Data processing: speech signal processing – linguistics – language – Speech signal processing – Recognition
Reexamination Certificate
1999-06-16
2001-12-04
Dorvil, Richemond (Department: 2741)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Recognition
C204S251000, C204S235000, C204S230600
Reexamination Certificate
active
06327566
ABSTRACT:
CROSS REFERENCE TO RELATED APPLICATIONS
STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT
BACKGROUND OF THE INVENTION
1. Technical Field
This invention relates to the field of computer speech recognition and more particularly to a method and system for correcting incorrectly recognized voice commands.
2. Description of the Related Art
Speech recognition is the process by which an acoustic signal received by microphone is converted to a set of text words by a computer. These recognized words may then be used in a variety of computer software applications for purposes such as document preparation, data entry and command and control.
With regard to command recognition, in the simplest possible command and control grammar, each function that the system can perform has no more than one speech phrase associated with it. At the other extreme is a command and control system based on natural language understanding (NLU). In an NLU system, the user can express commands using natural language, thereby providing total linguistic flexibility in command expression. Current command and control systems are beyond the simple one-function-one-speech-phrase grammar and are beginning to incorporate NLU.
Speech recognition systems, including NLU systems, have a difficult time accurately recognizing all the words spoken by a user. Speech recognition systems may incorrectly recognize words due to the dictation techniques and wide variety of pronunciations, accents and divergent speech characteristics of each individual speaker. For example, the speaker may speak very rapidly or softly, slur words or mumble. When transcribing speech dictation, this may result in: spoken words being converted into different words (“hold” recognized as “old”); improperly conjoined spoken words (“to the” recognized as “tooth”); and spoken words recognized as homonyms (“boar” instead “bore”). However, when controlling and navigating through speech-enabled applications by voice, incorrect recognition or non-recognition typically results in the execution of unintended commands or no command at all.
To rectify incorrectly recognized voice commands, conventional speech recognition systems include a user-initiated interface or window containing a list of possible commands. The list may be a listing of the entire speech command vocabulary, or a partial listing constrained by acoustic, language or context modeling techniques known in the art. The constrained lists are much more user friendly, since the speaker does not have to read through a lengthy list to find an intended command. These constrained lists can be generated, for example, by executing an algorithm, as is known in the art, one much like a spell checking program in word processing applications, to search a command grammar for words with similar characteristics as the incorrectly recognized words. Once the list is generated, the user may select the intended command by voice or input device. Alternatively, the user may key in the desired command in a text field within the user interface.
These command-listing methods can be effective for standard speech recognition systems, both informing the speaker of available and likely voice commands, as well as providing a simple means of executing the command. However, for NLU systems and more sophisticated systems that can recognize many hundreds of commands, command listing is impractical and cumbersome. For example, an NLU system may recognize spoken commands such as: “open the file please.”; would you kindly get the file for me?; “hey, computer, open the filet”; and “I want to see the contents of the file.” As can be seen, the phrasing can vary greatly and the numerous possible commands for performing each desired function would be too great to list.
Accordingly, there is a need to provide a quick and simple method of correcting incorrectly recognized voice commands in speech recognition systems, and for natural language understanding systems in particular.
SUMMARY OF THE INVENTION
The present invention provides a method and system for correcting incorrectly recognized commands that is particularly well-suited to natural language commands.
Specifically, the present invention operates on a computer system that is adapted for speech recognition to recognize voice commands for controlling a speech-enabled application running on the system. The method and system is performed by receiving a spoken command from a user and parsing the command to identify a paraphrased command. The paraphrased command is displayed to the user and user corrections to the paraphrased command are accepted.
The present invention provides the object and advantage of a quick and simple method of correcting incorrectly recognized commands. After being spoken, each command is parsed and a paraphrased command is generated. If the speech recognition system is unfamiliar with the spoken command, a mis-recognized command will be generated, which can subsequently be corrected by the user. Thus, the speaker will not have to re-dictate commands that are not recognized by the system. Moreover, the present invention provides another object and advantage in that it can be employed in speech recognition systems with natural language understanding capability to correct. incorrectly recognized natural language commands.
The method and system of the present invention segments the paraphrased command according to command language categories, including at least one of a command action category indicating an action to be performed;
an action object category indicating an application element subjected to the action; and a modifier category indicating modifications to the action or object. The present invention can also identify alternative voice commands for each command language segment of the paraphrased command. The paraphrased command is displayed in a user interface window in the segmented form. And, alternate commands for the recognized commands of each segment are displayed as a pop-up text window, a floating text window or a drop-down menu. The user can use an input device, such as a mouse, to select an appropriate alternative.
Yet another object and advantage is provided by the present invention in that the categorical segmenting of the recognized and paraphrased command along with the alternatives for each segment can reduce the thousands of possible word combinations in sophisticated recognition systems to several small and manageable sets. Still another object and advantage to this invention is that it provides for quick, accurate corrections, since the user can select the intended command simply by pointing and clicking without typing.
These and other objects, advantages and aspects of the invention will become apparent from the following description. In the description, reference is made to the accompanying drawings which form a part hereof, and in which there is shown a preferred embodiment of the invention. Such embodiment does not necessarily represent the full scope of the invention and reference is made therefore, to the claims herein for interpreting the scope of the invention.
REFERENCES:
patent: 5239617 (1993-08-01), Gardner et al.
patent: 5794196 (1998-08-01), Yegnanarayanan et al.
patent: 5875429 (1999-02-01), Douglas
patent: 5960384 (1999-09-01), Brash
JustVoice™ (Interactive Products Inc., “Voice Recognition for MS Windows 93,” © 1995, pp. 22-26).*
VoiceExpress™ (Lernout & Hauspie, “Installation & Getting Started Guide,” © 1992-1999, pp. 29-34).*
VoiceAssist™ (Creative Labs, “User Guide,” © 1993, pp. 4-4 to 4-8.
Lewis James R.
Ortega Kerry A.
Vanbuskirk Ronald E.
Wang Huifang
Dorvil Richemond
International Business Machines - Corporation
Nolan Daniel A.
LandOfFree
Method and apparatus for correcting misinterpreted voice... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for correcting misinterpreted voice..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for correcting misinterpreted voice... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2560937