Passport of joint Russian-U. S. project

I-11-CITIS-1-eng

Passport of joint Russian-U. S. project

1. Title

Technology and system for recognition and understanding of speech in any national language.

2. Abstract

At present on the world market there are no technologies and systems for recognition and understanding of speech in any national language that can ensure adequate conversion of speech into text information. Speech is identified disregarding the sense of the spoken communication.

Therefore, the niche of systems for speech recognition and understanding on the world IT market still remains vacant.

The developed products are designed for coping with the much greater range of problems compared to the available most well-known products. The developed technology permits to make changes in the man-equipment concept, to create new generations of voice-controlled facilities and also to develop systems for reliable search, analysis and translation of the information into various national languages.

This will ensure the technological leadership of the project participants in this research area, will facilitate consolidation of not only scientific-technical partnership, but development of the informational interstate interaction of the project members countries.

The analogs of the technologies and systems with the announced characteristics are not available on the world market, thus, it makes impossible the purchase of foreign technologies or licenses due to their absence.

3. Description of the expected project results

The product is named – technology and system for recognition and understanding of speech in any national language.

Kinds of the developed products are as follows:

- technology for recognition and understanding of speech of a certain language group;

- commercial versions of the system for recognition and understanding of speech designed for voice input and automatic representation of the information in the form of a sense-based marked text;

- internet portal providing services for voice interface of users with other information services: information search, translation of verbal messages from one national language into other, etc.;

- software development kits (SDK) for linguistic systems designed for realization of the speech recognition functions while creating outside systems for verbal information interpretation.

The developed products are designed for resolving the much greater range of problems compared to the available most well-known products. The developed technology permits to create systems for recognition and understanding of speech in any national language, thus, providing for realization of the principally new concept of the man-equipment interaction.

4. Novelty, description of competitive advantages of the developed technology

The most close to the developed technology are US corporations Microsoft and Google, but they do not have such products.

Specialists of the Federal State Research Establishment “Center for Information Technologies and Systems of Executive Power Bodies” are the authors of patents of RF, USA, China that will be used for protection of intellectual property in this project.

The rivals cannot manufacture similar products without breaching the patent rights of a contractor organization.

Technologically similar works are not conducted on the world IT market. The works on traditional technologies of speech recognition are carried out mostly within R&D and experimental developments of new approaches potentially capable to get over technological constraints of the well-known speech recognition technique.

The specific features of the technology developed in this project are principally new methods to ensure high quality of speech recognition and also understanding of the sense of a verbal communication by computers. Such methods provide higher technological indicators unattainable with traditional methods. Unlike the earlier applied approaches, the developed methods are based on the use of semantic-acoustic classification of phonemes, syllables and words and also on the full linguistic analysis of recognized messages.

5. Expected effect of the project results

Commercialization of this project is capable to improve significantly the efficacy of industrial production in general.

The results of this project will influence both the structure and efficiency of industrial estimates of experts and analytical agencies, the profit from application of new technologies of speech recognition already in the first year of project commercialization will exceed $150 million. Therefore, by the end of the first year of project commercialization the received profit will exceed more than five-fold the investments into this project. This will permit to increase the product sales and to reach the potential sales volume estimated on the basis of the existing demand on the world market for speech recognition systems which volume makes at present about $38 billion. In the future the potential world market of the speech recognition systems will be more than $50 billion due to widened applications of this technology and systems.

6. Proposed participants of the project consortium (research, education, commercial)

The initiator of this project is Federal State Research Establishment “Center for Information Technologies and Systems of Executive Power Bodies” (Russian Federation), Director Starovoitov Alexander Vladimirovich.

Among other participants of this project are the following research, educational and commercial organizations:

- International Center for Informatics and Electronics “InterEVM”, Director General Starovoitov Alexander Vladimirovich;

- Philological Department of the M. V. Lomonosov Moscow State University (Russian Federation.

Participation of various companies and their close interaction make it possible to realize the project potential by integrating the efforts of science, education and business.

7. Preliminary plan of project preparation and performance, main project milestones

In 2011 it is planned to perform outline and engineering design as well as preparation of working documentation for a trial specimen.

In 2012 it is planned to develop a trial system for recognition and understanding of speech of a given language group, to prepare a system of dictionaries for different languages and localization of system.

In the second half of 2012 it is planned to create and conduct preliminary and acceptance tests of the trial system for recognition and understanding of speech of a given language group and also to develop SDK and Internet portal.