Phrase-based dialogue modeling with particular application to creating recognition grammars for voice-controlled user interfaces
The invention enables creation of grammar networks that can regulate, control, and define the content and scope of human-machine interaction in natural language voice user interfaces (NLVUI). More specifically, the invention concerns a phrase-based modeling of generic structures of verbal interaction and use of these models for the purpose of automating part of the design of such grammar networks.
1. A speech recognition computer system comprising:
an acoustic signal analyzer that generates acoustic features from a digital speech signal representing spoken input;
a first recognition grammar based on a first language;
a second recognition grammar based on a second language;
wherein either one or both of the recognition grammars are coupled to a probabilistic phrase thesaurus including a data structure of phrases classified according to at least semantic similarity, and wherein either one or both of the recognition grammars specify legitimate word sequences representing conceptual meaning from alternative phrase variants of a language;
a phonetic dictionary having phonetic transcriptions of acoustic features; and
an acoustic decoder coupled with the acoustic signal analyzer, the recognition grammars, and the phonetic dictionary and that:
computes a set of most probable word hypotheses based on the acoustic features from the digital speech signal and the phonetic transcriptions in the phonetic dictionary;
constrains the word hypotheses to a narrowed set of commands according to the conceptual meaning of the legitimate word sequences from either one or both of the recognition grammars; and
provides the narrowed set of commands to a natural language processing component that translates the narrowed set of commands to a formalized set of instructions that can be processed by an application.
2. The system of claim 1 , wherein the acoustic signal analyzer is configured to obtain the digital speech signal from a microphone.
3. The system of claim 1 , wherein the acoustic signal analysis is configured to obtain a speech signal from a telephone.
4. The system of claim 1 , further comprising a speech recognition front end that comprises the acoustic signal analyzer, the acoustic decoder, the recognition grammars, and the phonetic dictionary.
5. The system of claim 1 , wherein the legitimate word sequences of either one or both of the recognition grammars comprise a set of legitimate variations.
6. The system of claim 1 , wherein the legitimate word sequences comprise a total set of user commands that are expected for a context.
7. The system of claim 1 , wherein either one or both of the recognition grammars comprise a recognition grammar network.
8. The system of claim 1 , wherein the system comprises at least two recognition grammars.
9. The system of claim 1 , wherein either one or both of the recognition grammars model interactions across different contexts.
10. The system of claim 1 , wherein either one or both of the recognition grammars model interactions across different domains.
11. The system of claim 1 , wherein the acoustic decoder is further configured to trigger a request for clarification from a user for a command not specified within a recognition grammar.
12. The system of claim 1 , wherein the narrowed set of commands includes text commands.
13. The system of claim 1 , further comprising the natural language processing component.
14. The system of claim 13 , further comprising a dialog manager that receives the narrowed set of commands from the natural language processing component and generates a response.
15. The system of claim 14 , further comprising a speech output back end that generates a verbal system response based on the response from the dialogue manager.
16. The system of claim 15 , wherein the verbal response comprises a request for clarification.
17. The system of claim 14 , wherein the dialogue manager is further configured to generate the response as an action to be performed by a device.
18. The system of claim 17 , wherein the action to be performed by the device includes accessing a database.
19. The system of claim 17 , wherein the action to be performed by the device includes retrieving a piece of information.
20. The system of claim 1 wherein said probabilistic phrase thesaurus contains pre-tagged descriptors for each phrase classification.
21. A non-transitory computer readable storage medium on which are stored software instructions executable by a processor to perform operations for speech recognition, the operations comprising:
generating acoustic features from a digital speech signal representing spoken input;
computing a set of most probable word hypotheses based on the acoustic features from the digital speech signal and phonetic transcriptions of the acoustic features derived from a phonetic dictionary;
constraining the word hypotheses to a narrowed set of commands according to a conceptual meaning represented by a legitimate word sequence, the legitimate word sequence specified from alternate phrase variants of a language by either one or both of a first recognition grammar based on a first language and a second recognition grammar based on a second language, the either one or both of the first and second recognition grammars being coupled to a probabilistic phrase thesaurus including a data structure of phrases classified according to at least semantic similarity; and
providing the narrowed set of commands to a natural language processing component that translates the narrowed set of commands to a formalized set of instructions that can be processed by an application.
22. A method of speech recognition using at least one processor and at least one non-transitory computer readable storage medium, the method comprising:
generating acoustic features from a digital speech signal representing spoken input;
computing a set of most probable word hypotheses based on the acoustic features from the digital speech signal and phonetic transcriptions of the acoustic feature derived from a phonetic dictionary;
constraining the word hypotheses to a narrowed set of commands according to a conceptual meaning represented by a legitimate word sequence, the legitimate word sequence specified from alternate phrase variants of a language by either one or both of a first recognition grammar based on a first language and a second recognition grammar based on a second language, the either one or both of the first and second recognition grammars being coupled to a probabilistic phrase thesaurus including a data structure of phrases classified according to at least semantic similarity; and
providing the narrowed set of commands to a natural language processing component that translates the narrowed set of commands to a formalized set of instructions that can be processed by an application.