly develop and adapt systems which will successful on the market, and accepted by the users.
Work description:
The work is subdivided into 8 work packages: WP0 incorporates all management tasks needed to keep the project up and running. At WP1 the project requirements will be determined, regarding user needs, wireless communication technology, hardware specifications, technical architecture selected, scientific contents, system functionality and usage scenarios. WP2 (Room & Speaker Perspectives) aims at the establishment of sites for recordings and tests, the hardware installation and the system features related to the signal processing part of the application. WP3 (Remote Access Simulation System) provides the hardware components for realistically simulating speech transmission channels and the user interfaces, which are used for the remote access. During WP4 (Dialogue Interaction), the dialogue component will be set up. WoZ experiments will be conducted to determine the dialogue flow for each language (Greek, German). A dialogue simulation tool will be developed and additional user modelling functionalities will be incorporated. WP5 focuses on the integration of the components built during the previous WPs to a whole prototype. Due to the complexity of the whole system and the uniformity of its components several test-and-correction cycles will be performed until acceptable performance is obtained. During WP6 (System Assessment, Usability & Acceptability Evaluation), diagnostic assessment of the performance will help to select the best-suited components and algorithms, and to identify system weaknesses and problems. Acceptability evaluation will be performed with pre-selected groups of potential users, in order to provide information on the fitness of the final system. Finally, at WP7, potential application, marketing and dissemination policies will be identified, in order to inform interested parties about the achievements of the project and ensure its commercial viability.
Milestones:
At M0+6 the test sites will have been established. At M0+12 the remote access simulation system will be available. At M0+18 the dialogue component will be available. At T0+21 the system prototype will be ready. Finally, at M0+30 all the assessment, usability and acceptability tests will have been completed and the market analysis, exploitation, dissemination and concentration reports will be ready.
|
