Does Chant SpeechKit support voice dependent programming?

Last reviewed: 11/18/1999

FAQ Article ID: F119902

Speaker Dependent Speech Recognition

Voice dependency is an attribute of the recognizer. There are different training facilities provided in SAPI that the recognizer may or may not support.

The simplest is word training. Most recognizers support this wizard API defined by SAPI. The words are stored in a speaker lexicon managed by the recognizer.

The next level is speaker training. SAPI provides API functions to develop an application to create/edit training profiles for speakers. MS recognizers do not provide this, however, you can use training utility in MS Dictate to create speaker profiles for the SAPI 4 recognizers. Vendors such as IBM, Dragon, and L&H provide speaker training utilities for their recognizers.

You can set the speaker for a session using the Chant components. You can invoke word training dialog through Chant components. You must use the recognizer utilities or write your own directly against SAPI if you want to do Speaker training. There is no facility for this in Chant components since it is very recognizer dependent and provided with the recognizer.