JP2005326677A - Voice memo printer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voice memo printer with which users can leave memorandums freely, without being held by preparation times in all scenes, and which can be utilized efficiently. <P>SOLUTION: The voice memory printer extracts a word, having a language feature nearest to the feature of a voice inputted from a microphone 52, from a language pattern dictionary 60 in which language features of words which are to be used in special applications are registered and outputs it as a result of voice recognition and prints it. As a result, since language features of the words to be used in the special application are registered in the language pattern dictionary 60, a voice memory printer 1, with which incorrect recognition about voice recognition of voices uttered in the special application can be suppressed as much as possible, can be realized with a simple constitution. Thus, the voice memory printer 1, with which users can leave the memorandums freely without being caught by the preparing time in all scenes, and which can be utilized efficiently, can be provided. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a voice memo printer equipped with a voice recognition function and capable of printing a voice recognition result.

  Voice recognition technology that analyzes voice digital data generated based on voice input from a microphone, etc., and converts human voice into text has been utilized in personal computers, etc., and has begun to become popular as an alternative to manual input using a keyboard Yes.

  On the other hand, there is known one that recognizes voice input from a microphone or the like and prints directly on a sheet (for example, see Patent Documents 1 and 2).

Japanese Patent Laid-Open No. 5-35428 JP-A-8-2015

  By the way, in recent years, brainstorming (Brain Stormimg Method) has become common as one of the conference methods for group ideas. In brainstorming (KJ method), the idea of the theme is written on sticky notes and small papers one after another in the initial stage, and each sticky note on which the idea is written is classified when the idea is exhausted. , Analyze themes.

  However, in conventional brainstorming in which writing is made on a sticky note each time a statement is made, the examination time is reduced due to the trouble of writing the contents of the statement on a sticky note or the like, and efficiency is lowered.

  Therefore, it is conceivable to directly print a human voice using the speech recognition technology as in Patent Documents 1 and 2 as described above, but this was conceived, for example, to create a meeting minutes. However, if it is intended to be used for brainstorming as described above, it takes a lot of labor to cut and paste the printed matter, and a reduction in efficiency cannot be denied.

  Further, according to the conventional speech recognition technology, in order to learn long sentences in advance and recognize unspecified speech without misrecognition, the device must be operated under the environment of a CPU having high performance. It is very expensive.

  An object of the present invention is to provide a voice memo printer that can easily leave a memo regardless of the creation time in any scene and can be used efficiently.

  SUMMARY OF THE INVENTION An object of the present invention is to realize a voice memo printer that suppresses erroneous recognition of voice recognition as much as possible with a simple configuration.

  The present invention relates to a microphone for inputting voice, A / D conversion means for converting voice analog data inputted from the microphone into voice digital data, and frequency conversion of the voice digital data converted by the A / D conversion means. Frequency analysis means for analyzing the voice, speech recognition means having a language pattern dictionary for specific applications, and printing means for printing the speech recognition result output from the voice recognition means based on the frequency analyzed by the frequency analysis means And comprising.

  Therefore, the word having the language feature closest to the acoustic feature of the voice input from the microphone is extracted from the language pattern dictionary, output as a voice recognition result, and printed. As a result, the language features of words used for specific purposes are registered in the language pattern dictionary, so a simple configuration of a voice memo printer that minimizes misrecognition of speech recognition for voices issued for specific purposes Can be realized.

  According to the present invention, since a language feature used for a specific application is registered in the language pattern dictionary, a simple configuration of a voice memo printer that suppresses misrecognition of voice recognition for a voice emitted for a specific application as much as possible. Therefore, it is possible to provide a voice memo printer that can easily leave a memo in any scene regardless of the creation time and can be used efficiently.

  An embodiment of the present invention will be described with reference to FIGS.

  Here, FIG. 1 is an external perspective view showing the voice memo printer 1 according to an embodiment of the present invention from the label discharge side, FIG. 2 is an external perspective view showing the voice memo printer 1 from the operator mounting side, and FIG. FIG. 2 is a horizontal sectional view showing the internal structure of the printer 1.

  As shown in FIGS. 1 to 3, a printer main body 1a of a voice memo printer 1 which is a portable portable printer includes a case 2 with one side opened, and a cover 3 for opening and closing the opened side of the case 2. It is made up of. The cover 3 is rotatably supported by a fulcrum shaft 4 provided on the case 2. The case 2 is formed with a hopper 6 for storing the long recording paper 5 wound in a roll shape with the cover 3 closed. In the present embodiment, the recording paper 5 is obtained by attaching a large number of labels 5b to the mount 5a at equal intervals, but other recording paper may be used. The label 5b is coated with adhesive having a weak adhesive force, and can be used as a sticky note P (see FIG. 7) after the printing is issued.

  In such a case 2, a paper guide 7 extending from the bottom of the hopper 6 toward the cover 3 is provided. A portion of the paper guide 7 near the cover 3 has a rotatable platen 8 and A label peeling body 9 along the longitudinal direction of the platen 8 is disposed.

  As shown in FIG. 3, a head support 11 having a thermal head 12 is provided on the inner surface of the cover 3 (on the hopper 6 side) so as to be rotatable around a support shaft 11a. The head support 11 is biased in one direction by a leaf spring 13, and the thermal head 12 comes into contact with the platen 8 with the cover 3 closed. That is, the printing unit 14 is formed by the platen 8 and the thermal head 12.

  Further, on both sides of the free end side of the cover 3, pinch rollers 16 that are pressed against the platen 8 by the urging force of the spring 15 are rotatably provided. Further, the cover 3 is formed with a label discharge port 17 that is disposed between the thermal head 12 and the pinch roller 16 and discharges the label 5b, and a paper presser 18 that presses the floating of the recording paper 5 in the hopper 6. ing. The case 2 is formed with a mount discharge port 19 for discharging the mount 5 a between the free end of the cover 3.

  On the upper surface of the case 2, a power switch 20 that declares ON / OFF of power supply from the battery 10 (see FIG. 3), a feed switch 21 that performs printing on the label 5 b, a lid 22, and a light receiving window that receives infrared rays. 23 is provided. The lid portion 22 opens and closes when the battery 10 is attached to and detached from the battery storage portion 30 (see FIG. 3) provided to be opened on the upper surface that is one surface of the case 2. Further, locking claws 24 are slidably provided on both sides of the cover 3 (see FIG. 1). These locking claws 24 are urged outward to be locked to the case 2, and when the cover 3 is opened, the locking claws 24 are slid inward as indicated by the arrow marks to engage with the case 2. Release the stop state.

  A built-in microphone 52 is provided on the same surface of the case 2 as the label discharge port 17. The voice memo printer 1 according to the present embodiment is equipped with a voice recognition function, and the built-in microphone 52 is used when executing the voice recognition function.

  In addition, an LED 56 is disposed on the upper surface of the case 2. The voice memo printer 1 of the present embodiment can notify the operator of the operation state of the voice memo printer 1 by turning on or blinking the LED 56.

  Further, as shown in FIG. 2, an arc surface 25 is formed on one surface of the printer body 1a opposite to the cover 3 so as to be in close contact with the operator's waist. On the other hand, a non-slip 26 for reducing slippage and a belt hook 27 which is hooked on the operator's belt so as to face the non-slip 26 are formed.

  With such a configuration, when the battery 10 is correctly stored in the battery storage unit 30, the terminal of the battery storage unit 30 and the terminal of the battery 10 come into contact with each other while the power switch 20 is ON. Thus, power is supplied from the battery 10 to each part such as the thermal head 12 that requires power supply.

  In such a voice memo printer 1, when the recording paper 5 is set, the cover 3 is opened, the recording paper 5 wound in a roll shape is stored in the hopper 6 of the printer main body 1a, and the cover 3 is opened. In this state, the front end of the recording paper 5 is pulled out to a position covering the platen 8 and the label peeling body 9 and the cover 3 is closed. As a result, as shown in FIG. 3, the leading end portion of the mount 5a of the recording paper 5 is pressed against the platen 8 by the thermal head 12 and the pinch roller 16, and the pull-out path of the mount 5a is formed by the label peeling member 9. The recording paper 5 is bent at an acute angle and the recording paper 5 is prevented from floating from the bottom surface of the hopper 6. The printer main body 1a on which the recording paper 5 is set can be used by placing it on a desk, but it can also be used even when it is usually worn on the operator's waist.

  Next, the connection of the control system of each part of the voice memo printer 1 will be described with reference to FIG. The voice memo printer 1 includes a CPU (Central Processing Unit) 41 that centrally controls each unit. The CPU 41 stores a ROM (Read Only Memory) in which fixed data such as a program executed by the CPU 41 is written. 42, a RAM (Random Access Memory) 43 in which variable data such as work data is renewably written, and a flash memory 44 for registering various information are connected via a bus line 45. Then, a thermal head driver 46 for driving the thermal head 12, a motor driver 48 for driving a motor 47 connected to the platen 8, a sensor circuit 50 to which various sensors 49 are connected, and a cover open that is turned on and off by opening and closing the cover 3. A switch circuit 54 to which the switch 51, the power switch 20 and the feed switch 21 are connected, an infrared interface 55, and a lighting control circuit 57 to which the LED 56 is connected are connected to the CPU 41. As described above, the circuit shown in FIG. 4 is formed on a substrate (not shown) provided inside the printer main body 1a. The infrared interface 55 is disposed inside the light receiving window 23 described above. The interface is denoted as I / F in the figure.

  The CPU 41 is connected with a voice input CODEC 53. A built-in microphone 52 is connected to the audio input CODEC 53. The audio input CODEC 53 functions as an A / D conversion unit, converts audio analog data input from the built-in microphone 52 into audio digital data, and outputs the audio digital data to the CPU 41.

  Further, a speech recognition engine 58 is connected to the CPU 41. The voice recognition engine 58 analyzes voice digital data input from the built-in microphone 52 and generated by the voice input CODEC 53, and converts human voice into text. Such a speech recognition engine 58 includes, for example, an acoustic dictionary 59 in which acoustic features (phonemes) of small units (phonemes) of human speech are registered, and a language pattern dictionary 60 in which language features of words to be recognized are registered. Voice recognition is performed using.

  The words to be recognized by speech registered in the language pattern dictionary 60 of the present embodiment are limited to specific applications. Since a specific word is often issued in a specific application, the language pattern dictionary 60 can be configured at low cost by registering only the words focused on the specific application in the language pattern dictionary 60 in this way. . Specifically, a list of conversations and utterance languages that will be used in the application to be used is collected and used as a use language table (not shown) for each application. For each language registered in this use language table for each application, the frequency of the language is analyzed and separated into speech features (phoneme information) and language features (phoneme series information). The language features separated in this way are registered in the language pattern dictionary 60.

The acoustic dictionary 59 is used as a dictionary related to voice recognition in general, not by use. The principle of speaking is
(1) “My throat trembles”
(2) “Passing through oral cavity / nasal cavity”
Therefore, the sound dictionary 59 stores information (acoustic features (phonemes) of small units (phonemes) of human utterance) that specify the shapes of (1) and (2) from the frequency of the voice.

  As shown in FIG. 5, the speech recognition engine 58 having such a configuration performs frequency conversion on speech digital data input from the built-in microphone 52 and generated by the speech input CODEC 53 by a frequency analysis unit 58a which is a frequency analysis means. The comparison unit 58b extracts acoustic features based on the acoustic dictionary 59 (speech feature extraction means). At this stage, only the shapes (1) and (2) described above can be specified, and any of the 50 sounds cannot be specified yet. Therefore, from the words registered in the language pattern dictionary 60, a word whose language feature is closest to the acoustic feature of the input speech is searched for and output as a speech recognition result (language feature extraction means). By comparing with the language pattern dictionary 60 in this way, “Aiueo” or the like can be specified for the first time. It is difficult to specify when a large number of unspecified words are uttered, but if you try to narrow down to words that are uttered for a specific purpose and compare the relationship of (1) and (2) above and the characteristics of the phoneme sequence waveform, The possibility of misrecognition can be suppressed as much as possible, and speech recognition is possible with such a simple mechanism.

  The language pattern dictionary 60 can be updated from an external device by connecting an external device (such as a personal computer) (not shown) to the voice memo printer 1 via the infrared interface 55. Further, the contents of the language pattern dictionary 60 are changed by exchanging the language pattern storage chip (dictionary) for storing the language pattern dictionary 60 or rewriting the registered contents by downloading the language pattern dictionary 60 from an external device (such as a personal computer) (not shown). It is also possible to change for each specific application. When a new language pattern dictionary 60 is downloaded via the infrared interface 55, the old language pattern dictionary 60 is deleted.

  Next, of the functions that the control program stored in the ROM 42 built in the voice memo printer 1 causes the CPU 41 to execute, the characteristic functions provided in the voice memo printer 1 of the present embodiment will be described.

  Here, the voice printing process executed by the CPU 41 of the voice memo printer 1 will be described. FIG. 6 is a flowchart showing the flow of the voice printing process. As shown in FIG. 6, when a digitized voice is input (Y in step S1), the process proceeds to step S2, and it is determined whether a recognition pattern registration process or an utterance voice recognition process.

  If it is determined that the speech recognition processing is utterance, the speech recognition processing by the speech recognition engine 58 is executed (step S3).

  If it is determined in the speech recognition process that the word is registered in the language pattern dictionary 60 (Y in step S4), the word whose language feature is closest to the acoustic feature of the input speech is searched and printed as a speech recognition result. The data is output to the section 14 and printed (step S5: printing means). Here, FIG. 7 is a plan view showing an example of the issued sticky note P. FIG. As shown in FIG. 7, the sticky note P is printed with the text “idea of XXX” when “speaking of an idea of XXX”.

  If it is determined in the voice recognition process that the word is not registered in the language pattern dictionary 60 (N in step S4), the process returns to step S1 without performing voice recognition.

  On the other hand, if it is determined that the process is a recognition pattern registration process, the process advances to step S6 to execute a recognition pattern registration process. In the recognition pattern registration process, conversations and utterance languages that will be used in the intended use are summarized in a list, and a use language table (not shown) for each use is registered, and each language registered in this use language table for each use Each time, the frequency of the language is analyzed and separated into speech features (phoneme information) and language features (phoneme sequence information). Then, the language features separated in this way are registered in the language pattern dictionary 60.

  Such a voice memo printer 1 can be used in any scene. For example, in brainstorming, when an idea is put out, it is issued in a state where the idea uttered is printed on the sticky note P just by uttering the idea. I'll do it. Also, when an order is received at a bento store or the like, the ordered product is issued on the sticky note P simply by speaking about the ordered product. It can be used as an order form. If the order slip is affixed to the product when the product is delivered, it is possible to prevent the product from being mixed.

  As described above, according to the present embodiment, the word having the language feature closest to the feature of the voice input from the built-in microphone 52 is extracted from the language pattern dictionary 60 and output as the voice recognition result for printing. Is done. As a result, the language features of the words used in the specific application are registered in the language pattern dictionary 60. Therefore, the voice memo printer 1 that suppresses the misrecognition of the voice recognition of the voice generated in the specific application as much as possible can be simplified. Since it can be realized with a simple configuration, it is possible to provide a voice memo printer 1 that can easily leave a memo regardless of the creation time in any scene and can be used efficiently.

  In the present embodiment, the single language pattern dictionary 60 that registers the linguistic features of words for specific applications to be recognized by speech is provided. However, the present invention is not limited to this. A plurality of language pattern dictionaries 60 each registering the language characteristics of the word may be provided. In this case, the language pattern dictionary 60 may be switched and used for each specific application. The language pattern dictionary 60 may be switched according to the input voice content or may be switched by a switch.

1 is an external perspective view showing a voice memo printer according to an embodiment of the present invention from a label discharge side. It is an external appearance perspective view which shows a voice memo printer from the operator mounting side. It is a horizontal sectional view showing the internal structure of the voice memo printer. It is a block diagram which shows the connection of the control system of each part of a voice memo printer. It is a block diagram which shows the structure of a speech recognition engine. It is a flowchart which shows the flow of an audio | voice printing process. It is a top view which shows an example of the issued sticky note paper.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Voice memo printer, 52 ... Microphone, 53 ... A / D conversion means, 58a ... Frequency analysis means, 59 ... Acoustic dictionary, 60 ... Language pattern dictionary

Claims (5)

A microphone for voice input,
A / D conversion means for converting audio analog data input from the microphone into audio digital data;
Frequency analysis means for frequency-converting and analyzing the audio digital data converted by the A / D conversion means;
A voice recognition means having a language pattern dictionary for specific applications;
Printing means for printing a voice recognition result output from the voice recognition means based on the frequency analyzed by the frequency analysis means;
Voice memo printer equipped with. A microphone for voice input,
A / D conversion means for converting audio analog data input from the microphone into audio digital data;
Frequency analysis means for frequency-converting and analyzing the audio digital data converted by the A / D conversion means;
A language pattern dictionary with linguistic features for specific uses;
Language feature extraction means for selecting one of the language pattern dictionaries based on the frequency analyzed by the frequency analysis means;
Printing means for printing the speech recognition result output by the language feature extraction means;
Voice memo printer equipped with. The language pattern dictionary is prepared for each specific application and can be exchanged according to the specific application.
The voice memo printer according to claim 1. A plurality of language pattern dictionaries are prepared for each specific application, and the language pattern dictionary is switched according to the specific application.
The voice memo printer according to claim 1. The registered contents of the language pattern dictionary can be rewritten from an external device.
The voice memo printer according to any one of claims 1 to 4.

Images