JPH09288560A - Speech recognition type portable data input device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device for recognizing speech and collecting data in a limited range by providing a speech recognition device and a portable com puter for comparing sequences with each other and storing data. SOLUTION: The correct sequence and the correct data range are previously registered. The meter values of meters, which an operator reads, are divided in the speech recognition device 24 by comparing them with the registered values. When they are correct, the contents are stored in the storage device of the portable computer 26. The order of start message utterance, meter number utterance, meter value utterance, meter number utterance, meter value utterance,... meter number utterance, meter value utterance to termination message utterance is registered to be correct in the sequence of reading. The numbers are previously given to the respective meters as the divided data range and they are registered in a data base on the storage device of the portable computer 26 with the meter value range of the unit being the read object.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data input device that collects data in a normal range by determining a voice signal from a microphone of a headset with a portable computer having a voice recognition device for recognizing voice. is there.

[0002]

2. Description of the Related Art Conventionally, for the purpose of prompting an input after vocalization,
The lamp was installed in the place where the speaker could see it. As an example of this, there is a method disclosed in Japanese Utility Model Laid-Open No. 3-73051. FIG.
0 is a block circuit diagram of the above headset. The present invention relates to a headset including a receiver for a wireless communication device and a microphone, and a switch for switching an antenna circuit from a receiving circuit to a transmitting circuit when a voice signal is input from the microphone into the headset, and a transmitting circuit. A light emitting diode that emits light when switched is provided in the microphone. The headset having the above configuration merely detects the average amplitude of the input voice or estimates the time to detect whether or not the input mode of the voice.

On the other hand, it is known that a stationary large-scale computer has a device for analyzing voice and inputting data.

[0004]

Since the conventional headset is constructed as described above and only detects the voice utterance period, whether the correct input is completed by analyzing the meaning of the voice. There was a problem that I could not confirm. Further, there is a problem that a computer that correctly analyzes input of each segment of voice is not suitable for carrying.

The present invention has been made to solve the above problems, and an object of the present invention is to obtain a portable data input device incorporating a voice recognition device with a limited analysis target range.

[0006]

A voice recognition portable data input device according to the present invention has a voice recognition device for analyzing an input voice and an input voice analyzed as a correct sequence compared with a pre-registered sequence. It was equipped with a portable computer that determines the result of the determination and, if correct, stores the following input voice as data.

Furthermore, the registration range of data is determined in advance, and if the registered sequence or data deviates from this registration range, an abnormal alarm is output.

Furthermore, the abnormal alarm output is made to correspond to the lamp display.

Furthermore, the abnormal alarm output is voice output.

Furthermore, the registration range of the data is determined in advance, and when the registered sequence and the data range match, the guidance for the next input is output.

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1. A voice recognition portable data input device according to the present invention includes a voice recognition device that analyzes voice input via a microphone and recognizes voice input for each segment, and a portable computer connected to the voice recognition device. , An output unit for outputting the result from the portable computer to the operator.
Further, the portable computer is pre-registered with the correct sequence and data range, and is equipped with a comparison / judgment and storage means for comparing the pre-registration and voice input to collect correct data.

FIG. 1A is a block diagram of a handset showing an example of a concrete structure of an input part and an output part. FIG.
(B) is a diagram showing a hardware configuration of a headset and a data input device that is a portable computer that analyzes voice input from the headset, collects and stores correct input data, and outputs the result as a monitor. is there. In these figures, 10 is a pad, 11 is a mounting band, 12 is a microphone support member, 13 is a connection cord, 14 is a microphone, and 15 is LE.
A D display member, 15a is a green LED display section for normal recognition, and 15b is a red LED display section for abnormal recognition.
Reference numeral 20 is a collective representation of the headset composed of the above-mentioned elements. Reference numeral 21 is a connection cable, 22 is a voice signal cable, 23 is an LED display cable, 24 is a voice recognition device, 25 is a control device for performing LED display, and 26 is a portable computer that determines the recognition result.

A voice signal entering the microphone 14 of FIG. 1A is taken into the voice recognition device 24 through the connection cord 13 and the voice signal cable 22 of FIG. 1B. Here, the voice signal is converted into text, and the portable computer 2
6 is stored in the storage device. At the time of storage, the text content is determined by the processor in the portable computer, and a signal indicating completion of recognition is passed through the LED display control device 25 and the LED display cable 23 to the LED display unit 15a or 15
b is turned on.

With the above hardware configuration, the portable computer performs, for example, the operation shown in FIG. 3 for correct data collection.
FIG. 3 is a flow chart showing the operation of the meter calculator data reading system described in this embodiment.
This system registers the correct sequence and the correct data range in advance in this calculator, compares the registered values with the meter values of the meter read by the operator, and separates them with a voice recognition device. It is stored in a storage device in the computer. The sequence of reading is registration that the order of start message utterance-> instrument number utterance-> instrument meter value utterance-> instrument number utterance-> instrument meter value utterance->...-> instrument number utterance-> instrument meter value utterance-> end message utterance is correct I'll do it. As the data range separated as described above, each instrument is given a number in advance and is registered in the database on the storage device of the portable computer together with the meter value range of the instrument to be read. If the sequence and the data range are correct, after the reading is completed, the storage device of the portable computer stores the start message and the text data of each read instrument number / meter value.

FIG. 2A is a diagram showing the boundaries of recognition items targeted by the voice recognition device 24, and FIG. 2B is a diagram showing an example of a range of registered data values. Since only a short voice input is required to be recognized, the scale of the voice recognition device 24 may be small. The operation of the meter reading system will be described with reference to FIGS. In FIG. 3, a step (hereinafter, the name of the step is omitted) 30 is a start of processing, 31 is a voice signal input from a microphone, 32 is a voice recognition and text conversion process on a voice recognition device, 33
Is the storage of the voice text data in the storage device in the portable computer, 34 is the determination processing on the portable computer, 34
a to 34e are flows in the determination process. Further, although it depends on the configuration of the computer, in the present embodiment, the process after 33 is a process performed by the processor or the like in the portable computer 26. In this flow chart, the main steps of the data input device are the steps 34a1, 34b1, 34c1, 34d1 for comparing the delimiter of the voice input with the registered text, and the step of checking if it is within the correct sequence and data range. 34a2, 34b2, 34c3, 34d3 and steps 34c4, 34d4 for storing correct data. It should be noted that, according to each step in which the abnormality is detected in the above step, in the present embodiment, the infrared diode (LED) displays a red color in step 36.

When the recognition result of the input voice is the start message of 34a1, if it is already started at 34a2, the red LED display output of 36 for abnormality recognition is performed, and the voice signal input from the microphone of 31 is waited again. If it is not started at 34a2, that is, if it is a correct sequence, the start message is stored in the storage device at 34a3, the green LED display output of 35a for normal recognition is performed, and the sound signal input from the microphone of 31 is returned. In the case where the recognition result is the end message of 34b1, if it does not start at 34b2, the red LED display output of the abnormality recognition of 36 is performed, and the sound signal input from the microphone of 31 is returned. If it is already started in 34b2, the green LED display output of 35b for normal recognition is performed, and the process ends.

When the recognition result is the inspection instrument number of 34c1, if the previous voice content is not the start message or instrument meter value in 34c2, 36 red LED display for abnormality recognition is output and the voice signal input from the 31 microphone. Return to.
If the previous voice content is the start message or meter meter value in 34c2, the process proceeds to the determination in 34c3. If the instrument number is not registered in the instrument database of 39 in 34c3, a red LED display output of 36 for abnormality recognition is performed, and the sound signal input from the microphone of 31 is returned. If the instrument number is registered in the instrument database of 39 at 34c3, the instrument number is stored in the storage device at 34c4, the green LED display output of 35a for normal recognition is performed, and the voice signal input from the microphone of 31 is returned.

When the recognition result is the meter meter value of 34d1, if the previous voice content is not the inspection meter number at 34d2, the abnormal LED recognition output of 36 is output and the voice signal input from the microphone of 31 is returned. If the previously uttered content is the test instrument number in 34d2, the process proceeds to the determination in 34d3. 3
In 4d3, if the meter value is not within the meter value range of the meter in the meter database of 39, the red LED display output of 36 for abnormality recognition is performed, and the sound signal input from the microphone of 31 is returned. At 34d3, if the meter value is within the meter value range of the meter in the meter database of 39, the meter value is stored in the memory device at 34d4, the green LED display output of 35a for normal recognition is performed, and the voice signal from the microphone of 31 is output. Return to input.

In the other cases where the recognition result is 34e1, 36
The red LED display for abnormal recognition is output and the sound signal input from the microphone 31 is returned. The text recognition item 38 in FIG. 4A is a collection of these recognition items.
The determination of the instrument number and the meter value is performed by adding "K" and "M" to the head.

The above system is intended for voice input of a data entry system. According to the above-mentioned configuration, the voice recognition may be a simple one for a limited one, and the output is the LED display section having the same line of sight as the microphone, so that it is possible to input an immediate voice without moving the line of sight. There is an effect that the input processing speed can be improved. Further, since the input voice signal is registered as a text in the storage device on the computer, it can be used for other work such as output of a form after completion of input work and word processing / spread calculation.

A case where the present embodiment is applied to another system will be described. That is, the input voice recognition is performed, the sequence and the data range are registered in advance, the input voice is checked, and if these are correct, it is stored as data, and if outside the range, LDE display is performed. Describe the periodic inspection record system for equipment. 4 and 5 are flowcharts showing the operation in the above system. In the flow of this processing operation, the sequence registered in advance is the start message utterance →
Device name utterance → Inspection status utterance → Device name utterance → Inspection status utterance → ・ ・ ・ ・ → Device name utterance → Inspection status utterance → End message utterance. When the equipment is normal, the inspection status utterance is "normal" utterance and the completion message utterance moves to the next equipment. In the case of equipment abnormality, "abnormal" speech → inspection location speech → action content speech → completion message speech moves to the next equipment. After the device inspection is completed, the storage device of the portable computer stores a start message if the sequence is correct, and text data of each device name / inspection status if the data range is correct.

FIG. 6A shows the recognition items targeted by the voice recognition device 24, and FIG. 6B shows the registration range of the device data.
FIG. 6C is a diagram showing the registration range of the treatment content data. The processing flow of this periodic inspection record is shown in FIGS.
Will be described with reference to the flowchart of FIG. 40 is a start of processing, 41 is a voice signal input from a microphone, 42 is a voice recognition and text conversion process on a voice recognition device, 43 is an acquisition of the voice text data into a storage device in a portable computer, and 44 is Determination processing on a portable computer, 44a-
44i is a flow in the determination process. Recognition result is 44a
In the case of the start message of 1, if it is already started at 44a2, a red LED display output of 46 for abnormality recognition is performed, and the process returns to the voice signal input from the microphone of 41. If not started at 44a2, the start message is stored in the storage device at 44a3, and the green LED display output of 45a for normal recognition is performed.
Return to the audio signal input from the No. 1 microphone. Recognition result is 44
In the case of the end message of b1, if it does not start at 44b2, a red LED display output of 46 for abnormality recognition is performed and 4
Return to the audio signal input from the No. 1 microphone. If it is already started at 44b2, the green LED display output of 45b for normal recognition is output, and the process ends.

When the recognition result is the device number of 44c1, 4
If the previous voice content is not a start message or a completion message in 4c2, a red LED display output of 46 for abnormality recognition is performed, and a voice signal input from the microphone of 41 is returned to. 44
If the previous utterance content is the start message or the completion message in c2, the process proceeds to the determination in 44c3. If the device number is not registered in the device database of 39a at 44c3, a red LED display output of 46 for abnormality recognition is performed, and the process returns to the voice signal input from the microphone of 41. If the device number is registered in the device database of the device number 39a at 44c3, the device number is stored in the storage device at 44c4, the green LED display output of the normal recognition of 45a is output, and the voice signal input from the microphone of 41 is returned. If the recognition result is a normal message of 44d1, and the previous voice content is not the device name in 44d2,
The red LED display output of 46 for abnormality recognition is performed, and the process returns to the voice signal input from the microphone of 41. If the previous voice content is the device number in 44d2, a normal message is stored in the storage device in 44d3, the green LED display output of 45a for normal recognition is performed, and the voice signal input from the microphone of 41 is returned.

When the recognition result is the abnormal message of 44e1, if the previous voice content is not the device name in 44e2,
The red LED display output of 46 for abnormality recognition is performed, and the process returns to the voice signal input from the microphone of 41. If the previous voice content is the device number in 44e2, the abnormal message is stored in the storage device in 44e3, the green LED display output of 45a for normal recognition is performed, and the voice signal input from the microphone of 41 is returned. In the case where the recognition result is the inspection location of 44f1, if the previous utterance content is not an abnormal message at 44f2, 46 of the abnormal recognition red L
ED display output is performed, and the process returns to the audio signal input from the microphone 41. If the previous speech content is an abnormal message at 44f2, the process proceeds to the determination at 44f3. There are 39 inspection points at 44f3
If it is not registered in the inspection location of the device database of a, the abnormal LED recognition output of 46 is output, and the sound signal input from the microphone of 41 is returned. If the inspection location is registered in the inspection location of the equipment database of 39a in 44f3, the inspection location is stored in the storage device in 44f4, and the 45a is stored.
The green LED display output for normal recognition is output and the sound signal input from the microphone 41 is returned to.

When the recognition result is the action content of 44g1, 4
If the previous voice content is not a check point at 4g2, a red LED display output of 46 for abnormality recognition is performed, and the flow returns to the voice signal input from the microphone of 41. If it is 44g2 and the previously uttered content is the inspection point, the process proceeds to the determination of 44g3. If the treatment content is not registered in the treatment content database of 39b at 44g3, a red LED display output of 46 for abnormality recognition is performed, and the sound signal input from the microphone of 41 is returned to. If the treatment content of 44g3 is registered in the treatment content database of 39b,
The content of the treatment is stored in the storage device at 44g4, the green LED display output of the normal recognition of 45a is output, and the sound signal input from the microphone of 41 is returned.

When the recognition result is the completion message of 44h1, if the previous voice is not the normal message or the content of the treatment in 44e2, the abnormal LED recognition output of 46 is output and the voice signal input from the microphone of 41 is returned. 44e2
If the previous voice is the normal message or the content of the treatment, the completion message is stored in the storage device at 44h3, the green LED display output of the normal recognition of 45a is output, and the voice signal input from the microphone of 41 is returned. In the other cases where the recognition result is 44i1, the abnormal LED recognition output of 36 is output and the processing returns to the audio signal input from the microphone of 31. The 48 text recognition items are a collection of these recognition items, and the device number, inspection point, and determination of the treatment content are each preceded by "
This is done by adding "K", "location", and "treatment".

The above system is portable, suitable for work in places where both hands cannot be used, and can be used for outdoor periodic inspection record work such as regular inspection of utility poles. Further, since the determination process is performed by the software program on the portable computer, the reading process in each field can be customized, and the application in a wide range of fields is possible.

Embodiment 2 It may be convenient to output the abnormal alarm output and the normal input display as voice output instead of the lamp equivalent. Further, if there is an operator guidance mechanism for instructing the next operation finger point in the sequence order, the input becomes easier. In this embodiment, a system to which these are applied will be described. The notification of the voice recognition completion by the LED display to the speaker can be used as the notification to the speaker by voice response. FIG. 7 is a hardware configuration diagram of the data input device according to the embodiment. FIG. 7A is a perspective view of a voice response type headset.
Reference numeral 0 is a receiver, 51 is a mounting band, 52 is a microphone support member, 53 is a connection cord, and 54 is a microphone. Figure 7
7B is a configuration diagram of a data input device to which a headset 60 including the above-described elements described in FIG. 7A is connected. In the figure, 61 is a connection cable, 62 is a voice recognition generation device, and 63 is a portable computer that determines the recognition result.

The voice signal input to the microphone 54 of FIG. 7A is captured by the voice recognition generation device 62 by the connection cord 53. Here, the voice signal is converted into text, and the portable computer 6
It is stored in the storage device in 3. Upon storage, the content of the text is judged on a portable computer, and an alarm output when an abnormality is detected and the next input instruction of a normal sequence are generated by the voice recognition generation device 62 in reverse to generate a voice, A voice response is made from the receiver 50.

A case where the target system is the meter instrument reading system according to the first embodiment will be described. As described above, this system is one in which the operator reads the meter value of the meter and stores the voice input in the storage device in the portable computer. The meter value range and the correct sequence of each recognition target instrument are registered in the database in advance. In the present embodiment, a voice dictionary for reply is also registered in the database on the storage device in the portable computer. The correct reading sequence is the same as in the first embodiment. In the utterance dictionary for voice response, a start message, an instrument number / alphanumeric value of a meter value, and a guidance message for prompting the next input are registered.

FIG. 8 is a flow chart showing the operation of the data input device having the above-mentioned configuration, and FIG. 9 is a diagram showing an example of a voicing dictionary for guidance and the like. The flow of this processing operation will be described with reference to FIGS. 70 is the start of processing,
Reference numeral 71 is a voice signal input from a microphone, 72 is voice recognition and text conversion processing on a voice recognition device, and 73 is the voice text data taken into a storage device in a portable computer.
74 is a determination process on the portable computer, and 74a to 74e are flows in the determination process. If the recognition result is the start message of 74a1, if it is already started in 74a2, 76
Then, 78e in the reply utterance dictionary 78 is returned to the receiver of the headset, and the sound signal input from the microphone 71 is returned.
If it does not start at 74a2, the start message is
a3, the response utterance dictionary 78
78a of the signal is returned to the receiver of the headset, and the sound signal input from the microphone 71 is returned. In the case where the recognition result is the end message of 74b1, if it does not start at 74b2, 78f in the reply utterance dictionary 78 is returned to the receiver of the headset at 76, and the voice signal input from the microphone of 71 is returned. If it is already started in 74b2, 78b in the reply utterance dictionary 78 is returned to the receiver of the headset in 75b, and the process is ended.

When the recognition result is the inspection instrument number of 74c1, if the previous voice content is not the start message or the instrument meter value in 74c2, 78g in the reply utterance dictionary 78 is returned to the receiver of the headset in 76, and 71 Return to audio signal input from the microphone. If the previous voice content is the start message or meter meter value at 74c2, the process proceeds to the determination at 74c3. If the instrument number is not registered in the instrument database in 74c3, 78h in the reply utterance dictionary 78 is returned to the receiver of the headset in 76, and the voice signal input from the microphone 71 is returned. If the instrument number is registered in the instrument database in 74c3, change the instrument number to 74c4.
Then, it is stored in the storage device, and at 75c, 78c in the response utterance dictionary 78 is returned to the receiver of the headset, and the voice signal input from the microphone 71 is returned.

In the case where the recognition result is the meter meter value of 74d1, if the previous voice content is not the test meter number at 74d2, 78i in the reply utterance dictionary 78 is returned to the receiver of the headset at 76 and the microphone from 71 is used. Return to audio signal input. If the previous utterance content is the test instrument number in 74d2, the process proceeds to the determination in 74d3. If the meter value is not within the meter value range of the meter in the meter database at 74d3, 76
Then, 78j in the reply utterance dictionary 78 is returned to the receiver of the headset, and the sound signal input from the microphone 71 is returned.
If the meter value at 74d3 is within the meter value range of the meter in the meter database, the meter value is stored at 74d4 in the storage device, and at 75d, 78c in the reply utterance dictionary 78 is returned to the receiver of the headset, and the microphone at 71 is stored. Return to the audio signal input from. 76 in the other cases where the recognition result is 74e1
Then, 78k in the reply utterance dictionary 78 is returned to the receiver of the headset, and the sound signal input from the microphone 71 is returned.

According to the operation of FIG. 8, since the abnormal alarm output and the normal operation output are performed by the voice response, it is possible to know the normal end at each input division, and the input can be performed more smoothly. Furthermore, since the following utterance instruction guidance notification is provided, there is an effect that the input by the operator is further simplified. Further, although the case where the voice guidance by the voice recognition generation device is used has been described above, the guidance may be displayed by characters on the liquid crystal display unit or the like.

[0035]

As described above, according to the present invention, since the voice recognition device and the sequence comparison and data storage means are provided, there is an effect that correct target data can be efficiently collected in the portable storage section.

Furthermore, since an alarm is output when an abnormal input is made, there is an effect that the data collection efficiency is improved.

Furthermore, when there is an abnormal input, a lamp is displayed, which has the effect of increasing the data collection efficiency.

Furthermore, since a voice warning is given when there is an abnormal input, there is an effect that the data collection efficiency is improved.

Furthermore, since guidance is output, there is an effect that data can be easily input.

[Brief description of drawings]

FIG. 1 is a diagram showing a hardware configuration of a headset and a data input device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing an example of recording recognition items and an example of special recording of a data value range of the voice recognition device according to the first embodiment.

FIG. 3 is an operation flowchart of the meter instrument data reading system according to the first embodiment.

FIG. 4 is an operation flowchart of the regular inspection recording system according to the first embodiment.

FIG. 5 is an operation flowchart of the regular inspection recording system according to the first embodiment.

FIG. 6 is a diagram showing an example of registration of voice recognition items and an example of registered data in the regular inspection recording system.

FIG. 7 is a diagram showing a hardware configuration of a headset and a data input device according to the second embodiment.

FIG. 8 is an operation flowchart of the system according to the second embodiment.

FIG. 9 is a diagram showing an example of registration of a voice dictionary in the data input device according to the second embodiment.

FIG. 10 is a configuration diagram of a conventional voice level detection system.

[Explanation of symbols]

15a LED display section (normal), 15b LED display section (abnormal), 24 voice recognition device, 26 portable computer,
34a1, 44a1, 74a1 start message detection step, 34b1, 44b1, 74b1 end message detection step, 34c1, 44c1, 74c1 inspection instrument number detection step, 34d1, 74d1 instrument meter detection step, 34a2, 44a2, 74a2 start sequence investigation step, 34b2 , 44b2, 74b
2 Sequence inspection step during inspection, 34c3, 44c
3,74c3 Instrument number range detection step, 34d3
74d3 Meter value data range detection step, 34c
4,74c4 Instrument number storage step, 34d4,74
d4 Meter value storage step, 36, 46 Abnormal warning output step, 38, 48 Voice recognition item registration database, 39 Normal data range registration database, 44d
2 device sequence checking step, 44c4 device name storing step, 44d3 normal message storing step,
44e2, 44f2, 44g2, 44h2, 44f3
44g3 Inspection sequence check step, 44e3, 4
4f4, 44g4, 44h3 Data storage step, 4
9a device number registration database, 49b treatment content registration database, 62 voice recognition generation device, 75a, 7
5c, 75d, 76 Guidance output step.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G10L 3/00 561 G10L 3/00 561G 571 571H

Claims (5)

[Claims] 1. A voice recognition device for analyzing an input voice and a sequence registered in advance to determine that the analyzed input voice is a correct sequence, and if so, the subsequent input voice as data. A voice recognition portable data input device comprising a portable calculator for storing. 2. The voice recognition portable data input device according to claim 1, wherein a registration range of data is determined in advance, and when the registered sequence or data is out of the registration range, an abnormality alarm is output. . 3. The voice recognition portable data input device according to claim 2, wherein the abnormality alarm output is an output corresponding to a lamp display. 4. The voice recognition portable data input device according to claim 2, wherein the abnormal alarm output is voice output. 5. The voice recognition mobile phone according to claim 1, wherein a registration range of data is determined in advance, and when the registered sequence and data range match, guidance for the next input is output. Type data input device.