FLEXIBLE SCHEMA FOR LANGUAGE MODEL CUSTOMIZATION

摘要

Customization of language modeling components for speech recognition is provided. The list of language modeling components may be made available by the computing device. The hint may then be sent to the recognition service provider to combine the plurality of language modeling components from the list. Hints may be based on a variety of different domains. A customized combination of language modeling components based on hints may then be received from the recognition service provider.