JPS59147400A - Voice recognition equipment - Google Patents
Voice recognition equipmentInfo
- Publication number
- JPS59147400A JPS59147400A JP58021046A JP2104683A JPS59147400A JP S59147400 A JPS59147400 A JP S59147400A JP 58021046 A JP58021046 A JP 58021046A JP 2104683 A JP2104683 A JP 2104683A JP S59147400 A JPS59147400 A JP S59147400A
- Authority
- JP
- Japan
- Prior art keywords
- dictionary
- section
- words
- recognition device
- voice
- 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
Links
Abstract
(57)【要約】本公報は電子出願前の出願データであるた
め要約のデータは記録されません。(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.
Description
【発明の詳細な説明】 孜朱分黒 本発明は、音声入力を識別するための装置に関する。[Detailed description of the invention] Keishu Bunkuro The present invention relates to a device for identifying audio input.
皿米肢貢
音声認識装置において、単語を登録しておく辞書部は、
該音声認識装置の認識率を左右する重要な部分である。In the voice recognition device, the dictionary section that registers words is
This is an important part that affects the recognition rate of the speech recognition device.
而して、音声認識装置においては、人の発声の仕方が刻
々と変化することから、認識が行われた単語に関して刻
々と辞書に登録されたパラメータを更新し、認識率を下
げないようにする方法がとられている。しかし、辞書が
更新されるためには認識されることが必要で、非常に使
用頻度が低い単語では殆んど更新されないことになり、
誤認識の原因となる。In a speech recognition device, since the way a person speaks changes every moment, the parameters registered in the dictionary are updated every moment regarding the recognized word to prevent the recognition rate from decreasing. A method is being taken. However, in order for the dictionary to be updated, it must be recognized, and words that are used very infrequently are rarely updated.
This may cause misrecognition.
１−一灯
本発明は、上記従来技術の欠点を解決するためになされ
たものであり、特に、入力音声によって辞書を書き替え
る際、突発的変動の影響を減じ、かつ、使用頻度の低い
単語でも十分更新できる学習方式を採用した音声認識装
置を提供するものである。1-1 Lamp The present invention was made in order to solve the above-mentioned drawbacks of the prior art.In particular, when rewriting a dictionary based on input speech, the present invention is intended to reduce the influence of sudden changes, and to reduce the influence of sudden changes, and to reduce the effects of sudden changes in words that are used infrequently. However, the present invention provides a speech recognition device that employs a learning method that can be updated sufficiently.
璽−一戒
本発明の構成に関し、以下、−・実施例に裁づいて説明
する。The structure of the present invention will be described below with reference to embodiments.
本発明の理解を容易にするために、最初に、通常の音声
認識装置について説明すると、通路の７４声認識装置に
おいては、まず、認識装置を起動し、フロッピーディス
ク等に記録さイ′した！４ｉＷｒＪ ｉｆｆ ｆｆｉを
荻置のメモリーへロードさせる。その後、認識装置とし
て使用され、メモリー内の辞書の使用頻度に応じて更新
される。装置の使用が終るとこのメモリー内の辞書は前
記記録媒体中へ格納され、装置は停止される。これによ
って、フロッピー内の特定の単語は刻々と更新されるが
、使用されなかった単語は更新されない欠点がある。本
発明は、このような欠点を解決するためになされたもの
で、特に、装置の使用が終ってメモリー内の辞書を記録
媒体中へ移す際のセーブの仕方に特徴がある。In order to make the present invention easier to understand, a typical speech recognition device will first be explained.In the 74-voice recognition device in the passage, first, the recognition device was started and recorded on a floppy disk or the like. 4iWrJ iff ffi is loaded into Ogi's memory. It is then used as a recognizer and updated according to the usage frequency of the dictionary in memory. When the use of the device is finished, the dictionary in this memory is stored in the recording medium and the device is stopped. As a result, specific words in the floppy disk are updated every moment, but unused words are not updated. The present invention has been made to solve these drawbacks, and is particularly characterized by the way the dictionary in the memory is saved when transferred to a recording medium after the use of the device.
而して、本発明においては、まず、メモリー内の辞書と
フロッピーに記録されている辞書の対応する単語を照合
比較し、その際に、両者に差があればフロッピーの対応
単語辞書上へメモリー内の特徴列を移し書き替え、両者
に差がなければ、その単語はメモリーにロードされてか
ら装置の使用が終るまでに更新されておらず、つまり、
使用されなり＼つだことになるので、その旨を表示など
により使用者に伝達して発声を促し、−回の使用時に全
ての単語の更新をした上で記録媒体へ記録するようにし
たものである。勿論、−単語毎に記録媒体へ移しても良
いし、全ての単語を更新した上で記録媒体へ移しても良
い。Therefore, in the present invention, first, the corresponding words in the dictionary in the memory and the dictionary recorded on the floppy disk are compared, and if there is a difference between the two, the words are transferred to the corresponding word dictionary on the floppy disk. If there is no difference between the two, the word has not been updated between the time it was loaded into memory and the end of use of the device, that is,
Since it will no longer be used, the user is notified of this through display, etc., encouraging the user to speak, and when the word is used, all words are updated and recorded on the recording medium. It is. Of course, the information may be transferred word by word to the recording medium, or all words may be updated and then transferred to the recording medium.
第１図は、本発明の一実施例を説明するための動作説明
図、第２図は、信号の流れ図で、図中、■はマイクロフ
ォン、２ば特徴抽出部、３は辞書部、４はフロッピーデ
ィ入り、５は照合部、６は判定部、７，７′は表示部、
８は比較部で、装置の使用に先たち、まず、フロッピー
ディスク４内に記録された各人の辞書を辞書部３にロー
ドする（信号の流れ■）。次に、マイク１からオペレー
ターの声を入力すると、その音声は特徴抽出部２へ入る
（信号の流れ■）。特徴量に変換された音声は照合部５
へ行き（信号の流れ■）、そこで、辞書部３から伝達さ
れる辞書登録単語（信号の流れ■）と照合される。その
結果は、判定部６へ伝達され（信号の流れ■）、認識結
果が表示部７にて表示されるとともに（信号の流れ■）
、辞書部３へ伝達され（信号の流れ■）で辞書の更新を
行う。これを必要回数だけ繰り返した後に、装置の終了
を指示する。終了に際し、辞書部３の単語を比較部８へ
伝達しく信号の流れ■）、ここで、フロッピーディスク
４内の更新前の示書単語と比較しく信号の流れ■）、両
者が一致しているものは、使用されず更新されなかった
単語として表示され、オペレータに発声を促す。オペレ
ータの発生で辞書部３の未更新単語は全て更新され（信
号の流１１［相］）、更新された後、辞書部３の特徴量
はフロッピーディスク４に記録され（信号の流れ■）、
斯様にして、古い辞書は全て書き換えられる。なお、言
うまでもないことであるが、オペレータの判断によって
本発明による装置を動作させるか否かのスイッチを設け
ることは好ましいことである。FIG. 1 is an operation explanatory diagram for explaining an embodiment of the present invention, and FIG. 2 is a signal flow diagram. In the figure, ■ is a microphone, 2 is a feature extraction section, 3 is a dictionary section, and 4 is a 5 is a verification section, 6 is a judgment section, 7, 7' is a display section,
Reference numeral 8 denotes a comparison section, which first loads each person's dictionary recorded on the floppy disk 4 into the dictionary section 3 before using the device (signal flow ■). Next, when the operator's voice is input from the microphone 1, the voice enters the feature extraction unit 2 (signal flow ■). The audio converted into features is sent to the verification unit 5.
(signal flow ■), where it is compared with dictionary registered words transmitted from the dictionary section 3 (signal flow ■). The result is transmitted to the determination unit 6 (signal flow ■), and the recognition result is displayed on the display unit 7 (signal flow ■).
, and is transmitted to the dictionary section 3 (signal flow ■) to update the dictionary. After repeating this a necessary number of times, an instruction is given to shut down the device. At the time of completion, the word in the dictionary section 3 is transmitted to the comparison section 8, and the signal flow (■) is compared with the written word in the floppy disk 4 before the update. The words are displayed as unused and unupdated words, prompting the operator to say them. All unupdated words in the dictionary section 3 are updated by the operator's input (signal flow 11 [phase]), and after being updated, the feature quantities in the dictionary section 3 are recorded on the floppy disk 4 (signal flow ■).
In this way, all old dictionaries will be rewritten. It goes without saying that it is preferable to provide a switch that allows the operator to operate the apparatus according to the present invention at his/her discretion.
紘−一来
以上の説明から明らかなように、本発明によると、常に
高い認識率を維持することのできる音声認識装置を提供
することができる。As is clear from the above explanation, according to the present invention, it is possible to provide a speech recognition device that can always maintain a high recognition rate.
第１図は１本発明の一実施例を説明するための動作説明
図、第２図は、信号の流れ図である。
■・・・マイクロフォン、２・・・特徴抽出部、３・・
・辞書部、４・・・フロッピーディスク、５・・・照合
部、６・・・判定部、７，７′・・・表示部、８・・・
比較部。FIG. 1 is an operation explanatory diagram for explaining one embodiment of the present invention, and FIG. 2 is a signal flow diagram. ■...Microphone, 2...Feature extraction unit, 3...
- Dictionary section, 4... Floppy disk, 5... Verification section, 6... Judgment section, 7, 7'... Display section, 8...
Comparison section.
Claims (1)
する特徴抽出部と、それらを比較照合する類似度計算部
とを有し、該類似度計算部の結果によって入力音声を識
別する音声認識装置において、認識装置内のメモリーに
ロードされた辞書を記録媒体にセーブする際、辞書中の
単語の特徴とそれに対応する記録媒体中の特徴を比較し
た後に、該記録媒体中にセーブするようにしたことを特
徴とする音声認識装置。A voice that has a dictionary section that accommodates voice features, a feature extraction section that extracts the features of input voice, and a similarity calculation section that compares and matches them, and that identifies the input voice based on the result of the similarity calculation section. In the recognition device, when saving the dictionary loaded in the memory in the recognition device to a recording medium, the characteristics of the words in the dictionary and the corresponding characteristics in the recording medium are compared before saving in the recording medium. A voice recognition device characterized by:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP58021046A JPS59147400A (en) | 1983-02-10 | 1983-02-10 | Voice recognition equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP58021046A JPS59147400A (en) | 1983-02-10 | 1983-02-10 | Voice recognition equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| JPS59147400A true JPS59147400A (en) | 1984-08-23 |
Family
ID=12043980
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP58021046A Pending JPS59147400A (en) | 1983-02-10 | 1983-02-10 | Voice recognition equipment |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JPS59147400A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS6175395A (en) * | 1984-09-20 | 1986-04-17 | 三洋電機株式会社 | Reference pattern input system |
-
1983
- 1983-02-10 JP JP58021046A patent/JPS59147400A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS6175395A (en) * | 1984-09-20 | 1986-04-17 | 三洋電機株式会社 | Reference pattern input system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6394709B2 (en) | SPEAKER IDENTIFYING DEVICE AND FEATURE REGISTRATION METHOD FOR REGISTERED SPEECH | |
| US6535849B1 (en) | Method and system for generating semi-literal transcripts for speech recognition systems | |
| US8949127B2 (en) | Recognizing the numeric language in natural spoken dialogue | |
| WO2007055233A1 (en) | Speech-to-text system, speech-to-text method, and speech-to-text program | |
| JPH073640B2 (en) | Apparatus and method for generating a voice pattern plate | |
| US20210090563A1 (en) | Dialogue system, dialogue processing method and electronic apparatus | |
| JP2002132287A (en) | Speech recording method and speech recorder as well as memory medium | |
| CN109300468B (en) | Voice labeling method and device | |
| JP2003186494A (en) | Voice recognition device and method, recording medium and program | |
| JP3081108B2 (en) | Speaker classification processing apparatus and method | |
| Liu et al. | An ASR-free fluency scoring approach with self-supervised learning | |
| JPS59147400A (en) | Voice recognition equipment | |
| JP2005308950A (en) | Speech processors and speech processing system | |
| JPS6326699A (en) | Continuous word recognition recording | |
| WO2021082084A1 (en) | Audio signal processing method and device | |
| JP2000148187A (en) | Speaker recognizing method, device using the method and program recording medium therefor | |
| JPH05265482A (en) | Information processor | |
| JPS6011897A (en) | Voice recognition equipment | |
| JPH11184492A (en) | Speaker recognition device and its method | |
| Akintola et al. | Speech Processing Algorithm for Automatic Speaker Recognition-Based Telephone Voice Dialing in Yorùbá | |
| JPS608898A (en) | Voice recognition equipment | |
| JPS6127592A (en) | Voice pattern collation system | |
| JPS58224392A (en) | Voice recognition system | |
| JPS59121399A (en) | Word voice registration system | |
| JPS62121499A (en) | Voice recognition equipment |