Back to results
US 11,145,296 B1Granted
Language and grammar model adaptation
AssigneeAmazon Technologies, Inc.
InventorsAnkur Gandhe, Ariya Rastrow, Gautam Tiwari, Ashish Vishwanath Shenoy, Chun Chen
Life of this patent
Estimating the term…
Abstract
Systems and methods described herein relate to adapting a language model for automatic speech recognition (ASR) for a new set of words. Instead of retraining the ASR models, language models and grammar models, the system only modifies one grammar model and ensures its compatibility with the existing models in the ASR system.
Claims20
A computer-implemented method comprising:
- receiving input data representing a first set of words;
- determining a first grammar model corresponding to a second set of words that are different than the first set of words, the first grammar model including first model weight data;
- processing the first grammar model using the input data to determine a rescoring model including second model weight data;
- determining a second grammar model based at least in part on a difference between the first model weight data and the second model weight data;
- generating a first finite state transducer (FST) corresponding to a third set of words, wherein at least one word of the third set of words is included in the first set of words but not in the second set of words;
- and
- storing a link to the first FST as associated with the first grammar model.
4 more elements
A system comprising:
- at least one processor;
- and
- at least one memory including instructions that, when executed by the at least one processor, cause the system to:
- receive input data representing a first set of words;
- determine a first grammar model corresponding to a second set of words that are different than the first set of words, the first grammar model including first model weight data;
- process the first grammar model using the input data to determine a rescoring model including second model weight data;
- determine a second grammar model based at least in part on a difference between the first model weight data and the second model weight data;
- generate a first finite state transducer (FST) corresponding to a third set of words, wherein at least one word of the third set of words is included in the first set of words but not in the second set of words;
- and
- store a link to the first FST as associated with the first grammar model.
7 more elements
Dependencies are inferred from the claim text.