EP1425737A1 - System and method for speech verification using a robust confidence measure - Google Patents

System and method for speech verification using a robust confidence measure

Info

Publication number
EP1425737A1
EP1425737A1 EP02794872A EP02794872A EP1425737A1 EP 1425737 A1 EP1425737 A1 EP 1425737A1 EP 02794872 A EP02794872 A EP 02794872A EP 02794872 A EP02794872 A EP 02794872A EP 1425737 A1 EP1425737 A1 EP 1425737A1
Authority
EP
European Patent Office
Prior art keywords
score
recognition
confidence measure
pseudo filler
calculator
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
EP02794872A
Other languages
German (de)
French (fr)
Other versions
EP1425737A4 (en
Inventor
Gustavo Hernandez Abrego
Xavier Menendez Pidal
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sony Electronics Inc
Original Assignee
Sony Electronics Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sony Electronics Inc filed Critical Sony Electronics Inc
Publication of EP1425737A1 publication Critical patent/EP1425737A1/en
Publication of EP1425737A4 publication Critical patent/EP1425737A4/en
Withdrawn legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/08—Speech classification or search

Definitions

  • This invention relates generally to electronic speech recognition systems, and relates more particularly to a system and method for speech verification using a robust confidence measure.
  • Voice-controlled operation of electronic devices is a desirable interface for many system users. For example, voice- controlled operation allows a user to perform other tasks simultaneously. For instance, a person may operate a vehicle and operate an electronic organizer by voice control at the same time. Hands-free operation of electronic systems may also be desirable for users who have physical limitations or other special requirements. Hands-free operation of electronic devices may be implemented by various speech-activated electronic systems. Speech-activated electronic systems thus advantageously allow users to interface with electronic devices in situations where it would be inconvenient or potentially hazardous to utilize a traditional input device. Electronic entertainment systems may also utilize speech recognition techniques to allow users to interact with a system by speaking to it.
  • Speech-activated electronic systems may be used in a variety of noisy environments such as industrial facilities, manufacturing facilities, commercial vehicles, passenger vehicles, homes, and office environments.
  • noisy environments such as industrial facilities, manufacturing facilities, commercial vehicles, passenger vehicles, homes, and office environments.
  • a significant amount of noise in an environment may interfere with and degrade the performance and effectiveness of speech-activated systems.
  • System designers and manufacturers typically seek to develop speech- activated systems that provide reliable performance in noisy environments.
  • sound energy detected by a speech-activated system may contain speech and a significant amount of noise.
  • the speech may be masked by the noise and be undetected. This result is unacceptable for reliable performance of the speech-activated system.
  • sound energy detected by the speech-activated system may contain only noise.
  • the noise may be of such a character that the speech-activated system identifies the noise as speech. This result reduces the effectiveness of the speech-activated system, and is also unacceptable for reliable performance. Verifying that a detected signal is actually speech increases the effectiveness and reliability of speech-activated systems.
  • a speech-activated system may have a limited vocabulary of words that the system is programmed to recognize.
  • the system should accurately respond to words or phrases that are in its vocabulary, and should not respond to words or phrases that are not in its vocabulary. Verifying that a recognized word is in the system's vocabulary increases the accuracy and reliability of speech-activated systems.
  • a search module from a speech recognizer may preferably access feature vectors for a given utterance, and may then compare the feature vectors to word models corresponding to a system vocabulary.
  • the search module may then responsively generate a series of potential recognition candidates with their corresponding recognition scores based on the foregoing comparison with the word models.
  • the search module may also rank the recognition candidates into a sorted N-best list according to their respective recognition scores.
  • a pseudo filler ' score calculator from a speech verifier may then access the sorted N-best list to calculate a pseudo filler score in accordance with the present invention.
  • the pseudo filler score calculator may determine a pseudo filler score in any effective manner.
  • the pseudo filler score calculator may preferably generate a pseudo filler score by averaging selected recognition scores from the N-best list, excluding only the highest or best recognition score and the lowest or worst recognition score.
  • a confidence measure calculator from the speech verifier may then calculate a confidence measure by utilizing any effective means or technique.
  • the confidence measure calculator may preferably access the pseudo filler score, as well as the highest or best recognition score and the lowest or worst recognition score from the N-best list in order to perform a confidence measure calculation.
  • a comparator from the speech verifier may then compare the confidence measure to a threshold value obtained from local threshold registers or from any other appropriate source.
  • the threshold value may include any selectable value. For example, a single simplified threshold value may be selected for use by the comparator. Alternately, multiple thresholds may be utilized depending upon the degree of accuracy required during the speech recognition process. In certain embodiments, a different threshold may be selected to correspond to each best recognition candidate that is evaluated by the speech verifier.
  • the speech verifier may preferably verify the best recognition candidate as a valid utterance. Conversely, if the comparator determines that the confidence measure is not greater than the threshold value, then the speech verifier may preferably reject the best recognition candidate as an invalid utterance.
  • the present invention advantageously utilizes only information that is normally provided by the search module (for example, the N-best list and corresponding recognition scores) to perform the speech verification process.
  • the present invention thus provides a non-intrusive system and method for performing a robust speech verification procedure.
  • FIG. 1 is a block diagram for one embodiment of a computer system, in accordance with the present invention.
  • FIG. 2 is a block diagram for one embodiment of the memory of FIG. 1 , in accordance with the present invention.
  • FIG. 3 is a block diagram for one embodiment of the speech detector of FIG. 2, in accordance with the present invention.
  • FIG. 4 is a block diagram for one embodiment of the recognizer of FIG. 3, in accordance with the present invention.
  • FIG. 5 is a block diagram of one embodiment of an N-best list, in accordance with the present invention.
  • FIG. 6 is a block diagram illustrating a speech verification procedure, in accordance with one embodiment of the present invention.
  • FIG. 7 is a flowchart of method steps for performing a speech verification procedure, according to one embodiment of the present invention.
  • the present invention relates to an improvement in speech recognition systems.
  • the following description is presented to enable one of ordinary skill in the art to make and use the invention and is provided in the context of a patent application and its requirements.
  • Various modifications to the preferred embodiments will be readily apparent to those skilled in the art, and the generic principles herein may be applied to other embodiments.
  • the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features described herein.
  • the present invention comprises a system and method for speech verification using a robust confidence measure, and may include a speech verifier which may preferably compare a confidence measure for a recognized word to a predetermined threshold value in order to determine whether the recognized word is valid, where the recognized word corresponds to a word model that produces a highest recognition score.
  • the foregoing confidence measure may preferably be calculated by advantageously utilizing the recognition score for the recognized word, a background score, and a pseudo filler score which may preferably be based upon selected average recognition scores from an N-best list of recognition candidates.
  • FIG. 1 a block diagram for one embodiment of a computer system 110 is shown, according to the present invention.
  • the FIG. 1 embodiment includes a sound sensor 112, an amplifier 116, an analog-to- digital converter 120, a central processing unit (CPU) 128, a memory 130, and an input/ output interface 132.
  • CPU central processing unit
  • Sound sensor 112 detects sound energy and converts the detected sound energy into an analog speech signal that is provided via line 114 to amplifier 116.
  • Amplifier 116 amplifies the received analog speech signal and provides the amplified analog speech signal to analog- to-digital converter 120 via line 118.
  • Analog- to-digital converter 120 then converts the amplified analog speech signal into corresponding digital speech data.
  • Analog-to- digital converter 120 then provides the digital speech data via line 122 to system bus 124.
  • CPU 128 may then access the digital speech data on system bus 124 and responsively analyze and process the digital speech data to perform speech detection according to software instructions contained in memory 130. The operation of CPU 128 and the software instructions in memory 130 are further discussed below in conjunction with FIGS. 2-7. After the speech data is processed, CPU 128 may then provide the results of the speech detection analysis to other devices (not shown) via input/ output interface 132.
  • Memory 130 may alternately comprise various storage-device configurations, including random access memory (RAM) and storage devices such as floppy discs or hard disc drives.
  • RAM random access memory
  • memory 130 includes, but is not limited to, a speech detector 210, model score registers 212, confidence measure register 214, threshold registers 216, pseudo filler score register 218, and N-best list registers 220.
  • speech detector 210 includes a series of software modules that are executed by CPU 128 to analyze and detect speech data, and which are further described below in conjunction with FIGS. 3-4. In alternate embodiments, speech detector 210 may readily be implemented using various other software and/ or hardware configurations.
  • Model score registers 212, confidence measure register 214, threshold registers 216, pseudo filler score register 218, and N-best list registers 220 contain respective variable values that are calculated and utilized by speech detector 210 to implement the speech verification method of the present invention.
  • the utilization and functionality of model score registers 212, confidence measure register 214, threshold registers 216, pseudo filler score register 218, and N-best list registers 220 are further discussed below in conjunction with FIGS. 3-7.
  • Speech detector 210 includes, but is not limited to, a feature extractor 310, an endpoint detector 312, and a recognizer 314.
  • Analog-to-digital converter 120 provides digital speech data to feature extractor 310 via system bus 124.
  • Feature extractor 310 responsively generates feature vectors, which are provided to recognizer 314 via path 320.
  • Feature extractor 310 further responsively generates speech energy to endpoint detector 312 via path 322.
  • Endpoint detector 312 analyzes the speech energy and responsively determines endpoints of an utterance represented by the speech energy. The endpoints indicate the beginning and end of the utterance in time. Endpoint detector 312 then provides the endpoints to recognizer 314 via path 324.
  • Recognizer 314 is preferably configured to recognize isolated words or commands in a predetermined vocabulary of system 110.
  • recognizer 314 is configured to recognize a vocabulary of approximately one hundred twenty five words or commands.
  • a vocabulary including any number of words or commands is within the scope of the present invention.
  • the foregoing vocabulary may correspond to any desired commands, instructions, or other communications for system 110.
  • verifying a recognized word or phrase as valid is significant for proper operation of system 110.
  • Valid recognized vocabulary words or commands are output to system 110 via path 332.
  • FIG. 4 a block diagram for one embodiment of the recognizer 314 of FIG. 3 is shown, according to the present invention.
  • Recognizer 314 includes, but is not limited to, a search module 416, a model bank 412, and a speech verifier 414.
  • Model bank 412 includes a word model for every word or command in the vocabulary of system 110. Each model may preferably be a Hidden Markov Model that has been trained to recognize a specific word or command in the vocabulary.
  • Search module 416 preferably receives feature vectors from feature extractor 310 via path 320, and receives endpoint data from endpoint detector 312 via path 324. Search module 416 compares the feature vectors for an utterance (the signal between endpoints) with each word model in model bank 412. Search module 416 produces a recognition score for the utterance from each model, and stores the recognition scores in model score registers 212. Search module 416 preferably ranks the recognition scores for the utterance from highest to lowest, and stores a specified number of the ranked recognition scores as an N-best list in N-best list registers 220.
  • the word model that corresponds to the highest recognition score is the first recognition candidate
  • the word model that corresponds to the next-highest recognition score is the second recognition candidate
  • the word model that corresponds to the third-highest recognition score is the third recognition candidate
  • word models in model bank 412 may be ranked.
  • the first recognition candidate is considered to be the recognized word.
  • Search module 416 outputs the ranked N-best list to speech verifier
  • speech verifier 414 calculates a confidence measure for the utterance, and then stores the confidence measure in confidence measure register 214. Speech verifier 414 next compares the confidence measure for the utterance to a threshold value stored in threshold registers 216.
  • the value of the confidence measure is a measure of the confidence that the recognized word is a valid vocabulary word of system 110.
  • the word identified as the first candidate, or recognized word is considered valid, and recognizer 314 outputs the result on path 332. If the confidence measure is less than the threshold value, then the first candidate is considered an invalid vocabulary word, and recognizer 314 outputs no result.
  • threshold values stored in threshold registers 216 typically depend on the vocabulary words of system 110. Different sets of threshold values may be used in speech verifier 414, each set of threshold values providing different advantages to system 110.
  • N-best list 510 may preferably include a recognition candidate 1 (512(a)) through a recognition candidate N (512(c)).
  • N-best list 510 may readily include various other elements or functionalities in addition to, or instead of, those elements or functionalities discussed in conjunction with the FIG. 5 embodiment.
  • N-best list 510 may readily be implemented to include any desired number of recognition candidates 512 that may include any required type of information.
  • FIG. 5 illustrates a block diagram of an N-best list 510, in accordance with one embodiment of the present invention.
  • N-best list 510 may preferably include a recognition candidate 1 (512(a)) through a recognition candidate N (512(c)).
  • N-best list 510 may readily include various other elements or functionalities in addition to, or instead of, those elements or functionalities discussed in conjunction with the FIG. 5 embodiment.
  • N-best list 510 may readily be implemented to include any desired number of recognition candidates 512 that may include any required type of information.
  • each recognition candidate 512 may preferably include a search result (a word, phrase, or command) in text format, and a corresponding recognition score.
  • the recognition candidates 512 of N-best list 510 are preferably sorted and ranked by their recognition score, with recognition candidate 1 (512(a)) having the highest or best recognition score, and recognition candidate N (512(c)) have the lowest or worst recognition score.
  • the utilization of N-best list 510 is further discussed below in conjunction with FIGS. 6 and 7.
  • search module 416 may preferably access feature vectors for a given utterance via path 320, and then may compare the feature vectors to word models that may be accessed via path 422. Search module 416 may then responsively generate a series of potential recognition candidates 512 with their corresponding recognition scores based on the foregoing comparison to word models. Search module 416 may also rank the recognition candidates 512 into a sorted N-best list according to their respective recognition scores.
  • a pseudo filler score calculator 618 may access the sorted N-best list via path 616 to calculate a pseudo filler score in accordance with the present invention. Pseudo filler score calculator 618 may determine a pseudo filler score in any effective manner. However, in the FIG. 6 embodiment, pseudo filler score calculator 618 may preferably generate a pseudo filler score by averaging the recognition scores from the N- best list 510, excluding only the highest or best recognition score and the lowest or worst recognition score (also referred to herein as the background score).
  • pseudo filler score calculator 618 may preferably calculate a pseudo filler score by determining an arithmetic average of selected recognition scores from N-best list 510 (omitting the best recognition score and the worst recognition score) in accordance with the following formula:
  • F is the pseudo filler score
  • Si is the recognition score of the "i” word of N-best list 510
  • "N" is a selectable integer value that is equal to "N" of the N-best list 510.
  • a confidence measure calculator 624 may then calculate a confidence measure (CM) by utilizing any effective means or technique.
  • confidence measure calculator 624 may preferably access the pseudo filler score via path 620, and may also access the highest or best recognition score (Si) from search module 416 via path 622 and the lowest or worst recognition score (SN) from search module 416 via path 621 to perform a confidence measure calculation.
  • confidence measure calculator 624 may preferably calculate a confidence measure in accordance with the following formula:
  • CM the confidence measure
  • Si the best recognition score of the first recognition candidate 512(a) of N-best list 510
  • SN the worst recognition score of the lowest-ranked recognition candidate 512(c) of N-best list 510
  • F the foregoing pseudo filler score
  • a comparator 628 may then access the confidence measure (CM) via path 626, and compare the confidence measure to a threshold value obtained from threshold registers 216 (FIG. 2) or from any other appropriate source.
  • the threshold value may include any selectable value. For example, a single simplified threshold value may be selected for use by comparator 628. Alternately, multiple thresholds may be utilized depending upon the degree of accuracy required during the speech recognition process. In certain embodiments, a different individual threshold may be selected to correspond to each best recognition candidate that is evaluated by speech verifier 414.
  • speech verifier 414 may preferably verify the first recognition candidate 512(a) as a valid utterance, as shown in block 630. Conversely, if comparator 628 determines that the confidence measure is not greater than the threshold value, then speech verifier 414 may preferably reject the first recognition candidate 512(a) as an invalid utterance, as shown in block 632.
  • the present invention advantageously utilizes only information normally provided by search module 416 (N-best list 510 and corresponding recognition scores) to perform the speech verification process. The present invention thus provides a non-intrusive system and method for performing a robust speech verification procedure.
  • FIG. 7 a flowchart of method steps for performing a speech verification procedure is shown, in accordance with one embodiment of the present invention.
  • the FIG. 7 embodiment is presented for purposes of illustration, and in alternate embodiments, the present invention may readily utilize various steps and sequences other than those discussed in conjunction with the FIG. 7 embodiment.
  • search module 416 may preferably generate a sorted N-best list 510 corresponding to a particular utterance. Then, in step 710, a pseudo filler score calculator of a speech verifier 414 may preferably access the sorted N-best list 510, and responsively calculate a pseudo filler score, as discussed above in conjunction with FIG. 6.
  • a confidence measure calculator 624 of speech verifier 414 may preferably determine a confidence measure corresponding to a potential recognized word.
  • Confidence measure calculator 624 may preferably utilize the foregoing pseudo filler score, a background score corresponding to the worst recognition score of a worst recognition candidate 512(c) of the N-best list 510, and a best recognition score corresponding to the best recognition candidate 512(a) of the N-best list 510 to determine the confidence measure, as discussed above in conjunction with FIG. 6.
  • a comparator from speech verifier 414 may preferably compare the confidence measure to a selected threshold value, as discussed above in conjunction with FIG. 6.
  • comparator 414 may preferably determine whether the confidence measure is greater than the threshold value. If comparator 628 determines that the confidence measure is greater than the threshold value, then speech verifier 414 may preferably verify the first recognition candidate 512(a) as a valid utterance, as shown in step 718. Conversely, if comparator 628 determines that the confidence measure is not greater than the threshold value, then speech verifier 414 may preferably reject the first recognition candidate 512(a) as an invalid utterance, as shown in step 720.

Abstract

The present invention comprises a system and method for speech verification (Fig.6) using a robust confidence measure, and includes a speech verifier 414 which compares a confidence measure for a recognized word to a predetermined threshold value in order to determine whether the recognized word is valid, where a recognized word corresponds to a word model that produces a highest recognition score. In accordance with the present invention, the foregoing confidence measure may be calculated using the recognition score for the recognized word, a background score, and a pseudo filler score that may be based upon selected average recognition scores from an N-best list 510 of recognition candidates 512.

Description

SYSTEM AND METHOD FOR SPEECH VERIFICATION USING A ROBUST CONFIDENCE MEASURE
CROSS-REFERENCE TO RELATED APPLICATIONS
This Application claims the benefit of U.S. Provisional Application No. 60/312,501, entitled "Robust And Efficient Confidence Measure For Isolated Command Recognition," filed August 14, 2001. The related Applications are commonly assigned.
BACKGROUND SECTION
1. Field of the Invention
This invention relates generally to electronic speech recognition systems, and relates more particularly to a system and method for speech verification using a robust confidence measure.
2. Description of the Background Art
Implementing a robust and efficient method for system users to interface with electronic devices is a significant consideration of system designers and manufacturers. Voice-controlled operation of electronic devices is a desirable interface for many system users. For example, voice- controlled operation allows a user to perform other tasks simultaneously. For instance, a person may operate a vehicle and operate an electronic organizer by voice control at the same time. Hands-free operation of electronic systems may also be desirable for users who have physical limitations or other special requirements. Hands-free operation of electronic devices may be implemented by various speech-activated electronic systems. Speech-activated electronic systems thus advantageously allow users to interface with electronic devices in situations where it would be inconvenient or potentially hazardous to utilize a traditional input device. Electronic entertainment systems may also utilize speech recognition techniques to allow users to interact with a system by speaking to it. Speech-activated electronic systems may be used in a variety of noisy environments such as industrial facilities, manufacturing facilities, commercial vehicles, passenger vehicles, homes, and office environments. A significant amount of noise in an environment may interfere with and degrade the performance and effectiveness of speech-activated systems. System designers and manufacturers typically seek to develop speech- activated systems that provide reliable performance in noisy environments. In a noisy environment, sound energy detected by a speech-activated system may contain speech and a significant amount of noise. In such an environment, the speech may be masked by the noise and be undetected. This result is unacceptable for reliable performance of the speech-activated system.
Alternatively, sound energy detected by the speech-activated system may contain only noise. The noise may be of such a character that the speech-activated system identifies the noise as speech. This result reduces the effectiveness of the speech-activated system, and is also unacceptable for reliable performance. Verifying that a detected signal is actually speech increases the effectiveness and reliability of speech-activated systems.
In addition, a speech-activated system may have a limited vocabulary of words that the system is programmed to recognize. The system should accurately respond to words or phrases that are in its vocabulary, and should not respond to words or phrases that are not in its vocabulary. Verifying that a recognized word is in the system's vocabulary increases the accuracy and reliability of speech-activated systems.
Therefore, for all the foregoing reasons, implementing a robust and efficient method for a system user to interface with electronic devices remains a significant consideration of system designers and manufacturers. SUMMARY
In accordance with the present invention, a system and method are disclosed for performing speech verification using a robust confidence measure. In one embodiment, initially, a search module from a speech recognizer may preferably access feature vectors for a given utterance, and may then compare the feature vectors to word models corresponding to a system vocabulary. The search module may then responsively generate a series of potential recognition candidates with their corresponding recognition scores based on the foregoing comparison with the word models. The search module may also rank the recognition candidates into a sorted N-best list according to their respective recognition scores.
A pseudo filler' score calculator from a speech verifier may then access the sorted N-best list to calculate a pseudo filler score in accordance with the present invention. The pseudo filler score calculator may determine a pseudo filler score in any effective manner. For example, the pseudo filler score calculator may preferably generate a pseudo filler score by averaging selected recognition scores from the N-best list, excluding only the highest or best recognition score and the lowest or worst recognition score. A confidence measure calculator from the speech verifier may then calculate a confidence measure by utilizing any effective means or technique. For example, the confidence measure calculator may preferably access the pseudo filler score, as well as the highest or best recognition score and the lowest or worst recognition score from the N-best list in order to perform a confidence measure calculation.
A comparator from the speech verifier may then compare the confidence measure to a threshold value obtained from local threshold registers or from any other appropriate source. In accordance with the present invention, the threshold value may include any selectable value. For example, a single simplified threshold value may be selected for use by the comparator. Alternately, multiple thresholds may be utilized depending upon the degree of accuracy required during the speech recognition process. In certain embodiments, a different threshold may be selected to correspond to each best recognition candidate that is evaluated by the speech verifier.
If the comparator determines that the confidence measure is greater than the threshold value, then the speech verifier may preferably verify the best recognition candidate as a valid utterance. Conversely, if the comparator determines that the confidence measure is not greater than the threshold value, then the speech verifier may preferably reject the best recognition candidate as an invalid utterance.
The present invention advantageously utilizes only information that is normally provided by the search module (for example, the N-best list and corresponding recognition scores) to perform the speech verification process. The present invention thus provides a non-intrusive system and method for performing a robust speech verification procedure.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram for one embodiment of a computer system, in accordance with the present invention;
FIG. 2 is a block diagram for one embodiment of the memory of FIG. 1 , in accordance with the present invention;
FIG. 3 is a block diagram for one embodiment of the speech detector of FIG. 2, in accordance with the present invention;
FIG. 4 is a block diagram for one embodiment of the recognizer of FIG. 3, in accordance with the present invention;
FIG. 5 is a block diagram of one embodiment of an N-best list, in accordance with the present invention;
FIG. 6 is a block diagram illustrating a speech verification procedure, in accordance with one embodiment of the present invention; and
FIG. 7 is a flowchart of method steps for performing a speech verification procedure, according to one embodiment of the present invention.
DETAILED DESCRIPTION
The present invention relates to an improvement in speech recognition systems. The following description is presented to enable one of ordinary skill in the art to make and use the invention and is provided in the context of a patent application and its requirements. Various modifications to the preferred embodiments will be readily apparent to those skilled in the art, and the generic principles herein may be applied to other embodiments. Thus, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features described herein.
The present invention comprises a system and method for speech verification using a robust confidence measure, and may include a speech verifier which may preferably compare a confidence measure for a recognized word to a predetermined threshold value in order to determine whether the recognized word is valid, where the recognized word corresponds to a word model that produces a highest recognition score. In accordance with the present invention, the foregoing confidence measure may preferably be calculated by advantageously utilizing the recognition score for the recognized word, a background score, and a pseudo filler score which may preferably be based upon selected average recognition scores from an N-best list of recognition candidates.
Referring now to FIG. 1 , a block diagram for one embodiment of a computer system 110 is shown, according to the present invention. The FIG. 1 embodiment includes a sound sensor 112, an amplifier 116, an analog-to- digital converter 120, a central processing unit (CPU) 128, a memory 130, and an input/ output interface 132.
Sound sensor 112 detects sound energy and converts the detected sound energy into an analog speech signal that is provided via line 114 to amplifier 116. Amplifier 116 amplifies the received analog speech signal and provides the amplified analog speech signal to analog- to-digital converter 120 via line 118. Analog- to-digital converter 120 then converts the amplified analog speech signal into corresponding digital speech data. Analog-to- digital converter 120 then provides the digital speech data via line 122 to system bus 124.
CPU 128 may then access the digital speech data on system bus 124 and responsively analyze and process the digital speech data to perform speech detection according to software instructions contained in memory 130. The operation of CPU 128 and the software instructions in memory 130 are further discussed below in conjunction with FIGS. 2-7. After the speech data is processed, CPU 128 may then provide the results of the speech detection analysis to other devices (not shown) via input/ output interface 132.
Referring now to FIG. 2, a block diagram for one embodiment of the memory 130 of FIG. 1 is shown, according to the present invention. Memory 130 may alternately comprise various storage-device configurations, including random access memory (RAM) and storage devices such as floppy discs or hard disc drives. In the FIG. 2 embodiment, memory 130 includes, but is not limited to, a speech detector 210, model score registers 212, confidence measure register 214, threshold registers 216, pseudo filler score register 218, and N-best list registers 220.
In the FIG. 2 embodiment, speech detector 210 includes a series of software modules that are executed by CPU 128 to analyze and detect speech data, and which are further described below in conjunction with FIGS. 3-4. In alternate embodiments, speech detector 210 may readily be implemented using various other software and/ or hardware configurations.
Model score registers 212, confidence measure register 214, threshold registers 216, pseudo filler score register 218, and N-best list registers 220 contain respective variable values that are calculated and utilized by speech detector 210 to implement the speech verification method of the present invention. The utilization and functionality of model score registers 212, confidence measure register 214, threshold registers 216, pseudo filler score register 218, and N-best list registers 220 are further discussed below in conjunction with FIGS. 3-7.
Referring now to FIG. 3, a block diagram for one embodiment of the speech detector 210 of FIG. 2 is shown, according to the present invention. Speech detector 210 includes, but is not limited to, a feature extractor 310, an endpoint detector 312, and a recognizer 314.
Analog-to-digital converter 120 (FIG. 1) provides digital speech data to feature extractor 310 via system bus 124. Feature extractor 310 responsively generates feature vectors, which are provided to recognizer 314 via path 320. Feature extractor 310 further responsively generates speech energy to endpoint detector 312 via path 322. Endpoint detector 312 analyzes the speech energy and responsively determines endpoints of an utterance represented by the speech energy. The endpoints indicate the beginning and end of the utterance in time. Endpoint detector 312 then provides the endpoints to recognizer 314 via path 324.
Recognizer 314 is preferably configured to recognize isolated words or commands in a predetermined vocabulary of system 110. In the FIG. 3 embodiment, recognizer 314 is configured to recognize a vocabulary of approximately one hundred twenty five words or commands. However, a vocabulary including any number of words or commands is within the scope of the present invention. The foregoing vocabulary may correspond to any desired commands, instructions, or other communications for system 110. Thus, verifying a recognized word or phrase as valid is significant for proper operation of system 110. Valid recognized vocabulary words or commands are output to system 110 via path 332.
Referring now to FIG. 4, a block diagram for one embodiment of the recognizer 314 of FIG. 3 is shown, according to the present invention.
Recognizer 314 includes, but is not limited to, a search module 416, a model bank 412, and a speech verifier 414. Model bank 412 includes a word model for every word or command in the vocabulary of system 110. Each model may preferably be a Hidden Markov Model that has been trained to recognize a specific word or command in the vocabulary.
Search module 416 preferably receives feature vectors from feature extractor 310 via path 320, and receives endpoint data from endpoint detector 312 via path 324. Search module 416 compares the feature vectors for an utterance (the signal between endpoints) with each word model in model bank 412. Search module 416 produces a recognition score for the utterance from each model, and stores the recognition scores in model score registers 212. Search module 416 preferably ranks the recognition scores for the utterance from highest to lowest, and stores a specified number of the ranked recognition scores as an N-best list in N-best list registers 220. The word model that corresponds to the highest recognition score is the first recognition candidate, the word model that corresponds to the next-highest recognition score is the second recognition candidate, the word model that corresponds to the third-highest recognition score is the third recognition candidate, and in this way, word models in model bank 412 may be ranked. Typically, the first recognition candidate is considered to be the recognized word. Search module 416 outputs the ranked N-best list to speech verifier
414 via path 420. In one embodiment, speech verifier 414 calculates a confidence measure for the utterance, and then stores the confidence measure in confidence measure register 214. Speech verifier 414 next compares the confidence measure for the utterance to a threshold value stored in threshold registers 216. The value of the confidence measure is a measure of the confidence that the recognized word is a valid vocabulary word of system 110.
If the confidence measure for the utterance is greater than or equal to the threshold value, then the word identified as the first candidate, or recognized word, is considered valid, and recognizer 314 outputs the result on path 332. If the confidence measure is less than the threshold value, then the first candidate is considered an invalid vocabulary word, and recognizer 314 outputs no result.
The threshold values stored in threshold registers 216 typically depend on the vocabulary words of system 110. Different sets of threshold values may be used in speech verifier 414, each set of threshold values providing different advantages to system 110.
Referring now to FIG. 5, a block diagram of an N-best list 510 is shown, in accordance with one embodiment of the present invention. In the FIG. 5 embodiment, N-best list 510 may preferably include a recognition candidate 1 (512(a)) through a recognition candidate N (512(c)). In alternate embodiments, N-best list 510 may readily include various other elements or functionalities in addition to, or instead of, those elements or functionalities discussed in conjunction with the FIG. 5 embodiment. In the FIG. 5 embodiment, N-best list 510 may readily be implemented to include any desired number of recognition candidates 512 that may include any required type of information. In the FIG. 5 embodiment, each recognition candidate 512 may preferably include a search result (a word, phrase, or command) in text format, and a corresponding recognition score. In the FIG. 5 embodiment, the recognition candidates 512 of N-best list 510 are preferably sorted and ranked by their recognition score, with recognition candidate 1 (512(a)) having the highest or best recognition score, and recognition candidate N (512(c)) have the lowest or worst recognition score. The utilization of N-best list 510 is further discussed below in conjunction with FIGS. 6 and 7.
Referring now to FIG. 6, a block diagram illustrating a speech verification procedure is shown, in accordance with one embodiment of the present invention. In alternate embodiments, the present invention may readily perform speech verification using various other elements or functionalities in addition to, or instead of, those elements or functionalities discussed in conjunction with the FIG. 6 embodiment. In the FIG. 6 embodiment, initially, search module 416 may preferably access feature vectors for a given utterance via path 320, and then may compare the feature vectors to word models that may be accessed via path 422. Search module 416 may then responsively generate a series of potential recognition candidates 512 with their corresponding recognition scores based on the foregoing comparison to word models. Search module 416 may also rank the recognition candidates 512 into a sorted N-best list according to their respective recognition scores.
In the FIG. 6 embodiment, a pseudo filler score calculator 618 may access the sorted N-best list via path 616 to calculate a pseudo filler score in accordance with the present invention. Pseudo filler score calculator 618 may determine a pseudo filler score in any effective manner. However, in the FIG. 6 embodiment, pseudo filler score calculator 618 may preferably generate a pseudo filler score by averaging the recognition scores from the N- best list 510, excluding only the highest or best recognition score and the lowest or worst recognition score (also referred to herein as the background score).
In certain embodiments, pseudo filler score calculator 618 may preferably calculate a pseudo filler score by determining an arithmetic average of selected recognition scores from N-best list 510 (omitting the best recognition score and the worst recognition score) in accordance with the following formula:
ι AM
F = — — ∑S,
N-2 i '
where F is the pseudo filler score, Si is the recognition score of the "i" word of N-best list 510, and "N" is a selectable integer value that is equal to "N" of the N-best list 510.
In the FIG. 6 embodiment, a confidence measure calculator 624 may then calculate a confidence measure (CM) by utilizing any effective means or technique. In the FIG. 6 embodiment, confidence measure calculator 624 may preferably access the pseudo filler score via path 620, and may also access the highest or best recognition score (Si) from search module 416 via path 622 and the lowest or worst recognition score (SN) from search module 416 via path 621 to perform a confidence measure calculation.
In certain embodiments, confidence measure calculator 624 may preferably calculate a confidence measure in accordance with the following formula:
CM = Sι -F
S! - SΛ
where CM is the confidence measure, Si is the best recognition score of the first recognition candidate 512(a) of N-best list 510, SN is the worst recognition score of the lowest-ranked recognition candidate 512(c) of N-best list 510, and F is the foregoing pseudo filler score.
In the FIG. 6 embodiment, a comparator 628 may then access the confidence measure (CM) via path 626, and compare the confidence measure to a threshold value obtained from threshold registers 216 (FIG. 2) or from any other appropriate source. In accordance with the present invention, the threshold value may include any selectable value. For example, a single simplified threshold value may be selected for use by comparator 628. Alternately, multiple thresholds may be utilized depending upon the degree of accuracy required during the speech recognition process. In certain embodiments, a different individual threshold may be selected to correspond to each best recognition candidate that is evaluated by speech verifier 414.
In the FIG. 6 embodiment, if comparator 628 determines that the confidence measure is greater than the threshold value, then speech verifier 414 may preferably verify the first recognition candidate 512(a) as a valid utterance, as shown in block 630. Conversely, if comparator 628 determines that the confidence measure is not greater than the threshold value, then speech verifier 414 may preferably reject the first recognition candidate 512(a) as an invalid utterance, as shown in block 632. The present invention advantageously utilizes only information normally provided by search module 416 (N-best list 510 and corresponding recognition scores) to perform the speech verification process. The present invention thus provides a non-intrusive system and method for performing a robust speech verification procedure.
Referring now to FIG. 7, a flowchart of method steps for performing a speech verification procedure is shown, in accordance with one embodiment of the present invention. The FIG. 7 embodiment is presented for purposes of illustration, and in alternate embodiments, the present invention may readily utilize various steps and sequences other than those discussed in conjunction with the FIG. 7 embodiment.
In the FIG. 7 embodiment, in step 708, search module 416 may preferably generate a sorted N-best list 510 corresponding to a particular utterance. Then, in step 710, a pseudo filler score calculator of a speech verifier 414 may preferably access the sorted N-best list 510, and responsively calculate a pseudo filler score, as discussed above in conjunction with FIG. 6.
In step 712, a confidence measure calculator 624 of speech verifier 414 may preferably determine a confidence measure corresponding to a potential recognized word. Confidence measure calculator 624 may preferably utilize the foregoing pseudo filler score, a background score corresponding to the worst recognition score of a worst recognition candidate 512(c) of the N-best list 510, and a best recognition score corresponding to the best recognition candidate 512(a) of the N-best list 510 to determine the confidence measure, as discussed above in conjunction with FIG. 6.
In step 714, a comparator from speech verifier 414 may preferably compare the confidence measure to a selected threshold value, as discussed above in conjunction with FIG. 6. In step 716, comparator 414 may preferably determine whether the confidence measure is greater than the threshold value. If comparator 628 determines that the confidence measure is greater than the threshold value, then speech verifier 414 may preferably verify the first recognition candidate 512(a) as a valid utterance, as shown in step 718. Conversely, if comparator 628 determines that the confidence measure is not greater than the threshold value, then speech verifier 414 may preferably reject the first recognition candidate 512(a) as an invalid utterance, as shown in step 720.
The invention has been explained above with reference to preferred embodiments. Other embodiments will be apparent to those skilled in the art in light of this disclosure. For example, the present invention may readily be implemented using configurations and techniques other than those described in the preferred embodiments above. Additionally, the present invention may effectively be used in conjunction with systems other than those described above as the preferred embodiments. Therefore, these and other variations upon the preferred embodiments are intended to be covered by the present invention, which is limited only by the appended claims.

Claims

WHAT IS CLAIMED IS:
1. A system for performing a speech verification of an utterance, comprising: a search module 416 configured to generate an N-best list 510 of recognition candidates 512 corresponding to said utterance; a pseudo filler score calculator 618 coupled to said search module 416 for determining a pseudo filler score that is based upon recognition scores corresponding to said recognition candidates 512; a confidence measure calculator 624 for utilizing said pseudo filler score, a background score of a worst recognition candidate 512(a), and a best recognition score of a best recognition candidate 512(c) to determine a confidence measure; and a comparator 628 configured to compare said confidence measure to a threshold value for determining whether to verify said best recognition candidate 512(c).
2. The system of claim 1 wherein said system comprises a speech recognizer 314 that includes said search module 416, a model bank 412 of word models corresponding to a vocabulary of said system, and a speech verifier 414 that includes said pseudo filler score calculator 618, said confidence measure calculator 624, and said comparator 628.
3. The system of claim 1 wherein said search module 416 compares feature vectors corresponding to said utterance with word models to thereby generate said recognition candidates 512 and said recognition scores that each correspond to a different one of said recognition candidates 512.
4. The system of claim 1 wherein said search module 416 ranks said N- best list 510 according to said recognition scores that each correspond to a different one of said recognition candidates 512, said N-best list 510 including said best recognition candidate 512(c) having said best recognition score, said worst recognition candidate 512(a) having said background score, and a pre-determined additional number of said recognition candidates 512.
5. The system of claim 1 wherein said pseudo filler score calculator 618 determines said pseudo filler score by omitting said best recognition score and said background score, and then analyzing remaining ones of said recognition scores from said N-best list 510.
6. The system of claim 1 wherein said pseudo filler score calculator 618 determines said pseudo filler score by analyzing only selected ones of said recognition scores from said N-best list 510.
7. The system of claim 1 wherein said pseudo filler score calculator 618 determines said pseudo filler score by calculating an arithmetic average of selected ones of said recognition scores from said N-best list 510.
8. The system of claim 1 wherein said pseudo filler score calculator 618 determines said pseudo filler score according to a following formula:
1 JV-1
F = — ∑St
where F is said pseudo filler score, Si is a recognition score of an "i" word of said N-best list 510, and "N" is a selectable integer value that is equal to "N" of said N-best list 510.
9. The system of claim 1 wherein said pseudo filler score calculator 618 determines said pseudo filler score by calculating an arithmetic mean of said recognition scores from said N-best list 510 after omitting said background score and said best recognition score.
10. The system of claim 1 wherein said background score is equal to a worst score corresponding to said a lowest-ranked recognition candidate of said N-best list 510.
11. The system of claim 1 wherein said confidence measure calculator 624 determines said confidence measure corresponding to said best recognition candidate 512(c) by comparing said pseudo filler score and at least one of said recognition scores from said N-best list 510.
12. The system of claim 1 wherein said confidence measure calculator 624 determines said confidence measure by calculating a ratio that compares said pseudo filler score and said background score from said N-best list 510.
13. The system of claim 1 wherein said confidence measure calculator 624 determines said confidence measure by dividing said best recognition score minus said pseudo filler score by said best recognition score minus said background score.
14. The system of claim 1 wherein said confidence measure calculator 624 determines said confidence measure according to a following formula:
CM = -^ ~ F
S! ~ SΛ
where CM is said confidence measure, Si is said best recognition score of said best recognition candidate 512(c) of said N-best list 510, SN is a worst recognition score of a lowest- ranked recognition candidate of said N-best list 510, and F is said pseudo filler score.
15. The system of claim 1 wherein said comparator 628 selectively accesses said threshold value from threshold registers 216, and then compares said confidence measure to said threshold value to generate a verification result corresponding to said best recognition candidate 512(c).
16. The system of claim 1 wherein said comparator 628 utilizes a single threshold value that is selected for efficiently and economically performing said speech verification.
17. The system of claim 1 wherein said threshold value may be selected from several threshold options to optimally achieve pre-determined performance objectives while performing said speech verification.
18. The system of claim 1 wherein said comparator 628 selectively utilizes different threshold values that each correspond to a different best recognition candidate 512 (c) .
19. The system of claim 2 wherein said speech verifier 414 designates said best recognition candidate 512(c) as valid whenever said confidence measure exceeds said threshold value, said speech verifier 414 rejecting said best recognition candidate 512(c) as invalid whenever said confidence measure fails to exceed said threshold value.
20. The system of claim 2 wherein said speech verifier 414 non-intrusively utilizes only information routinely provided by said search module 416 in order to perform said speech verification, said information including said N- best list 510 and said recognition scores.
21. A method for performing a speech verification of an utterance, comprising the steps of: generating an N-best list 510 of recognition candidates 512 corresponding to said utterance by using a search module 416; utilizing a pseudo filler score calculator 618 to determine a pseudo filler score that is based upon recognition scores corresponding to said recognition candidates 512; determining a confidence measure with a confidence measure calculator 624 by using said pseudo filler score, a background score of a worst recognition candidate 512(a), and a best recognition score of a best recognition candidate 512(c); and comparing said confidence measure to a threshold value with a comparator 628 to thereby determine whether to verify said best recognition candidate 512(c).
22. The method of claim 21 wherein said system comprises a speech recognizer 314 that includes said search module 416, a model bank 412 of word models corresponding to a vocabulary of said system, and a speech verifier 414 that includes said pseudo filler score calculator 618, said confidence measure calculator 624, and said comparator 628.
23. The method of claim 21 wherein said search module 416 compares feature vectors corresponding to said utterance with word models to thereby generate said recognition candidates 512 and said recognition scores that each correspond to a different one of said recognition candidates 512.
24. The method of claim 21 wherein said search module 416 ranks said N- best list 510 according to said recognition scores that each correspond to a different one of said recognition candidates 512, said N-best list 510 including said best recognition candidate 512(c) having said best recognition score, said worst recognition candidate 512(a) having said background score, and a pre-determined additional number of said recognition candidates 512.
25. The method of claim 21 wherein said pseudo filler score calculator 618 determines said pseudo filler score by omitting said best recognition score and said background score, and then analyzing remaining ones of said recognition scores from said N-best list 510.
26. The method of claim 21 wherein said pseudo filler score calculator 618 determines said pseudo filler score by analyzing only selected ones of said recognition scores from said N-best list 510.
27. The method of claim 21 wherein said pseudo filler score calculator 618 determines said pseudo filler score by calculating an arithmetic average of selected ones of said recognition scores from said N-best list 510.
28. The method of claim 21 wherein said pseudo filler score calculator 618 determines said pseudo filler score according to a following formula:
where F is said pseudo filler score, Si is a recognition score of an "i" word of said N-best list 510, and "N" is a selectable integer value that is equal to "N" of said N-best list 510.
29. The method of claim 21 wherein said pseudo filler score calculator 618 determines said pseudo filler score by calculating an arithmetic mean of said recognition scores from said N-best list 510 after omitting said background score and said best recognition score.
30. The method of claim 21 wherein said background score is equal to a worst score corresponding to said a lowest- ranked recognition candidate of said N-best list 510.
31. The method of claim 21 wherein said confidence measure calculator 624 determines said confidence measure corresponding to said best recognition candidate 512(c) by comparing said pseudo filler score and at least one of said recognition scores from said N-best list 510.
32. The method of claim 21 wherein said confidence measure calculator 624 determines said confidence measure by calculating a ratio that compares said pseudo filler score and said background score from said N-best list 510.
33. The method of claim 21 wherein said confidence measure calculator 624 determines said confidence measure by dividing said best recognition score minus said pseudo filler score by said best recognition score minus said background score.
34. The method of claim 21 wherein said confidence measure calculator 624 determines said confidence measure according to a following formula:
St -F
CM =
S! ~ SΛ
where CM is said confidence measure, Si is said best recognition score of said best recognition candidate 512(c) of said N-best list 510, SN is a worst recognition score of a lowest-ranked recognition candidate of said N-best list 510, and F is said pseudo filler score.
35. The method of claim 21 wherein said comparator 628 selectively accesses said threshold value from threshold registers 216, and then compares said confidence measure to said threshold value to generate a verification result corresponding to said best recognition candidate 512(c).
36. The method of claim 21 wherein said comparator 628 utilizes a single threshold value that is selected for efficiently and economically performing said speech verification.
37. The method of claim 21 wherein said threshold value may be selected from several threshold options to optimally achieve pre-determined performance objectives while performing said speech verification.
38. The method of claim 21 wherein said comparator 628 selectively utilizes different threshold values that each correspond to a different best recognition candidate 512(c).
39. The method of claim 22 wherein said speech verifier 414 designates said best recognition candidate 512(c) as valid whenever said confidence measure exceeds said threshold value, said speech verifier 414 rejecting said best recognition candidate 512(c) as invalid whenever said confidence measure fails to exceed said threshold value.
40. The method of claim 22 wherein said speech verifier 414 non- intrusively utilizes only information routinely provided by said search module 416 in order to perform said speech verification, said information including said N-best list 510
41. A computer-readable medium comprising program instructions for performing a speech verification of an utterance, by performing the steps of: generating an N-best list 510 of recognition candidates 512 corresponding to said utterance by using a search module 416; utilizing a pseudo filler score calculator 618 to determine a pseudo filler score that is based upon recognition scores corresponding to said recognition candidates 512; determining a confidence measure with a confidence measure calculator 624 by using said pseudo filler score, a background score of a worst recognition candidate 512(a), and a best recognition score of a best recognition candidate 512(c); and comparing said confidence measure to a threshold value with a comparator 628 to thereby determine whether to verify said best recognition candidate 512(c).
42. A system for performing a speech verification of an utterance, comprising: means for generating an N-best list 510 of recognition candidates 512 corresponding to said utterance; means for calculating a pseudo filler score that is based upon recognition scores corresponding to said recognition candidates 512; means for determining a confidence measure by using said pseudo filler score, a background score of a worst recognition candidate 512(a), and a best recognition score of a best recognition candidate 512(c); and means for comparing said confidence measure to a threshold value to thereby determine whether to verify said best recognition candidate 512(c).
43. A method for performing a speech verification of an utterance, comprising the steps of: generating a list 510 of recognition candidates 512 corresponding to said utterance; determining a confidence measure from a pseudo filler score, a background score of a worst recognition candidate 512(a), and a best recognition score of a best recognition candidate 512(c) from said list 510; and comparing said confidence measure to a threshold value to thereby determine whether to verify said best recognition candidate
512(c).
44. A system for performing a speech verification of an utterance, comprising: a search module 416 configured to generate an N-best list 510 of recognition candidates 512 corresponding to said utterance; a pseudo filler score calculator 618 coupled to said search module 416 for determining a pseudo filler score that is based upon recognition scores corresponding to said recognition candidates 512, said pseudo filler score calculator 618 determining said pseudo filler score according to a following formula:
1 Λ-l
F = —∑S,
where F is said pseudo filler score, Si is a recognition score of an "i" word of said N-best list 510, and "N" is a selectable integer value that is equal to "N" of said N-best list 510; a confidence measure calculator 624 for utilizing said pseudo filler score, a background score of a worst recognition candidate 512(a), and a best recognition score of a best recognition candidate 512(c) to determine a confidence measure, said confidence measure calculator 624 determining said confidence measure according to a following formula:
where CM is said confidence measure, Si is said best recognition score of said best recognition candidate 512(c) of said N-best list 510, SN is a worst recognition score of a lowest- ranked recognition candidate of said N-best list 510, and F is said pseudo filler score; and a comparator 628 configured to compare said confidence measure to a threshold value for determining whether to verify said best recognition candidate 512(c).
EP02794872A 2001-08-14 2002-08-13 System and method for speech verification using a robust confidence measure Withdrawn EP1425737A4 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US31250101P 2001-08-14 2001-08-14
US312501P 2001-08-14
PCT/US2002/025577 WO2003017253A1 (en) 2001-08-14 2002-08-13 System and method for speech verification using a robust confidence measure

Publications (2)

Publication Number Publication Date
EP1425737A1 true EP1425737A1 (en) 2004-06-09
EP1425737A4 EP1425737A4 (en) 2005-11-09

Family

ID=23211748

Family Applications (1)

Application Number Title Priority Date Filing Date
EP02794872A Withdrawn EP1425737A4 (en) 2001-08-14 2002-08-13 System and method for speech verification using a robust confidence measure

Country Status (4)

Country Link
EP (1) EP1425737A4 (en)
JP (1) JP2005500580A (en)
GB (1) GB2394590B (en)
WO (1) WO2003017253A1 (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3819896B2 (en) * 2003-11-14 2006-09-13 日本電信電話株式会社 Speech recognition method, apparatus for implementing this method, program, and recording medium
TWI466101B (en) * 2012-05-18 2014-12-21 Asustek Comp Inc Method and system for speech recognition
CN103426428B (en) * 2012-05-18 2016-05-25 华硕电脑股份有限公司 Audio recognition method and system
JP6585022B2 (en) 2016-11-11 2019-10-02 株式会社東芝 Speech recognition apparatus, speech recognition method and program

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5842163A (en) * 1995-06-21 1998-11-24 Sri International Method and apparatus for computing likelihood and hypothesizing keyword appearance in speech
US20010049600A1 (en) * 1999-10-21 2001-12-06 Sony Corporation And Sony Electronics, Inc. System and method for speech verification using an efficient confidence measure
EP1189202A1 (en) * 2000-09-18 2002-03-20 Sony International (Europe) GmbH Duration models for speech recognition

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5737489A (en) * 1995-09-15 1998-04-07 Lucent Technologies Inc. Discriminative utterance verification for connected digits recognition

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5842163A (en) * 1995-06-21 1998-11-24 Sri International Method and apparatus for computing likelihood and hypothesizing keyword appearance in speech
US20010049600A1 (en) * 1999-10-21 2001-12-06 Sony Corporation And Sony Electronics, Inc. System and method for speech verification using an efficient confidence measure
EP1189202A1 (en) * 2000-09-18 2002-03-20 Sony International (Europe) GmbH Duration models for speech recognition

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of WO03017253A1 *

Also Published As

Publication number Publication date
GB0402781D0 (en) 2004-03-10
GB2394590A (en) 2004-04-28
EP1425737A4 (en) 2005-11-09
JP2005500580A (en) 2005-01-06
WO2003017253A1 (en) 2003-02-27
GB2394590B (en) 2005-02-16

Similar Documents

Publication Publication Date Title
US6778959B1 (en) System and method for speech verification using out-of-vocabulary models
US7103543B2 (en) System and method for speech verification using a robust confidence measure
Prabhavalkar et al. Automatic gain control and multi-style training for robust small-footprint keyword spotting with deep neural networks
US6850886B2 (en) System and method for speech verification using an efficient confidence measure
US9633652B2 (en) Methods, systems, and circuits for speaker dependent voice recognition with a single lexicon
US10013985B2 (en) Systems and methods for audio command recognition with speaker authentication
US6735562B1 (en) Method for estimating a confidence measure for a speech recognition system
US6226612B1 (en) Method of evaluating an utterance in a speech recognition system
US7447634B2 (en) Speech recognizing apparatus having optimal phoneme series comparing unit and speech recognizing method
US6505155B1 (en) Method and system for automatically adjusting prompt feedback based on predicted recognition accuracy
TWI475558B (en) Method and apparatus for utterance verification
US20120191455A1 (en) System and Method for an Endpoint Detection of Speech for Improved Speech Recognition in Noisy Environments
US20050182628A1 (en) Domain-based dialog speech recognition method and apparatus
JPH10319988A (en) Speaker identifying method and speaker recognizing device
US6473735B1 (en) System and method for speech verification using a confidence measure
Herbig et al. Self-learning speaker identification for enhanced speech recognition
US6970818B2 (en) Methodology for implementing a vocabulary set for use in a speech recognition system
WO2004086357A2 (en) System and method for speech recognition utilizing a merged dictionary
WO2019097215A1 (en) Enrolment in speaker recognition system
CN101123090A (en) Speech recognition by statistical language using square-rootdiscounting
EP1425737A1 (en) System and method for speech verification using a robust confidence measure
US6823304B2 (en) Speech recognition apparatus and method performing speech recognition with feature parameter preceding lead voiced sound as feature parameter of lead consonant
KR101229108B1 (en) Apparatus for utterance verification based on word specific confidence threshold
US7231352B2 (en) Method for computer-supported speech recognition, speech recognition system and control device for controlling a technical system and telecommunications device
Lee et al. Dynamic time warping-based Korean spoken word detection system using euclidean distance in intelligent personal assistants

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 20040225

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR IE IT LI LU MC NL PT SE SK TR

AX Request for extension of the european patent

Extension state: AL LT LV MK RO SI

A4 Supplementary search report drawn up and despatched

Effective date: 20050928

17Q First examination report despatched

Effective date: 20081111

GRAP Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOSNIGR1

RIC1 Information provided on ipc code assigned before grant

Ipc: G10L 17/00 20060101ALI20100208BHEP

Ipc: G10L 15/08 20060101AFI20100208BHEP

RBV Designated contracting states (corrected)

Designated state(s): DE FR GB

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN

18D Application deemed to be withdrawn

Effective date: 20100713