JPS5835639A - Data input device - Google Patents

Data input device

Info

Publication number
JPS5835639A
JPS5835639A JP56133531A JP13353181A JPS5835639A JP S5835639 A JPS5835639 A JP S5835639A JP 56133531 A JP56133531 A JP 56133531A JP 13353181 A JP13353181 A JP 13353181A JP S5835639 A JPS5835639 A JP S5835639A
Authority
JP
Japan
Prior art keywords
input
displayed
character
voice
characters
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
JP56133531A
Other languages
Japanese (ja)
Inventor
Shoichiro Yoshimura
吉村 昌一郎
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NEC Corp
Original Assignee
NEC Corp
Nippon Electric Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by NEC Corp, Nippon Electric Co Ltd filed Critical NEC Corp
Priority to JP56133531A priority Critical patent/JPS5835639A/en
Publication of JPS5835639A publication Critical patent/JPS5835639A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/16Sound input; Sound output

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)

Abstract

PURPOSE:To make a more accurate and high-speed data input possible, by selecting primarily characters by the input of a continuous speaking voice to display plural results on a display part and designating an objective character string in them. CONSTITUTION:The operator registers preliminarily single sounds in a processor 2. Next, he inputs data with voice through a microphone 1 in the continuous speaking system. Discrimination results of this spoken voice are displayed on a CRT 4 immediately. In this case, the recognition result of every single sound is not a single answer, and plural results are selected in order of likelihood and are arranged and displayed in accordance with likelihood. The operator indicates a correct character string out of character groups, which are displayed on the CRT 4, with a light pen 5 to select characters finally, and selected characters are displayed.

Description

【発明の詳細な説明】 本発明はデータ入力装置に関する。[Detailed description of the invention] The present invention relates to data input devices.

近年、鍵盤入力に代わる新しい入力方式として音声入力
が注目され、コンピュータ入力用にあるいはワードプロ
セッサー用に研究されつつあるが、しかしまだ本格的な
実用には至っていない状況にある。その理由の一つは、
−文字画シ又は−I!I当シの縞1梢度(または認識率
)が90ないし9〇−と低く結果の確認や、入力の訂正
1反復などを必豪とするので、従来の鍵盤入力やベンタ
ッチ入力等に比べ能率が却って低下し冥用上の効果が上
らないからである。
In recent years, voice input has attracted attention as a new input method to replace keyboard input, and is being studied for use in computer input or word processors, but it has not yet been put into full-scale practical use. One of the reasons is
-Character stroke or -I! The stripe 1 degree (or recognition rate) of this method is low at 90 to 90-, making it necessary to confirm the results and correct the input once, making it more efficient than conventional keyboard input or Bentouch input. This is because on the contrary, it decreases and the effect on the dark side does not increase.

本発明の目的は、音声入力方式の特長と欠点とをよくわ
きまえ、これを従来のベンタッチ式入力と合理的に組合
わせることにより、より正確で高速なコンビ、−タ用あ
るいはワードプロセッサー用のデータ入力方式を提供す
ることである。
An object of the present invention is to understand the advantages and disadvantages of the voice input method, and to rationally combine it with the conventional Bentouch type input, thereby achieving more accurate and faster data input for computers, computers, and word processors. The purpose is to provide a method.

本発明によれば、連続発声音声入力により、文字の一次
選択を高速に行い、これをCRTにマトリ、クス状に表
示し、次いで表示された文字群の中からベンタッチ方式
によシ目的の文字列を引自出すことにより、正確で能率
的な入力を実現することができるデータ入力装置が得ら
れる。すなわち、連続発声式音声入力部と選択可能でマ
) +3ツクス状に配列した文字表示部とを有し、第一
段階でまとまった長さのデータないしメツセージを連続
発声によシ音声入力し、そのときの各単位音声の−識結
果のうち、尤度の最も高い複数ケの文字を列に表示する
一方、一連のデータないしメツセージの各単位音声を行
に配列して文字表示部上に文字マトリ、クスを形成しs
ag2段階にて、その文字マトリックスのなかから、正
しいデータないしメツセージ列を選択することを%像と
する。
According to the present invention, the primary selection of characters is performed at high speed by continuous voice input, the selection is displayed in a matrix or box shape on a CRT, and then the desired character is selected from among the displayed characters using the Bentouch method. By extracting columns, a data input device is obtained that allows accurate and efficient input. That is, it has a continuous voice input section and a selectable character display section arranged in a +3x pattern, and in the first stage, data or messages of a certain length are inputted by continuous voice input, Among the recognition results for each unit sound at that time, the characters with the highest likelihood are displayed in a column, while each unit sound of a series of data or messages is arranged in a row and the characters are displayed on the character display area. matori, form a kusu
In step ag2, the goal is to select the correct data or message string from the character matrix.

本発明の詳細な説明に先立ち、音声入力の原理と、いく
つかの用語の足部についてまず説明する。
Prior to a detailed explanation of the present invention, the principle of voice input and some terminology will first be explained.

音声入力装置は一般にマイクロフォンと音声入力装置を
最小構成嶽累としている。砧省から発声が69た場合、
音声入力装置は通常発声f:0.5〜2秒程置装連続し
た発声単位に分解し、その発声単位毎に認識処理を行う
。各発声単位の中味にトウキ審り、オオサカ、ナゴヤ等
の単鎖でもよいし。
A voice input device generally has a minimum configuration of a microphone and a voice input device. If there is a 69 utterance from Minato Minato,
The voice input device usually divides the utterance f into continuous utterance units for about 0.5 to 2 seconds, and performs recognition processing for each utterance unit. The content of each vocalization unit may be a single chain such as Touki Jujiri, Osaka, Nagoya, etc.

イチ、二、サン等の数字でもよい、また、イ、口、ハな
どの単一の音(以下単音と云う)でもよい。
It may be a number such as 1, 2, or san, or it may be a single sound such as i, ku, or ha (hereinafter referred to as a single sound).

音声入力装置は予めそれらの標準的な発声t−登録して
おく、即ち、それらの音O特黴や形管メモリ上に記憶し
ておく、これを以vkf声パターンと呼ぶことにする1
次いで実際に発声の入力があった場合にその入力音声パ
ターンと登録された音声パターンとの類似度を比較し、
最も類似度の高い組合わせを検出することによって入力
音声の認識を行う0以上が音声入力装置の一般的なMI
l&理である。この場合音声入力処理装置は処理に先立
ち。
The voice input device registers these standard vocalizations in advance, i.e., stores them in the special memory or shape memory, which will hereinafter be referred to as a VKF voice pattern.
Next, when there is an actual voice input, the degree of similarity between the input voice pattern and the registered voice pattern is compared,
0 or more is a general MI of a voice input device that recognizes the input voice by detecting the combination with the highest degree of similarity.
It is l & reason. In this case, the audio input processing device performs the processing prior to processing.

先づ音声を各発声単位毎に切シ分けなければならないが
、これが容易に行なわれるよう、各発声単位相互間には
少くとも0.3秒81度の空白区間(ポーズ)を設ける
よう話者に義務づけるやシ方がある。これを離散方式の
音声入力装置と云っている。
First, the voice must be divided into individual vocalization units, but in order to make this easier, the speaker should create a blank interval (pause) of at least 0.3 seconds and 81 degrees between each vocalization unit. There is a way to make it compulsory. This is called a discrete voice input device.

離散方式の音声入力装置には大量く分けて2つの使い方
がおる。即ち、#述の例のように「トウキ、つ」「オオ
サカ」「ナゴヤ」等の各単語毎に登録し、これらを単位
として認識する方法と、イ。
There are two ways to use discrete voice input devices. That is, there is a method of registering each word such as "Touki, Tsu", "Osaka", "Nagoya", etc. as in the example described in # and recognizing these as a unit;

口、八等の各単音毎に登録しwt繊する方法とである。This is a method of registering each single note such as ku, 8th, etc. and making wt fibers.

str省の場合は各話の発声の所景時間に高々1秒程鼠
であるから、カナ1文字画)の所畳時間は0.2〜0.
3秒程度にな9.けん盤入力に比べ、可成り高速と云え
るが、一方では、(1)音声入力装置装置の内部のメモ
リ容量上の制約から限られた種類の言葉しか登録できな
いこと、(2)胎教が多ければ、登録されたパターンの
間相互に類似のものの組合せが多くなシ、認識の精度が
低下する。0)大量の結の登録に長時間装する等の欠点
がある。後者の方法によれば五十音だけの登録をするこ
とにより、これらの組合わせによるめらゆる讃の入力が
でき、カナ鍵盤の如き自由な入力が可能となる。
In the case of str, the duration of the utterance of each story is about 1 second at most, so the duration of one kana character stroke is 0.2 to 0.
About 3 seconds 9. It can be said that it is considerably faster than keyboard input, but on the other hand, (1) only a limited number of words can be registered due to the internal memory capacity constraints of the voice input device, and (2) if there are many prenatal If there are many combinations of mutually similar registered patterns, the recognition accuracy will decrease. 0) There are drawbacks such as the long time it takes to register a large number of knots. According to the latter method, by registering only the Japanese 50 syllables, it is possible to input a variety of hymns using combinations of these, and it becomes possible to input as freely as on a kana keyboard.

しかしこの方法の致命的な欠点は各単音を離散的に発声
せねばならぬので、発声の能率が極めて低下することで
あシ、認識nFjtの不充分さもめって。
However, the fatal drawback of this method is that each single note must be uttered discretely, resulting in extremely low utterance efficiency, and often resulting in insufficient recognition nFjt.

カナ鍵盤の能率には一歩及ばないのが現状である。Currently, it is not even close to the efficiency of the kana keyboard.

上記の欠点をカバーすべく近年連続発声方式の音声入力
装置が開発されている。連続発声方式では数ケの単位発
声が切れ切れでなく、互に連なって発声され喪場合にお
いても、それらt−ig織処理の過程でうまく切り分け
て認識する方式である。
In order to overcome the above-mentioned drawbacks, continuous voice type voice input devices have been developed in recent years. In the continuous vocalization method, even when several unit vocalizations are uttered consecutively, rather than in pieces, even in the case of mourning, they are successfully separated and recognized in the process of T-IG processing.

例えば、ト、つ、キ、!、つなと五十音だけの登録管予
めしておくことにより「トウー?、つ」の連続−語の発
声管処理し、1#!誠することが可能である。この方式
によれば前記に列挙した層数方式の2つの使い方の各々
の欠点tカッく−するかに見られるが併しなお重大な欠
点が残されている。即ち。
For example, t, tsu, ki,! , tsuna and the Japanese syllabary are registered in advance, and the continuous -word ``tou?, tsu'' is processed in the vocal tube, and 1#! It is possible to do so sincerely. According to this method, although the drawbacks of each of the two uses of the number-of-layer method enumerated above can be seen, there still remains a serious drawback. That is.

gImf11度がなお不充分であり、NgLを向上させ
るため、入力結果の確認ないし修正、追加に余計な時間
を必景とし、これが音声入力の効能を大巾に減殺してい
る。特にイ、口、I・などの単音の発声パターンVユ「
トウキ、つ」「オオサカ」「ナゴヤjなどの複合音の発
声に比し、音の変化が単調で特長がつかみずらいことか
ら%認識精度が一段と低く、ある実験によればカナー文
字当シ高々80から90s前後と云われる。「トウキ、
つ」などカナをnヶ組合せたー#当りの認識率は理鍮上
各−文字当ffm鐵率rのn乗とされるので「トウキ。
gImf 11 degrees is still insufficient, and in order to improve NgL, extra time is required to confirm, correct, and add input results, which greatly reduces the effectiveness of voice input. Especially the vocal patterns of single sounds such as i, mouth, I, etc.
Compared to the utterance of complex sounds such as ``Touki, tsu'', ``Osaka'', and ``Nagoya j'', the change in sound is monotonous and it is difficult to grasp the characteristics, so the percentage recognition accuracy is much lower. It is said to be around 80's to 90's.
The recognition rate for n combinations of kana such as "tsu" is the n-th power of the rate r for each character, so it is "tōki".

つ」の仇では(0,1)’ 4;0.59.即ち59−
でめヤ、実用上大いに問題である。
In the enemy of 'tsu'(0,1)'4;0.59. That is, 59-
This is a big problem in practical terms.

本発明においてに%音声線繊のもつ上記の利点と欠点と
を充分認識し、音声入力の利点をうまく活用し、かつそ
の欠点を従来の方式と夛わけベンタッチ方式によってう
まく補遺し、両方式の合理的な結合によシ、正確かつ能
率的な入力装置が得られる。
The present invention fully recognizes the above-mentioned advantages and disadvantages of voice input, makes good use of the advantages of voice input, and supplements the disadvantages with the conventional method and Bentouch method. Through rational combination, an accurate and efficient input device can be obtained.

次に本発明を図面全参照しなから許細に説明する。第1
図は本発明の一実施例ケ示す図でおり。
Next, the present invention will be described in detail with reference to all the drawings. 1st
The figure shows one embodiment of the present invention.

音声入力用マイクロフォン1、連続発声音声Iv!鐵処
理用プロセッサー2(以下単にプロセッサーと云う)、
通常のカナ文字用鍵盤3、ベンタッチ入力可能なCRT
ディスプレー装置4、ライトベン5、及び全体制御部6
を基本的構成要素とする。
Microphone 1 for voice input, continuous voice voice Iv! Iron processing processor 2 (hereinafter simply referred to as processor),
3 keyboards for regular kana characters, CRT capable of Bentouch input
Display device 4, light ben 5, and overall control unit 6
are the basic components.

本装置は次のように動作するよう構成される。即ち、オ
ペレータはプロセッサー2に予め単音の登録を行う、単
音の種顛としては通常50音に濁音。
The device is configured to operate as follows. That is, the operator registers a single sound in the processor 2 in advance, and the standard of the single sound is usually 50 sounds and voiced sounds.

半濁音を加えた合計78種(ジとヂ、ズとヅは同一カテ
ゴリーとして扱う)、数字10種、アルファベット26
種及びいくつかの記号の合計120徳程度を考慮すれば
よい1以上の準備ケ済ませた上でオペレータは原稿文t
−飲み上げ、マイクロフォン1を通じて音声入力を行う
。仁の際のやυ方は例えば「フルイケヤ」「カワズ」「
トビコム」「ミズノオト」の如く、各単音をいくつかつ
ないで単語とし、これを連続発声方式で入力する0本実
施例ではプロセッサー2は最大器単音までの連続発声を
処理できるものとした。従りて、長すぎる言*rz最大
が5単音になるよう適当に分けて発声すればよい0本発
明では以上のように発声された発声の諸織Wi呆t−C
RT4上に直ちに表示する。その形式は第2図の如く行
う、即ち、各単音毎の靴織結釆は単一の答でなく、その
尤度の高い順に複数ケ1−:sんで、その尤度の順に配
列表示せしめる。
A total of 78 types including semi-voiced sounds (jitoji and zutozu are treated as the same category), 10 numbers, and 26 alphabets.
After completing one or more preparations that require consideration of a total of about 120 symbols and some symbols, the operator reads the manuscript text t.
- Drink and input voice through microphone 1. For example, ``Furuikeya'', ``Kawazu'', ``Kawazu'',
In this embodiment, the processor 2 is capable of processing continuous utterances up to the maximum number of single tones, such as ``tobicom'' and ``mizu no oto'', which are input by connecting several individual sounds to form a word and using a continuous utterance method. Therefore, the word *rz that is too long should be uttered in appropriate parts so that the maximum number is 5 single tones.
Immediately display on RT4. The format is as shown in Figure 2. In other words, the shoe weave button for each single note is not a single answer, but a plurality of digits 1-:s arranged in order of likelihood. .

第2図による本実施例では尤度につき第5順位のカテゴ
リーまでを表示している。
In this embodiment shown in FIG. 2, categories up to the fifth rank are displayed in terms of likelihood.

ここで尤度について更に詳しく説明するに、音声認識装
置においては、他の一般的なパターン認識の方式と同縁
、予めカテゴリー毎に標準的なパターン(ここでに音声
パターン)1に登録しておき、入力信号がめったときに
そのパターンと登録パターンとの類似度を定量的に比軟
し、もつとも高い類似度が得られた登録パターンを決定
し、それが11するカテゴリーをもりてili!i繊結
米とするのが普通である。
To explain the likelihood in more detail here, similar to other general pattern recognition methods, in speech recognition devices, standard patterns (here speech patterns) 1 are registered in advance for each category. Then, when an input signal rarely occurs, the degree of similarity between that pattern and the registered pattern is quantitatively compared, and the registered pattern with the highest degree of similarity is determined. It is common to use i-fiber rice.

本発明でに音声認識装置に、そのような最終選択を行な
うことなく、単に類似度の高いものt−ある数(複数−
)だけ選びだしてe g ’r上に類似度の順序で配列
するものでおる。即ち音声入力で予備選択をするもので
ある。このようにや\低い順位マチのカテゴリーを併記
配列することにより正しい結果はたとえ第1順位のカテ
ゴリーとして選ばれないまでも、低い方の側位迄の中に
は必ず含まれてCRT上に表示されるであろうことはu
Rのメカニズムからも容易に推論されるところでおり。
In the present invention, the speech recognition device does not perform such a final selection, but simply selects t - a certain number (plural -
) are selected and arranged on e g 'r in order of similarity. That is, a preliminary selection is made by voice input. In this way, by arranging categories with lower rankings, even if the correct result is not selected as the first category, it will always be included in the lower rankings and displayed on the CRT. What will be done is u
This can be easily inferred from the mechanism of R.

正しい答がCRTの表示から脱落するMIfU極めて少
い。
There are very few MIfUs in which the correct answer disappears from the CRT display.

さて、第2図は本実施例において「フルイケヤ」「カワ
ズ」「トビコム」「ミズノオト」を入力した場合のCR
T表示結果の一例ケ示すものであるが本例では一文字当
p12)認識精度をや\控え目に70−と仮定している
。従って正しい文字が左漏の一文字目に出現する如度は
70%であシ、残りの30tsの確率で第2文字又は第
3文字以降に出現している。さて、一連のデータ又はメ
ツセージの入力が終った所でオペレータは%音声人力錫
塩からベンタッチ入力処理の作業に移向する。
Now, Figure 2 shows the CR when "Furuikeya", "Kawazu", "Tobicom", and "Mizunooto" are input in this example.
This is an example of the T display result, but in this example, the recognition accuracy per character is conservatively assumed to be 70-. Therefore, the probability that the correct character will appear as the first character of the left leak is 70%, and the remaining probability of 30ts is that it will appear as the second or third character. Now, when the input of a series of data or messages is completed, the operator shifts from the voice input process to the touch input process.

ここではオペレータはCRT上にマトリ、クス状に表示
された文字群の中から正しい文字列をライトベンで指示
することにより1文字の最終選択を行いこれを入力する
。即ち、第2図の○印はオペレータがライトベンにより
選択する文字を示し、史に選択操作の順序が矢印にて示
されている。このようにしてオペレータがCRTの文字
マトリ。
Here, the operator makes a final selection of one character and inputs it by indicating the correct character string from a group of characters displayed in a matrix or box shape on the CRT using a light bar. That is, the circles in FIG. 2 indicate the characters selected by the operator using the light bar, and the order of selection operations is indicated by arrows in the history. In this way, the operator can use the CRT character matrix.

クスの中から「フルイケヤ カワズ トビコムミズイオ
ト」の選択を完了したとき同文の入力が完了する。万が
一必要な文字がCRT上に見出せず、脱落しているとき
は、これを別途設けたカナ鍵盤から同文字を入力してラ
イトベンによる文字列の選択の中に割シ込ませればよい
、即ち、本発明ではデータの入力を2段に分けて行う。
When the selection of "Furuikeya Kawazu Tobicom Mizuioto" is completed from among the boxes, the input of the same sentence is completed. In the unlikely event that the required character cannot be found on the CRT and is missing, you can input the same character from a separately provided kana keyboard and insert it into the character string selection by Light Ben. In the present invention, data input is performed in two stages.

前段では音声入力により1文字の一次選択を能率よく行
い、′s択の嘱を高々数文字に絞り込み、入力データ又
はメツセージに関する文字マトリックスを表示する。欠
いで纂2段階ではライトベンによυ文字マド13ックス
の中から正しい文字列を最終的に選択するものである。
In the first stage, one character is efficiently selected by voice input, the selections are narrowed down to a few characters at most, and a character matrix related to the input data or message is displayed. In the second stage, the correct character string is finally selected from among the 13 υ character mados by Light Ben.

カナタイプライタ、更に一般的にはカナ鍵盤(フルキー
ボードとも云われる)の操作は数字だけの鍵盤(いわゆ
るテンキーボード)に比べ格段に操作能率が低いといわ
れる。それは選択すべ数文字の柚類が多く1選択に時間
を簀するためでめりて、%に釧#!を充分積んでいない
不馴れなオペレータでは平均して1分間に20文字以下
の能率でめシ、−文字画シのPj′r景時間は3秒を超
える。
It is said that operating a kana typewriter, or more generally a kana keyboard (also called a full keyboard), is much less efficient than a keyboard with only numbers (so-called numeric keyboard). It's because there are so many characters to choose, it's hard to save time in one selection, so it's % #! An inexperienced operator who does not have sufficient knowledge of the characters can write less than 20 characters per minute on average, and the Pj'r view time for character strokes exceeds 3 seconds.

本発明によれば次のような利点がめる。即ち、第一次の
文字選択は連続発声の音声入力によシ自然に近い発声速
度で行う故に、その能率は毎分汎〜goo文字%1文字
@シ0.2〜0.25秒と非常に高い、ま九第2段階の
文字の選択はCRT上に表示された文字マトリックスの
上をライトペンにより行うが、この作業もある程Wまと
まった長さのデータ又はメツセージについて連続的に行
うので非常に能率的である。実験の結果でに通常の人で
全く予備的な訓練なしで1分間当990文字即ち1文字
画シ0.7秒程度の速度で可能である0%に本文字マト
リ、クスでは目的の文字が左端の第一順位の位置にある
S度が高く、選択の能率を高める理由にもなっている。
According to the present invention, the following advantages can be obtained. In other words, since the first character selection is performed using continuous voice input at a speech rate close to natural, the efficiency is extremely high at 0.2 to 0.25 seconds per minute. The character selection in the second stage is done using a light pen over the character matrix displayed on the CRT, but this work is also done continuously for data or messages of a certain length. Very efficient. As a result of the experiment, a normal person can write 990 characters per minute, that is, one character stroke at a speed of about 0.7 seconds, without any preliminary training. The first ranking position on the far left has a high degree of S, which is also the reason for increasing selection efficiency.

また、選択の幅は本実施例では島々横に配列した5文字
であり、常にオペレータの視界に同時に這入るので、1
00文字程度が広がシをもって配列したツルキーボード
に比べ格段に選択が谷易だからである。
In addition, in this example, the range of selection is 5 characters arranged horizontally as islands, and since they always enter the operator's field of vision at the same time, 1 character can be selected.
This is because it is much easier to select than a crane keyboard in which about 00 characters are spread out and arranged in a square pattern.

従って本方式によれば第1段階と第2段階とを併せた1
文字画9所要時間rc0.9ないし0.95秒でめ9、
各段階間の切替えに多少のむだ時間は発止するもののこ
れが全体に占める割合は極めて少く無視し得るので全体
的にカナタイプライタ−等に比べてよシ速い能率で入力
操作ができることになる。しか1けん盤の配列を憶える
などの必要性がなく、素人でも殆んど予備訓練なしに速
い操作ができるところに大きな特徴がるる、従来のフル
キーボードに比べ単に連続音声入力装置とCRTなどの
表示装置を追加することによってこれだけの性能向上が
得られるので本方式はコンビ、−タの入力端末、ワード
プaセッサー、カナタイプライタ−など従来フルキーボ
ードが使われていたあらゆる分野に応用が可能でおる。
Therefore, according to this method, the first stage and the second stage are combined into one
Character stroke 9 required time rc0.9 to 0.95 seconds9,
Although some dead time occurs when switching between each stage, the proportion of this time to the total is extremely small and can be ignored, so overall input operations can be performed more efficiently than with a kana typewriter or the like. The main feature is that there is no need to memorize the layout of a single keyboard, and even an amateur can operate it quickly without any preliminary training.Compared to a conventional full keyboard, it is simply a continuous voice input device and a display such as a CRT. Because such performance can be improved by adding additional devices, this method can be applied to all fields where full keyboards have traditionally been used, such as combinations, computer input terminals, word processors, and kana typewriters. .

また最近ではカナ、洟字変換f:%徴とするワードプロ
セッサーが開発され実用化されているが、この場合のカ
ナ入力手段としても轟然応用できる。*するに本発明の
ねらいとするところUlつには連続発声の音声入力の高
速性を利用して、文字の一次選択を能率よく行うことで
めり、従っであるまとまった長さのデータないしメツセ
ージを単位として処理が行なわれることを特徴としてい
る。
In addition, recently, a word processor has been developed and put into practical use that uses kana and shuji conversion f:% characters, and it can also be used as a kana input means in this case. *One of the aims of the present invention is to make use of the high speed of continuous voice input to efficiently perform the primary selection of characters, and therefore to process large amounts of data or data. It is characterized in that processing is performed on a message-by-message basis.

また次の段階で行う最終選択についても、上記のまとま
った長さのデータないしメツセージの一次選択の結果を
文字マトリックス上にとらえ、文字マトリックス上から
連続的に文字列を引き出すことによって能率の上昇をね
らっている。既存の技術としては、マトリックスによら
ず、単−文字毎、又は−回の選択毎に選択対象文字群を
CRT上九表示し、その中から表示の都度、塩1ヶを選
択するデーI入力方式が実用された例はめるが、これt
文字毎に反復操作したのでは明らかに能率が悪く、疲労
fも高まるので本発明との差は歴然としている。
In addition, regarding the final selection to be made in the next step, efficiency can be increased by capturing the results of the primary selection of the above-mentioned data or messages on a character matrix and continuously drawing character strings from the character matrix. Aiming. The existing technology is data input, which displays nine selection target characters on a CRT for each single character or each selection, and selects one character each time it is displayed, regardless of the matrix. I will include an example of the method being put into practice, but this is
If the operation is repeated for each character, it is clearly inefficient and increases fatigue f, so the difference with the present invention is obvious.

本実施例では選択可能な文字表示部としてベンタッチ入
力可能なCRTディスプレー装置を使用したが、同目的
に対してはこのほかにも代替の手段がいろいろ考えられ
る0例えばLED液晶、プラズマディスプレーなどによ
る文字表示手段があシ、これらの単位文字表示体をマト
リックス状に配列すると共に、各表示体に選択用のスイ
ッチS。
In this embodiment, a CRT display device capable of touch-sensitive input was used as the selectable character display unit, but there are various other alternative means for the same purpose. The display means has these unit character display bodies arranged in a matrix, and each display body is provided with a selection switch S.

〜龜・を配し、第3図のように構成しても同様に目的を
達成することができるが本発明の主旨は単に選択可能な
文字群をマトリック状に配することによシ成立するもの
であって表示手段を特定するものでにない、また、第一
段階にて選択すべき文字の数は予め固定した数を決めて
おくやり方のほか、単に類似度の基準だけを決めて、こ
の基準を超える類似度のもののみを全部表示するやシ方
がめる。この場合、表示文字の数は勿論常に一定せず1
時に数が過大になシ1選択能率をおとす可能性はあるが
他方目的の文字が脱落する余地が少くなる利点がある0
本発明の具体化に当っては音声M織処理用プロセッサー
の特性を考慮し、それによく合歓した方式1に選ぶとよ
い。
Although it is possible to achieve the same objective by arranging the characters as shown in Fig. 3, the gist of the present invention is achieved simply by arranging selectable character groups in a matrix. In addition to determining the number of characters to be selected in advance as a fixed number in the first step, it is also possible to simply determine the similarity criterion. If only all the items with similarities exceeding this standard are displayed, it will be discontinued. In this case, the number of displayed characters is of course not always constant;
Although there is a possibility that the number may become too large and reduce the selection efficiency, on the other hand, there is an advantage that there is less room for the desired character to be dropped.
When implementing the present invention, it is preferable to consider the characteristics of the audio M-text processing processor and select method 1, which is well suited to the characteristics.

段階毎にまとめて処理するデータないしメツセージの適
切な長さに関しては決定的な基準にないが、ここでは次
のことが明iできる即ち、囚データ長(又はメッセージ
長)を3ないし5文字即ち−gtrtm度に燦くした場
合には各段階への切管が如禦とな9%各作業の連続性が
失われる一方、各段階間切替時に生じがちのむだ時間の
比率が無視できないものとなシ、全体的な能率が落ちる
。(jデータ長(又はメッセージ長)が長すぎる場合に
は。
Although there is no definitive standard as to the appropriate length of data or messages to be processed collectively at each stage, it is clear here that the prisoner data length (or message length) should be 3 to 5 characters, i.e. - If the speed is set to 90%, the continuity of each work will be lost by 9% due to the disconnection to each stage, but the proportion of dead time that tends to occur when switching between each stage cannot be ignored. Unfortunately, overall efficiency will decrease. (If the j data length (or message length) is too long.

#I2段pHKToりては第1段階にて飲んだデータ(
又はメツセージ)の記憶が失われてお9%鳳稿の読み直
しを必要とするので、この意味から却って能率の低下を
きたす。従って結論として上記(A)、(2)の中間に
適切な値があることが云えよう。
#I 2nd stage pHK Tori is the data taken in the 1st stage (
The memory of messages (or messages) is lost and it is necessary to reread the manuscript by 9%, which actually reduces efficiency. Therefore, it can be concluded that there is an appropriate value between (A) and (2) above.

【図面の簡単な説明】[Brief explanation of the drawing]

第1図は本発明の一実施例を示す図、第2図は本発明の
表示部の一例を示す図、第3図は本発明の表示部の他の
例を示す図。 第3 図
FIG. 1 is a diagram showing an embodiment of the present invention, FIG. 2 is a diagram showing an example of the display section of the present invention, and FIG. 3 is a diagram showing another example of the display section of the present invention. Figure 3

Claims (1)

【特許請求の範囲】[Claims] 連続発声式音声入力部と、前記音声入力部に入った音声
入力をg*し、類似度の高いlI数の結果を出力する音
声認a!部と%前記音声紹緘部からの複数の結果全表示
する表示部と、前記表示部に表示された複数の結果の1
つを指定する手段と全具備する仁とを特徴とするデータ
入力装置。
A continuous utterance type voice input unit and a voice recognition a! which converts the voice input into the voice input unit into g* and outputs the result of the lI number with high similarity. a display section that displays all of the plurality of results from the audio introduction section; and one of the plurality of results displayed on the display section.
A data input device characterized by a means for specifying an item and a data input device having all the functions.
JP56133531A 1981-08-26 1981-08-26 Data input device Pending JPS5835639A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP56133531A JPS5835639A (en) 1981-08-26 1981-08-26 Data input device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP56133531A JPS5835639A (en) 1981-08-26 1981-08-26 Data input device

Publications (1)

Publication Number Publication Date
JPS5835639A true JPS5835639A (en) 1983-03-02

Family

ID=15106972

Family Applications (1)

Application Number Title Priority Date Filing Date
JP56133531A Pending JPS5835639A (en) 1981-08-26 1981-08-26 Data input device

Country Status (1)

Country Link
JP (1) JPS5835639A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS6121526A (en) * 1984-07-10 1986-01-30 Nippon Signal Co Ltd:The Voice recognition input device

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS5377402A (en) * 1976-12-21 1978-07-08 Fujitsu Ltd Voice input device
JPS5396703A (en) * 1977-02-04 1978-08-24 Fujitsu Ltd Voice input device
JPS53147405A (en) * 1977-05-27 1978-12-22 Fujitsu Ltd Audio input unit
JPS562041A (en) * 1979-06-20 1981-01-10 Fujitsu Ltd Audio input system
JPS562039A (en) * 1979-06-20 1981-01-10 Fujitsu Ltd Audio input system

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS5377402A (en) * 1976-12-21 1978-07-08 Fujitsu Ltd Voice input device
JPS5396703A (en) * 1977-02-04 1978-08-24 Fujitsu Ltd Voice input device
JPS53147405A (en) * 1977-05-27 1978-12-22 Fujitsu Ltd Audio input unit
JPS562041A (en) * 1979-06-20 1981-01-10 Fujitsu Ltd Audio input system
JPS562039A (en) * 1979-06-20 1981-01-10 Fujitsu Ltd Audio input system

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS6121526A (en) * 1984-07-10 1986-01-30 Nippon Signal Co Ltd:The Voice recognition input device

Similar Documents

Publication Publication Date Title
US6092043A (en) Apparatuses and method for training and operating speech recognition systems
US5915236A (en) Word recognition system which alters code executed as a function of available computational resources
CN1137449C (en) Method for identifying character/numeric string in Chinese speech recognition system
TWI295783B (en) Text inputting device for mobile communication device and method thereof
JP2005078211A (en) Chinese input program
JP3995093B2 (en) Hangul character input method, Hangul character input device, Hangul character input program, and computer-readable medium
US6281884B1 (en) General Chinese phonetic keyboard setting apparatus
JPS6139152A (en) Flat character selecting system
JPS5835639A (en) Data input device
KR100657366B1 (en) Processing method and apparatus for inputting chinese character
JPH09282418A (en) Device and method for composing recognition system
KR101095864B1 (en) Apparatus and method for generating N-best hypothesis based on confusion matrix and confidence measure in speech recognition of connected Digits
JPH08512162A (en) Cursive writing analysis method
JPH049320B2 (en)
JPS58189777A (en) Character input system
KR100255640B1 (en) Character recognizing method
KR910007745B1 (en) Method for selecting flat characters
JPS61150087A (en) Recognizing device
JPH04162098A (en) Regular voice synthesizing device
CN111611773A (en) Digital coding method for Chinese and foreign languages and its use
JPS6059416A (en) Input system of japanese word information
Raper Descriptive Zulu placenames of San origin
KR20040078319A (en) Alphabet input device and method in keypad
JPS6118066A (en) Word extracting system
JPS63115264A (en) Document processor