JPH01161400A - Voice input control system - Google Patents

Abstract

PURPOSE: To eliminate the need for voice input indicating an unnecessary end when a word representing a plane name or article name is spoken and inputted and lighten the burden on an operator by sending out a code sent out by a voice input device to a host computer immediately unless the code sent out by the voice input device is a code representing a number. CONSTITUTION: A controller 14 refers to a previously set code table 16 according to an indication from a code discrimination indicating means 15, decides which of the code of a word representing a number and the code of a word representing a place name, an article name, etc., the code sent out by the voice input device 2 is, and sends the code indicating the place name or article name to the host computer 4, or the code representing the number to the host computer 4 when the codes of words are stored by a number specified previously by the host computer 4. Consequently, the code representing the place name or article is sent out to the host computer immediately without indicating the end, the throughput of a voice input system is increased, and the burden on the operator is lightened.

Description

[Detailed Description of the Invention] [Summary] This is a voice input system used in package distribution centers, etc., which allows words such as place names and product names that are mixed with numerical words to be extracted and input into a host computer. Regarding the voice input control method, the purpose is to increase the processing efficiency of the voice input system and reduce the burden on the operator. In a voice input system consisting of a control device that counts codes sent by a voice input device and sends the code to a host computer when a predetermined number of words is reached, and a host computer that instructs the control device to write a predetermined number of words, □ Instructs the device to refer to a code table indicating whether or not a code represents a number, and to check whether or not the code sent by the voice input device is a code representing a number. If the code sent by the voice input device is not a code representing a number, the code sent by the voice input device is immediately sent to the host computer.

[Industrial application field]

The present invention relates to a voice input system used in package distribution centers, etc., and in particular, a voice input system that allows words such as place names and product names that are mixed with numerical words to be extracted and input into a host computer. Regarding control method.

With recent advances in computer technology, data processing using voice has come to be performed. An example of such voice-based data processing is a package collection and delivery system, but these systems usually require the input of place names, product names, etc. into a host computer in order to sort packages by destination at a package distribution center. ing.

In this case, it is possible to manually input words such as place names and product names into the voice input device, but mainly,
Instead of uttering these place names and product names, a number string consisting of multiple digits is assigned, and by uttering this number string, the place name and product name are recognized. However, place names and product names that are frequently input by voice do not use the assigned number string, but are input by voice inputting the words of the place name or product name. , the number strings assigned to these place names and product names are input by voice.

Therefore, in addition to the voice input of words representing number strings, the voice input system includes voice input of place names and product names, but the words in the number string are predetermined numbers representing place names and product names. It is necessary to accumulate up to 1000 words before sending it to the host computer.Even if there is only one word for a place name or product name, it needs to be sent to the host computer.

Therefore, it is necessary that the voice input is not a burden to the operator.

[Conventional technology]

FIG. 4 is a block diagram illustrating the conventional technique, and FIG. 5 is a flowchart illustrating the operation of FIG. 4.

When a word is uttered, the microphone 1 sends an analog signal forming the sound pattern of the word to the filter group 5 of the audio input device 2 . The filter group 5 converts the power of each predetermined frequency of this analog signal into a digital value and sends it to the audio cutting section 6.

The voice cutting unit 6 detects the power level of the voice pattern based on a predetermined threshold value, and determines a word break corresponding to the duration of the power level below the darkness value. Then, the period during which the power level above the threshold continues is cut out as a voice pattern portion of the word, and the characteristic parameters of this cut out voice pattern portion are extracted and sent to the matching unit 7.

The matching unit 7 uses the feature parameters stored in the dictionary 8,
It compares the feature parameters sent by the voice extraction unit 6, and sends a code indicating a voice pattern having the most similar feature parameters stored in the dictionary 8 to the control device 3 as a result of the comparison.

The processor 9 of the control device 3 operates by reading the program stored in the ROM10, and calculates the number of words notified from the host computer 4 via the input/output port 13 by R.
Store in AMII. Then, as shown in FIG. 5, the code input from the voice input device 2 is monitored, and the input port 1
When the code sent by the voice input device 11!2 is received through step 2, it is stored in RAM II.

The processor 9 checks whether the code stored in the RAM II is a code instructing termination, and if it is not a code instructing termination, the processor 9 counts the number of words represented by the code stored in the RAM II, and reads the word specified by the host computer 4. Check to see if the number has been reached. If the specified number is not reached, the routine returns to monitor the input of the next code, and when the specified number is reached, the multiple codes accumulated in RAM II are sent to the host computer 4 via the input/output port 13. Send.

Furthermore, if the code input through the input port 12 is a code specifying termination, the number of words represented by the code stored in the RAM 11 has not reached the number of words specified by the host computer 4. However, the code stored in the RAMII is sent to the host computer 4 via the input/output port 13.

[Problem that the invention seeks to solve]

For example, if the place name to be input by voice is "Hachioji", the number string assigned to it is rlJ rOJ rOJ "2".
Suppose that When input as a string of numbers from the host computer 4, the number of digits in the string must be "4" in order to recognize the place name "Hachioji".
"4" is stored in I as a specified value.

As explained in FIG.
Every time the code sent by ← is input, this code is stored in RAM.
II, the number of digits of the number represented by this code,
That is, when the number of words is counted and reaches "4" specified by the host computer 4, the number rlJ rOJ rO
A code string representing J r2J is sent to the host computer 4.

However, when "Hachioji" is input by voice, since it is a - word, the processor 9 does not send it to the host computer 4 because the number of words does not reach "4". but,
Even if the word "Hachioji" is input by voice, it is possible to recognize the place name, and it is necessary to send this word to the host computer 4. It is necessary to send a specific code to instruct, for example, a code indicating termination from the voice input device 2 to the control device 3.

By the way, as mentioned above, the operator does not use the assigned number string to input the place name or product name by voice, but instead inputs the words of the place name or product name by voice. In order to send out a message, it is necessary to input a specific word, such as the end, after inputting a place name or product name, and since the extra words are input into the voice input system, the number of voice inputs increases and the host computer 4 There is a problem in that the time it takes to send the code to the computer becomes slow and it becomes a burden on the operator.

In view of these problems, the present invention has been developed so that the processor 9 of the control device 3 does not need to be instructed to terminate, for example, by voice input.
The purpose of this system is to improve the processing efficiency of the voice input system and reduce the burden on the operator by recognizing a code representing a place name or product name, and immediately sending it to the host computer 4 if the code does not represent a number. .

[Means for solving problems]

FIG. 1 is a block diagram of the principle of the present invention.

FIG. 1 shows a code identification instruction means 15 for the control device 3 of FIG.
and code table 16 have been added.

Based on the instruction from the code identification instruction means 15, the control device 14 refers to a preset code table 16 and determines whether the code sent from the voice input device 2 is a word code representing a number, a place name, a product name, etc. If the code is a code representing a place name or product name, it is sent to the host computer 4. If the code is a code representing a number, the host computer 4 sends the code to a pre-specified number of words. 1 is sent to the host computer 4 when the code is stored.

[Effect]

With the above configuration, the control device 14 compares the code sent by the voice input device 2 with the contents of the code table 16 according to instructions from the code identification instruction means 15, and determines whether it is a code representing a number or a place name. Since it is possible to determine whether the code represents a product name, the code representing a place name or product name can be immediately sent to the host computer 4 without being instructed to terminate.

〔Example〕

FIG. 2 is a block diagram of a circuit showing an embodiment of the present invention.
FIG. 3 is a flowchart illustrating the operation of FIG. 2.

The same reference numerals as in FIG. 4 indicate the same functions. Control device 1
The processor 9 of No. 4 reads a program from the ROM 10 and operates, and stores the number of words inputted from the host computer 4 through the input/output port 13 into the RAM II. Then, as shown in Fig. 3, the code input from the input port 12 is monitored, and when the code is input, the RAM
Store in II.

Here, the processor 9 determines whether the code stored in the RAM II is a code representing a number, a place name or a product name, based on instructions from the code identification instruction program 17 read from the ROM 10, or a code table preset in the RAM 11. 16.

If the code stored in the RAM II is a code representing a number, the processor 9 checks whether the host computer 4 stored in the RAM II has reached the specified number of words;
If the specified number has not been reached, the routine returns to monitor the code input through the input port 12, and when the specified number is reached, the code is stored in the host computer 4 through the input/output port 13 and stored in RAM II. Send the code.

If the code stored in RAMII is a code representing a place name or product name, the processor 9 stores the code in advance in the host computer 4.
Regardless of the number of words specified, the code stored in RAMII is immediately sent to the host computer 4 via the input/output port 13.

〔Effect of the invention〕

As described above, the present invention eliminates the need for an extra voice input indicating the end when a word representing a place name or product name is voice input, thereby reducing the burden on the operator.
Processing efficiency can be improved by eliminating the need for extra voice input to the voice input system.

[Brief explanation of the drawing]

FIG. 1 is a principle block diagram of the present invention, FIG. 2 is a block diagram of a circuit showing an embodiment of the present invention, FIG. 3 is a flowchart explaining the operation of FIG. 2, and FIG.
The figure is a block diagram explaining the conventional technique, and FIG. 5 is a flowchart explaining the operation of FIG. 4. In the figure, 1 is a microphone, 2 is an audio input device, 3.14 is a control device, 4 is a host computer, 5 is a filter group, 6 is an audio extraction unit, 7 is a collation unit,
8 is a dictionary, 9 is a processor, and 10 is a ROM. 11 is a RAM 5; 12 is an input port; 13 is an input/output port; 15 is code identification instruction means; 16 is a code table; and 17 is a code identification instruction program. The principle of the sun and the moon in the sun and the moon 2 Flow to say "T" and the moon to flow + yard grass 3 Figure half Otsu to the early 4th prisoner, Kaigo and the moon to Flora - F-tei
5 mouths

Claims (1)

[Claims] A voice input device (2) that recognizes words input by voice, converts them into codes, and sends them; and a voice input device (2) that counts the codes sent by the voice input device (2) to reach a predetermined number of words. When the code is reached, the control device (1) sends the code to the host computer (4).
4) and a host computer (4) that instructs the control device (14) to specify the predetermined number of words; a code table (16) indicating whether or not the code table (16) is present; and a code identification instruction to refer to the code table (16) to check whether or not the code sent by the voice input device (2) is a code representing a number. and an instruction means (15), and if the code sent by the voice input device (2) is not a code representing a number, the code sent by the voice input device (2) is immediately sent to the host computer (4). An audio input control method characterized by transmitting signals to