Methods and apparatus for natural spoken language speech recognition
A word prediction apparatus and method that improves the precision accuracy, and a speech recognition method and an apparatus therefor are provided. For the prediction of a sixth word “?”, a partial analysis tree having a modification relationship with the sixth word is predicted. “sara-ni sho-senkyoku no” has two partial analysis trees, “sara-ni” and “sho-senkyoku no”. It is predicted that “sara-ni” does not have a modification relationship with the sixth word, and that “sho-senkyoku no” does. Then, “donyu”, which is the sixth word from “sho-senkyoku no”, is predicted. In this example, since “sara-ni” is not useful information for the prediction of “donyu”, it is preferable that “donyu” be predicted only by “sho-senkyoku no”.
1. A speech recognition apparatus comprising:
a combination of hardware and software configured to implement:
an acoustic processor that converts an input analog speech signal into a digital signal;
at least one memory that stores an acoustic model and a dictionary, the dictionary indicating appearance frequencies of words relative to other words and/or word sequences; and
a recognizer that predicts a first word from the digital based on at least one other word and/or word sequence in a phrase recognized from the digital signal, wherein the recognizer calculates a probability value for the phrase including the first word using an appearance frequency indicated by the dictionary of the first word relative to the at least one other word and/or word sequence, wherein the recognizer does not predict the first word based on a second word that immediately precedes the first word in the phrase unless the second word belongs to a partial analysis tree that grammatically modifies or is grammatically modified by the first word in a sentence structure of the phrase.
2. The speech recognition apparatus according to claim 1 , further comprising:
an arrangement of hardware and software configured to return at least said first word to a user as a recognition result.
3. The speech recognition apparatus according to claim 1 , wherein the at least one other word and/or word sequence comprises a partial analysis tree.
4. The speech recognition apparatus according to claim 1 , wherein the at least one other word and/or word sequence grammatically modifies or is grammatically modified by the first word in the sentence structure.
5. The speech recognition apparatus according to claim 1 , wherein the recognizer predicts a word for each of multiple potential sentence structures for the phrase.
6. The speech recognition apparatus according to claim 4 , wherein said recognizer specifies a modification direction between the first word and the at least one other word and/or word sequence in the sentence structure.
7. The speech recognition apparatus according to claim 2 , wherein the arrangement of hardware and software is configured to return an entire sentence to said user when the recognizer predicts the last word of the sentence.
8. The speech recognition apparatus according to claim 2 , further comprising:
a storage medium configured to store said recognition result.
9. A speech recognition method comprising:
receiving an input speech signal; and
predicting a first word from the input speech signal based on at least one other word and/or word sequence in a phrase recognized from the input speech signal, wherein the first word is not predicted based on a second word that immediately precedes the first word in the phrase unless the second word belongs to a partial analysis tree that grammatically modifies or is grammatically modified by the first word in a sentence structure of the phrase.
10. The method according to claim 9 , further comprising:
returning at least said first word to a user as a recognition result.
11. The method according to claim 9 , wherein the at least one other word and/or word sequence comprises a partial analysis tree.
12. The method according to claim 9 , wherein the at least one other word and/or word sequence grammatically modifies or is grammatically modified by the first word in the sentence structure.
13. The method according to claim 9 , further comprising predicting a word for each of multiple potential sentence structures for the phrase.
14. The method according to claim 12 , further comprising:
specifying a modification direction between the first word and the at least one other word and/or word sequence in the sentence structure.
15. The method according to claim 10 , further comprising:
returning an entire sentence to said user when the last word of the sentence is predicted.
16. The method according to claim 10 , further comprising:
storing said recognition result in a memory.
17. A program storage device readable by computer, the program storage device tangibly embodying a program of instructions executable by the computer to perform a method comprising:
receiving an input speech signal; and
predicting a first word from the input speech signal based on at least one other word and/or word sequence in a phrase recognized from the input speech signal, wherein the first word is not predicted based on a second word that immediately precedes the first word in the phrase unless the second word belongs to a partial analysis tree that grammatically modifies or is grammatically modified by the first word in a sentence structure of the phrase.
18. The method according to claim 9 , further comprising calculating a probability value for the phrase including the first word, based on an appearance frequency of the first word relative to the at least one other word and/or word sequence.