Sirko: automatic speech recognition

The innovation enterprise "Announced the systems of speech communications in April, 2011 commercial availability of the latest technology of automated speech recognition. It is a result of long-term work of group of scientists of the Nizhny Novgorod state linguistic university.

The new method of phonetic decoding of words which allowed to achieve powerful benefits in comparison with existing solutions in the field of speech recognition is the cornerstone of technology. Here some of key features of new development:

• Real-time operation with the dictionary in several thousands of words.
• Minimum need for computing resources. For the dictionary of 1000 words the prize makes more than 200 times in comparison with analogs.
• Automatic procedure of formation and updating of the working phonetic dictionary from the topical written text of any volume. For replenishment of the dictionary on 1000 words only 3 minutes there are enough!
• Possibility of the choice of national language from the wide list.
• Independence of speech recognition of specific features and age of the announcer.

The greatest problem for all known technologies of automatic speech recognition is formations of the working dictionary. Adding or change in the working dictionary even of one position turns into a difficult and labor-consuming procedure. The technology developed by the Nizhny Novgorod scientists provides growth of efficiency on several orders.

"Show characteristics of our technology to any specialist in the field of speech recognition, and he will tell you that it is impossible." – the head of group of developers, the Doctor of Engineering, professor Vladimir Savchenko says. – "Nobody to us implemented the procedure of automatic formation of the sound working dictionary from the written text as it is well known that the oral and written language is, in fact, two different languages. We in the project reached impossible, and that says it all!"

The provided technology can be used in a broad spectrum of applications. There are some examples:

• Voice information search in thematic databases via the Internet and telephone networks
• Systems of voice actuation
• Directory services with hands-off processing of requests by phone
• Systems of automatic secretaries and system of automatic speech translation
• Online stores and the trading systems of orders by phone.

Thanks to procedures of automatic formation and updating of the working dictionary, the new technology offers wide prospects of work with any databases with constantly changing contents. It can be, for example, directories of Online stores, information bases of contact centers, holding sociological polls and many other things. Independence of speech recognition of specific features and age of the announcer allows to apply new technology in different solutions for mass audience.