Multi-phoneme streamer and knowledge representation speech recognition system and method
A new approach to speech recognition that reacts to concepts conveyed through speech, which shifts the balance of power in speech recognition from straight sound recognition and statistical models to a more powerful and complete approach determining and addressing conveyed concepts. A probabilistically unbiased multi-phoneme recognition process is employed, followed by a phoneme stream analysis process that builds the list of candidate words derived from recognized phonemes, followed by a permutation analysis process that produces sequences of candidate words with high potential of being syntactically valid, and finally, by processing targeted syntactic sequences in a conceptual analysis process to generate the utterance's conceptual representation that can be used to produce an adequate response. Applications include improving accuracy or automatically generating punctuation for transcription and dictation, word or concept spotting in audio streams, concept spotting in electronic text, customer support, call routing and other command/response scenarios.
1. A method for processing speech, comprising:
receiving an input representing a stream of semantic information;
permuting a set of candidate parts of speech associated with elements of the stream of semantic information, the permuting resulting in at least two alternates;
determining at least one potentially valid syntactic structure from the permuted set of candidate parts of speech; and
storing a conceptual representation of the at least one potentially valid syntactic structure substantially independent of a syntax of the input.
2. The method of claim 1 , wherein the permuting is implemented as at least one of: a bottom-up parsing process, a top-down parsing process, an Earley parsing process, a finite-state parsing process, and a CYK parsing process.
3. The method of claim 1 , further comprising applying at least one anomaly criterion to the at least one potentially valid syntactic structure.
4. The method of claim 1 , further comprising generating a representation of a conceptually appropriate response to the input based on at least the stored conceptual representation.
5. The method of claim 1 , wherein the input comprises an audio signal representing speech.
6. The method of claim 1 , wherein the input comprises a text message.
7. The method of claim 1 , further comprising applying a set of semantic rules and a dictionary during said determining.
8. A method for processing a semantic stream, comprising:
receiving a semantic information stream comprising words in a syntax;
generating a set of candidate parts of speech for each word;
permuting at least a portion of the candidate parts of speech to produce a plurality of syntactically valid structures;
disambiguating at least two potentially syntactically valid structures based on the corresponding conceptual representations and a context; and
storing a conceptual representation of the semantic information stream, substantially independent of the syntax, in dependence on said disambiguated at least two potentially syntactically valid structures.
9. The method of claim 8 , wherein the permuting is implemented by at least one of a bottom-up parsing process, a top-down parsing process, an Earley parsing process, a finite-state parsing process, and a CYK parsing process.
10. The method of claim 8 , further comprising using the conceptual representation to formulate a response to an inquiry.
11. The method of claim 8 , further comprising applying a set of semantic rules, wherein the set of semantic rules comprises conceptual relationships between words.
12. The method of claim 11 , wherein the set of semantic rules comprises an interpreted language.
13. The method of claim 11 , wherein the set of semantic rules comprises a predicate builder scripting language.
14. The method of claim 11 , wherein the set of semantic rules comprises a compiled language.
15. A semantic information processing system, comprising:
a memory; and
a processor configured to:
receive semantic information having a syntax;
identify words within the received semantic information;
generate a set of candidate parts of speech from a sequence of identified words;
produce a plurality of potentially valid syntactic structures from a plurality of permutations of the candidate parts of speech; and
store a syntax-free conceptual representation of at least one of the potentially valid syntactic structures.
16. The system of claim 15 , wherein the plurality of permutations are generated by at least one of a bottom-up parsing process, a top-down parsing process, an Earley parsing process, a finite-state parsing process, and a CYK parsing process.
17. The system of claim 15 , wherein the processor is further configured to formulate a response to the received semantic information based on the conceptual representation.
18. The system of claim 15 , wherein the processor is further configured to retrieve a dictionary and a set of semantic rules.
19. The system of claim 15 , wherein the processor is further configured to determine at least one anomaly criterion for at least one permutation, and to disambiguate between a plurality of permutations in dependence on the determined at least one anomaly criterion.
20. The system of claim 15 , wherein the processor is further configured to apply a Hidden Markov Model (HMM) technique to the received semantic information.