System and method for generating user models from transcribed dialogs
View Patent ↗Disclosed herein are systems, computer-implemented methods, and computer-readable storage media for generating personalized user models. The method includes receiving automatic speech recognition (ASR) output of speech interactions with a user, receiving an ASR transcription error model characterizing how ASR transcription errors are made, generating guesses of a true transcription and a user model via an expectation maximization (EM) algorithm based on the error model and the respective ASR output where the guesses will converge to a personalized user model which maximizes the likelihood of the ASR output. The ASR output can be unlabeled. The method can include casting speech interactions as a dynamic Bayesian network with four variables: (s), (u), (r), (m), and encoding relationships between (s), (u), (r), (m) as conditional probability tables. At each dialog turn (r) and (m) are known and (s) and (u) are hidden.
1. A method comprising:
receiving speech recognition output associated with speech from a user;
receiving an error model characterizing how automatic speech recognition transcription errors are made;
generating, via a processor, guesses of a true transcription based on an error model-based algorithm, wherein the error model-based algorithm uses the error model and the speech recognition output; and
generating, based on the guesses of the true transcription, a personalized user model associated with a user voiceprint of the user, wherein the generating of the personalized user model comprises iteratively guessing the true transcription until a threshold is met.
2. The method of claim 1 , wherein generating of the guesses further comprises repeating, until a threshold is met, steps comprising:
guessing the true transcription from a current guess of the personalized user model, to yield a current guess of the true transcription; and
guessing the personalized user model based on the current guess of the true transcription.
3. The method of claim 1 , wherein the error model-based algorithm estimates conditional probabilities of hidden variables.
4. The method of claim 1 , wherein generating of the guesses is further based on a set of manual transcriptions of speech interactions with the user.
5. The method of claim 1 , wherein the personalized user model comprises a Bayesian network.
6. The method of claim 1 , further comprising generating a personalized automatic speech recognition model based on the personalized user model.
7. The method of claim 1 , further comprising recognizing additional speech from the user using the personalized user model.
8. The method of claim 7 , further comprising iteratively improving the personalized user model based on the additional speech.
9. A system comprising:
a processor; and
a computer-readable storage medium having instructions stored which, when executed by the processor, cause the processor to perform operations comprising:
receiving speech recognition output associated with speech from a user;
receiving an error model characterizing how automatic speech recognition transcription errors are made;
generating guesses of a true transcription based on an error model-based algorithm, wherein the error model-based algorithm uses the error model and the speech recognition output; and
generating, based on the guesses of the true transcription, a personalized user model associated with a user voiceprint of the user, wherein the generating of the personalized user model comprises iteratively guessing the true transcription until a threshold is met.
10. The system of claim 9 , wherein generating of the guesses further comprises repeating, until a threshold is met, steps comprising:
guessing the true transcription from a current guess of the personalized user model, to yield a current guess of the true transcription; and
guessing the personalized user model based on the current guess of the true transcription.
11. The system of claim 9 , wherein the error model-based algorithm estimates conditional probabilities of hidden variables.
12. The system of claim 9 , wherein generating of the guesses is further based on a set of manual transcriptions of speech interactions with the user.
13. The system of claim 9 , wherein the personalized user model comprises a Bayesian network.
14. The system of claim 9 , the computer-readable storage medium having additional instructions stored which, when executed by the processor, cause the processor to perform operations comprising generating a personalized automatic speech recognition model based on the personalized user model.
15. The system of claim 9 , the computer-readable storage medium having additional instructions stored which, when executed by the processor, cause the processor to perform operations comprising recognizing additional speech from the user using the personalized user model.
16. The system of claim 15 , the computer-readable storage medium having additional instructions stored which, when executed by the processor, cause the processor to perform operations comprising iteratively improving the personalized user model based on the additional speech.
17. A computer-readable storage device having instructions stored which, when executed by a computing device, result in the computing device performing operations comprising:
receiving speech recognition output associated with speech from a user;
receiving an error model characterizing how automatic speech recognition transcription errors are made;
generating guesses of a true transcription based on an error model-based algorithm, wherein the error model-based algorithm uses the error model and the speech recognition output; and
generating, based on the guesses of the true transcription, a personalized user model associated with a user voiceprint of the user, wherein the generating of the personalized user model comprises iteratively guessing the true transcription until a threshold is met.
18. The computer-readable storage device of claim 17 , wherein generating of the guesses further comprises repeating, until a threshold is met, steps comprising:
guessing the true transcription from a current guess of the personalized user model, to yield a current guess of the true transcription; and
guessing the personalized user model based on the current guess of the true transcription.
19. The computer-readable storage device of claim 17 , wherein the error model-based algorithm estimates conditional probabilities of hidden variables.
20. The computer-readable storage device of claim 17 , wherein generating of the guesses is further based on a set of manual transcriptions of speech interactions with the user.