Robust Affordable Speech Recognition
Vocabulary Size
The Vestec Speech Engine has been designed for small vocabulary continuous speech applications. These comprise your typical “command-and-control” type IVR applications and constitute the vast majority of speech deployments today.
We offer two ASR models – Tier-1 and Tier-2 – that support vocabulary sizes of 500 and 2,500 items per recognition, respectively. There is no limit on the number of distinct grammars – each containing the maximum recognizable terms allowed by ASR model – that can be created for a single speech application.
Work is underway on development of large vocabulary engines.
NOTE: Vocabulary size refers to the number of distinct terms – whether keywords, names or digits – that can be recognized by the speech engine during a single speaker interaction. There can be dozens of distinct speaker interactions – each corresponding to a unique grammar, representing up to 500-2,500 distinct terms for recognition purposes – in a single application.