Honeywell Adaptive Speech Recognition Feature

The Honeywell Adaptive Speech Recognition feature is available in both BlueStreak and PnG speech recognition. BlueStreak adaptation is offered in all versions of VoiceCatalyst. PnG adaptation was introduced in A700x/VoiceCatalyst 4.7.

Adaptive Speech Recognition means that as the worker uses the system over time, the recognizer uses samples of his or her speech in different environments, at different volumes, and so on to automatically update the worker's templates to account for these changes in speaking patterns. This feature allows recognition to increase the more the worker uses the system.

Note that in order for this feature to work, the provider of your Honeywell Voice system must include special recognizer hints called response expressions when developing the application. Only words used in response expressions automatically adapt.

There is a limit to how far from the original template the recognizer can automatically adapt. If, after an extended period of usage, a worker is repeatedly having problems with a word not being recognized, he or she should retrain that word.

For PnG, for words that are used in response expressions (typically digits and in some apps, some other words), waiting for adaptation to occur may yield better results than retraining, as adaptation relies on a larger sample of examples before adding pronunciations.