Speech Recognition

Speech Recognition (Automatic Speech Recognition ASR) is a procedure that processes speech and determines what was said. Speech recognizers suitable for computer dialog systems (IVR) are speaker-independent, i.e. no previous training is required for a particular user's voice.

Besides the speech itself, this type of speech recognizer takes so called grammar as an input. The output is then a text representation of the speech and a representation of its meaning. Grammar limits the set of utterances that can be recognized by the system, making this type of speech recognition different from the speech-to-text transcription. The speech-to-text transcription recognizes "anything" but with a significantly higher error rate and with no possibility to assign the meaning representation to the utterance.

Thanks to its own implementation of an MRCPv2 client protocol, OptimSys' products support ASR systems of all major vendors covering over 50 languages and dialects as well as systems of some local vendors specialized in less common languages.