Recognizing the numeric language in natural spoken dialogue
View Patent ↗A system and a method are provided. A speech recognition processor receives unconstrained input speech and outputs a string of words. The speech recognition processor is based on a numeric language that represents a subset of a vocabulary. The subset includes a set of words identified as being for interpreting and understanding number strings. A numeric understanding processor contains classes of rules for converting the string of words into a sequence of digits. The speech recognition processor utilizes an acoustic model database. A validation database stores a set of valid sequences of digits. A string validation processor outputs validity information based on a comparison of a sequence of digits output by the numeric understanding processor with valid sequences of digits in the validation database.
1. A method comprising:
receiving unconstrained input speech from a user;
converting, via a processor, only the unconstrained input speech corresponding to single digits into a string of words, wherein each word in the string of words is modeled using a three segment structure comprising a plurality of heads and a plurality of tails;
converting the string of words into a sequence of digits using classes of rules and according to an acoustic model database in which Markov models characterize acoustic features of numeric words;
comparing the sequence of digits to a plurality of valid sequences of digits, to yield validity information; and
providing the validity information to a device associated with the user.
2. The method of claim 1 , wherein the plurality of valid sequences of digits comprise one of valid credit card numbers, valid telephone numbers, and valid social security numbers.
3. The method of claim 1 , wherein the three segment structure further comprises a body.
4. The method of claim 1 , wherein converting the unconstrained input speech into the string of words further comprises:
using a language model database that stores data describing a structure and a sequence of words and phrases.
5. The method of claim 1 , wherein the Markov models are applied left-to-right with no skip states.
6. The method of claim 1 , wherein the classes of rules varies depending upon one of a city and a country.
7. A system comprising:
a processor; and
a computer-readable storage device having instructions stored which, when executed by the processor, cause the processor to perform operations comprising:
receiving unconstrained input speech from a user;
converting, via a processor, only the unconstrained input speech corresponding to single digits into a string of words, wherein each word in the string of words is modeled using a three segment structure comprising a plurality of heads and a plurality of tails;
converting the string of words into a sequence of digits using classes of rules and according to an acoustic model database in which Markov models characterize acoustic features of numeric words;
comparing the sequence of digits to a plurality of valid sequences of digits, to yield validity information; and
providing the validity information to a device associated with the user.
8. The system of claim 7 , wherein the plurality of valid sequences of digits comprise one of valid credit card numbers, valid telephone numbers, and valid social security numbers.
9. The system of claim 7 , wherein the three segment structure further comprises a body.
10. The system of claim 7 , wherein converting the unconstrained input speech into the string of words further comprises:
using a language model database that stores data describing a structure and a sequence of words and phrases.
11. The system of claim 7 , wherein the Markov models are applied left-to-right with no skip states.
12. The system of claim 7 , wherein the classes of rules varies depending upon one of a city and a country.
13. A computer-readable storage device having instructions stored which, when executed by a computing device, cause the computing device to perform operations comprising:
receiving unconstrained input speech from a user;
converting, via a processor, only the unconstrained input speech corresponding to single digits into a string of words, wherein each word in the string of words is modeled using a three segment structure comprising a plurality of heads and a plurality of tails;
converting the string of words into a sequence of digits using classes of rules and according to an acoustic model database in which Markov models characterize acoustic features of numeric words;
comparing the sequence of digits to a plurality of valid sequences of digits, to yield validity information; and
providing the validity information to a device associated with the user.
14. The computer-readable storage device of claim 13 , wherein the plurality of valid sequences of digits comprise one of valid credit card numbers, valid telephone numbers, and valid social security numbers.
15. The computer-readable storage device of claim 13 , wherein the three segment structure further comprises a body.
16. The computer-readable storage device of claim 13 , the computer-readable storage device having additional instructions stored which, when executed by the computing device, result in the operations further comprising:
using a language model database that stores data describing a structure and a sequence of words and phrases.
17. The computer-readable storage device of claim 13 , wherein the Markov models are applied left-to-right with no skip states.