ASR


 

AlfaNum ASR is a system for the recognition of continuous speech in Serbian and kindred languages. It is one of the basic software components that put the AlfaNum company ahead of its competition. ASR software has the following characteristics:

  • Recognition is speaker-independent
  • Accuracy is above 98% on a set of 2000 words
  • The size of vocabulary is up to several thousand words
  • Optimized for telephone quality recordings
  • The system provides information on reliability of recognition of individual words
  • The system works fast and reliably and, supports modern multi-core and multiprocessor platforms
  • The system is completely software-based and does not require any additional hardware
  • Several interfaces are supported: C++ library, COM, SAPI, MRCP, IP server
  • The system supports distribution over multiple computers and load balancing, which enables its application in very demanding environments

ASR software component and can be purchased separately by anyone who wishes to upgrade their system by adding the speech recognition feature. However, it can be built into a multitude of our stand-alone products and applications, such as Word Spotter.

We recommend you to take a look at the AlfaNum Speech Recognition System demo, or read more about practical implementation of ASR systems.

 

Young businessman speaking on a cell phone