Lingsoft® Speech Recognizer
|Lingsoft Speech Recognizer is a technology product for companies who wish to add speech recognition to their own applications. Speech Recognizer is available for Windows operating systems as a DLL-library with C language interface. Ask also for MRCP server and Linux-version of Speech Recognizer.|
Lingsoft Speech Recognizer is a phoneme-based speech recognition program, which is mainly intended for voice commands in telecom services and in PC software.
The speech recognition engine itself is language-independent, and the adjustment to the phonetics of each language is achieved with language-specific speech models. Currently available is the Finnish and Finland Swedish speech model. Ask for additional speech models.
Structure of Lingsoft's Speech Recognizer
Lingsoft Speech Recognizer inputs raw audio data (8 or 16 kHz, 16 bit linear PCM) which the hosting application retrieves from sound card, telephony card etc. Recognizer outputs result strings and confidence values.
The package also contains tools for developing vocabularies to recognize. With automatic tool it's possible to generate vocabularies from various data sources: databases, web pages, phone books etc. An application for fine tuning or building more complex grammars is also available.
Hardware requirementsMinimum requirements for Lingsoft Speech Recognizer are 20-50 Mb of RAM per recognition process and a 400 MHz processor for vocabulary size of 50 words and a 1,5 GHz for 1000 words. Thus with a 1000 word vocabulary a typical 3 Ghz processor can handle two simultaneous recognition and cover 15 telephone channels.
More informationYou can get more information by mailing us.
Copyright ©1986-2017, Lingsoft Ltd.