Telephonic communications – Audio message storage – retrieval – or synthesis – Voice activation or recognition
Reissue Patent
2000-02-16
2003-04-29
Hong, Harry S. (Department: 2642)
Telephonic communications
Audio message storage, retrieval, or synthesis
Voice activation or recognition
C379S189000, C379S199000, C379S216010, C379S361000, C704S246000, C704S275000, C704S270100
Reissue Patent
active
RE038101
ABSTRACT:
FIELD OF THE INVENTION
The present invention is directed to telephone systems and, more particularly, to methods and apparatus for activating telephone services in response to speech.
BACKGROUND OF THE INVENTION
Telephones are used to provide a host of services in addition to basic calling services. Such telephone services include services such as repeat dialing and call return where security is not of concern. Telephone services also include banking and financial services where security is of concern.
Voice controlled dialing systems such as the one described in U.S. Pat. No. 5,165,095 permit a user to place a call verbally without knowing the number of the person being called. In accordance with the known system a user first speaks a command, e.g., the word “call” followed by a destination identifier. Once the command is identified using speaker independent voice recognition techniques, the system accesses speaker dependent or independent templates to recognize the destination identifier. The recognized destination identifier associated with the person making the call is then used to access a data base associated with the caller and to obtain therefrom a telephone number.
The known system uses both speaker independent and speaker dependent speech recognition.
Speaker dependent speech recognition uses a computer that has been “trained” to respond to the manner in which a particular person speaks. In general, the training involves one person speaking a sound to generate an analog speech input, converting the speech input into signal data, generating a template representing the sound, and indexing the template to appropriate response data, such as a computer instruction to perform an action. During real time applications, input data is compared to the user's set of templates and the best match results in an appropriate response.
Speaker independent speech recognition uses a computer that stores a composite template or cluster of templates that represent the same sound spoken by a number of different persons. The templates are derived from numerous samples of signal data to represent a wide range of pronunciations.
The requirement for steering words in the known system preceding, e.g., a destination identifier such as a name, places constraints on a user in terms of how a user must format a spoken request to place a call. Such format constraints require a user to speak in a manner that may be unnatural or uncomfortable for a user.
In order to provide voice controlled telephone services in a more user friendly manner, it is desirable that the need for commands or “steering words” be eliminated for basic calling functions. In addition, it is desirable that a user be able to place calls without excessive artificial constraints being placed on the words or sequence of words required to, e.g., place a call.
In addition to placing calls, it is desirable that other telephone services, e.g., call forwarding, repeat dial, call return, etc. also be available to a user through the use of speech control. It is also desirable that telephone services, e.g., banking transactions, involving matters requiring various degrees of security, also be supported without the need for a user to supply security codes or other personal identification number (PIN) information.
SUMMARY OF THE PRESENT INVENTION
The present invention is directed to telephone systems and, more particularly, to methods and apparatus for activating telephone services in response to speech.
In accordance with the present invention a directory of names and telephone numbers is maintained for each customer. A speaker dependent speech template is created for each name as it is added by a customer to the customer's personal telephone directory. The speaker dependent templates are stored in a data base which can be downloaded when a customer initiates a telephone call.
The present invention permits a user, e.g., customer, to place telephone calls using speech as opposed to telephone numbers without the use of steering words, e.g., the command “call” being-spoken before a destination identifier such as the name of an individual to be called. It accomplishes this by interpreting speech which matches a name in a customer's personal directory, absent an additional command, as being an attempt to initiate a call. Thus, upon detecting a match between a spoken name and a name in the customer's directory, the telephone number associated with the name in the customer's personal directory is dialed absent the receipt of an explicit command.
Identification of the customer for purposes of accessing the customer's personal telephone directory may be achieved using the telephone number of the telephone used to initiate the telephone transaction, by using a calling card number to initiate the telephone call or by a plurality of other known customer identification methods.
In addition to speaker dependent speech templates the present invention uses a plurality of speaker independent templates. The speaker independent templates are used to identify commands, e.g., REDIAL, CALL RETURN, etc., yes
o responses; and/or speaker independent names such as, e.g., home, which may be used by a plurality of customers.
In accordance with the present invention, speaker dependent speech recognition, e.g., to identify names in a customer's telephone directory, is performed in parallel with speaker independent speech recognition. The speaker independent recognition is used to detect, e.g., explicit commands.
After both speaker dependent and speaker independent speech recognition is performed on a received utterance which may be a word, e.g., the name John, or a phrase, e.g., Call forwarding to Mary, the results of the speech recognition steps are analyzed. In the event that only a speaker dependent name is recognized, a call is placed to the phone number associated with the recognized name in the customer's personal directory. In this manner, a customer can place a call by simply speaking a name. Placing a call in this manner may be thought of as issuing an “implicit command” to dial since the system of the present invention will interpret a speaker dependent name, in the absence of other commands, as a command to dial the number associated with the name.
When a speaker independent command is detected, and no speaker independent name has been detected, the command is performed unless additional input is required, e.g., a name in the case of call forwarding, or security is of concern, such as in the case of message retrieval where the phone company is providing voice mail services. When security is of concern, a voice verification step is performed and the customer's identity is verified. Accordingly, the present invention provides a flexible system where voice verification is performed on an as needed basis and not necessarily on all calls.
The system and method of the present invention supports commands which require additional inputs of either a name or another command. When such commands are detected by the speaker independent recognition step, a check is made to determine if the additional information was received as part of the same utterance in which the command was received. If it was, the command is processed accordingly. Otherwise, the needed additional input is requested and processed using speaker dependent and speaker independent speech recognition in parallel as described above.
Since both speaker dependent and speaker independent speech recognition is performed in parallel, both speech recognition steps may produce possibly valid recognition results which would result in a conflicting interpretation of the command or action to be performed. To resolve any disputes between the outcomes of the speech recognition steps, an arbiter is used to analyze the results and determine which course of action, e.g., command, is most likely the correct one. The course of action that is selected by the arbiter that will be followed in the case of potentially conflicting speech recognition results.
In order to assist the arbit
Asadi Ayman O.
Lubensky David M.
Naik Jayant M.
Raman Vijay R.
Vysotsky George J.
Hong Harry S.
Straub Michael P.
Suchyta Leonard C.
Swingle Loren C.
Telesector Resources Group Inc.
LandOfFree
Methods and apparatus for performing speaker independent... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and apparatus for performing speaker independent..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and apparatus for performing speaker independent... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3102812