SEE TRIGRAM LANGUAGE MODEL DATA SHEET

What is a Trigram Specialty Vocabulary?

The most widely used statistical model of language is the trigram model, in which an estimate of the likelihood of a word is made solely on the identity of the preceding two words in the utterance. The strengths of the trigram model come from its success at capturing local constraints, the ease by which it may be constructed from text corpora, and from its computational efficiency in use. Such a model has been at the heart of speech recognition systems since the pioneering work at IBM in the early 1980s. Trigram models have also been applied in optical character recognition applications, and in machine translation.

Now let's try this again in plain English. Speech recognition programs are really just playing the odds to predict what you actually said. They narrow down what you say based on the sounds you make and how high on the vocabulary scale the words are that match the sounds. A short list of possible words are then compared to a data set (the language model) and the speech program further narrows down what it thinks you meant using trigram's of based on context of the language model.

So, if you were a physician and dictated “Chest clear bilaterally” the recognition of the word “bilaterally” would be made based on the previous two words “chest clear” that is how trigram's work in speech recognition programs. So, the speech program is very dependent on the language model for accuracy. The speech program relies heavily on the language model for accuracy.

Creating language models and practicing medicine have one thing in common, experience. Trigram Technology has the real world experience of creating language models with every available speech engine since 1996. As in plastic surgery, for example, there is an “Art” to the creation of language models that shows in the end result.

Contact us for more information