| Welcome to conversational computing! In the world of speech recognition technology, there are two basic types: Dictation Dictation uses very large speech engines that can intelligently convert sound signals into written text. This allows for arbitrary text to be spoken and recognized. Because the SR system has the difficult task of matching your phonemes to known words in the English language, each user typically must "train" the system to understand his/her unique way of speaking. The endless possible combination of words may result in misrecognitions. Command & Control Command & Control recognition will only recognize pre-defined words and phrases, otherwise known as "language models". While this presents some obvious limitations, it also tends to be more accurate and does not require user voice training. Since the user's utterances are restricted to a known list, there is less room for misrecognition errors. SUI Pack uses PlainTalk speech recognition, which is Command & Control. SUI Pack helps you manage the task of defining language models, which can be as simple as a list of ten colors, or they may contain "embedded language models", which SUI Pack supports: <TopLM> = Send <manager> a copy of the <reportname> <manager> = Jane | John | Joe | Kathy | Mary | Robert <reportname> = Sales Summary | Year End Report SUI Pack uniquely offers complex language models for all major data types. There are built-in models for numeric, date and time values. For example, if you say "June fifth two thousand one" your date field would be populated with 06/05/2001. This is an advanced implementation of Command & Control! Note: The term "language model" is often abbreviated in this manual as "LM". | ||