JPS605337A - Voice inputting system - Google Patents

Abstract

PURPOSE:To reduce the frequency of wrong recognition at a CPU, by blocking the transmission of a recognized result to the CPU when the decline of the recognition rate is expected. CONSTITUTION:The recognizing operation of the voice inputting system of this invention is performed when a voice inputter inputs a voice through a microphone MIC. At this time, the difference between the recognized result and a standard value is evaluated by using a numeral ''score'' at a comparison processing section CP. Plural candidate words are fetched from the high order of the score in the order of the 1st candidate, 2nd candidate, -. A display device DSP displays the names of the 1st and 2nd candidates. The candidate name is correlated with corresponding tables of stored words and candidate names. The corresponding tables are installed to the inside of a signal pass control device PTC. Since the absolute value of the score is known in advance when the device is manufactured and increase in the recognition rate cannot be expected when the score of the 1st candidate is not quite high, a gate circuit with a threshold level is installed to the signal pass control device PTC and the 1st candidate is controlled so as to inhibit the pass of the 1st candidate to a CPU. Therefore, no wrongly recognized signal is transmitted to the CPU.

Description

DETAILED DESCRIPTION OF THE INVENTION (1) Technical Field of the Invention The present invention relates to a voice input method that can control voice input words with a low recognition rate not to be transmitted to a central processing unit.

(2) Prior art and problems The voice input device shown in Fig. 1 uses a microphone MIC, etc. to form a voice pattern in the input processing unit PU based on information input by the voice input person. The standard pattern stored in the storage unit MM as the input voice is compared with the comparison processing unit CP to calculate the distance. As a result, the word determined to have the smallest distance is notified to the central processing unit CPU, for example, in the form of a stored word number. At this time, even if a relatively similar word was not found as a result of the comparison, the word that was determined to be the most similar among them was used as the voice input word, resulting in many misrecognitions.

(3) Object of the Invention An object of the present invention is to improve the above-mentioned drawbacks and to provide a voice input method that can control voice input words with a low recognition rate not to be transmitted to the central processing unit.

(4) Structure of the Invention The structure of the present invention to achieve the above-mentioned object is to calculate the distance between the input cover pattern and the standard pattern when voice is input, select a plurality of candidates from the one with the smallest distance, A device that outputs a score indicating the quality of recognition calculated based on the stored word number and distance, a signal passage control device inserted into the line that transmits the recognition result information to the central processing unit, and a device that outputs a score indicating the recognition quality calculated based on the stored word number and distance. and a control device capable of controlling passage of the signal passage control device, and the signal passage control device is controlled so as to pass only when the value exceeds a built-in threshold level, or the voice input person Passage control is performed by looking at the display.

(5) Embodiment of the Invention FIG. 2 is a block diagram showing the configuration of an embodiment of the invention. The same reference numerals as in FIG. 1 indicate similar parts. I)S
P includes a display device capable of displaying candidate names, PTC a signal passage control device, and K13 a manual control device that can be operated by a voice input person using a keyboard L. When a voice input person inputs from the microphone MIC, the recognition operation is performed in the same manner as in the conventional case. At that time, it is known that the comparison processing unit cp evaluates the difference from the standard value using a number called a "score." For example, a score of 200 indicates a complete match, a value smaller than that indicates a poor recognition degree, and a score of O indicates a complete match. A plurality of candidate words can be extracted from the top of the score, such as a first candidate and a second candidate. The display device DSP displays the first candidate name and the second candidate name as shown in FIG. This candidate name is a name uniquely defined for the stored word number, and the degree to which the candidate name is displayed is determined by indexing the correspondence table of stored word numbers and candidate names. This correspondence table is built into the signal passage control device 1)TC. The recognition rate is highest when the first candidate is close to the old point and the second candidate has a small value and the score difference is large, and when the first candidate is not very threshold and the score difference with the second candidate is small. Recognition rate is low. The absolute value of the score is known in advance when the device is molded, so if the first candidate's score is less than 120,
Since there is no hope of improving the recognition rate, a gate circuit with a threshold level is provided in the signal passage control device 1) TC, and a zero limit is applied to prohibit the passage of the signal to the central processing unit CPU. As a result, the erroneously recognized signal is not transmitted to the central processing unit CI)U. In addition, when the voice inputter looks at the display device I) S P immediately after inputting,
If the first candidate is above the threshold level, only the first candidate is displayed, otherwise multiple candidates are displayed and can be confirmed. If there is no recognition level near the threshold F level and the score difference between the second candidate is small, it is possible that the second candidate is legitimate. The comparator CP and the signal 3M overcontrol device PTC are controlled as necessary so as to pass only those selected from a plurality of candidates.

Note that if the score is less than or equal to the threshold level, the recognition rate is considered to be high and passage will not be prohibited.

(6) Effects of the Invention In this way, according to the present invention, when the recognition rate is expected to be low, the recognition result is not transmitted to the central processing unit, or the input person operates it to make it more legitimate. Since similar objects are transmitted, there is an effect of reducing misrecognition in the central processing unit.

[Brief explanation of the drawing]

FIG. 1 is a diagram for explaining a conventional voice input method. FIG. 2 is a block diagram showing the configuration of an embodiment of the present invention. FIG. 3 is a diagram showing an example of the display of the display device in FIG. 2. . MI C-m-Microphone MM-Storage section cp--Comparison processing section MPU-Control section KB-Keyboard DSP-Display device PTC-Signal passage control device IPU-Human power processing section Patent applicant Fujitsu Limited Agent People Patent Attorney Eisuke Suzuki

Claims (1)

[Claims] A device that calculates the difference between an input speech pattern and a standard pattern and outputs word numbers for multiple candidates and a score indicating the quality of the recognition results, and a line that transmits recognition result information to the central processing unit. The signal passage control device is equipped with a signal passage control device inserted into the device, and a control device that allows an audio input person to control the passage of the signal passage control device, and the signal passage control device is equipped with a built-in thread/control device.
A voice input method characterized in that the voice input method is controlled so as to pass only when the value exceeds 7 points, or the voice input person controls the passage by looking at the candidate name display.