JPH0344316B2 - - Google Patents
Info
- Publication number
- JPH0344316B2 JPH0344316B2 JP59024123A JP2412384A JPH0344316B2 JP H0344316 B2 JPH0344316 B2 JP H0344316B2 JP 59024123 A JP59024123 A JP 59024123A JP 2412384 A JP2412384 A JP 2412384A JP H0344316 B2 JPH0344316 B2 JP H0344316B2
- Authority
- JP
- Japan
- Prior art keywords
- candidate
- distance
- matching
- recognition
- pattern
- 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.)
- Expired - Lifetime
Links
- 238000010586 diagram Methods 0.000 description 3
- 238000000034 method Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 1
Description
【発明の詳細な説明】
〔発明の技術分野〕
本発明は音声認識装置に係り、特に複数の候補
を認識する音声認識装置においてその認識を更ら
に正確に行うようにしたものに関する。DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a speech recognition device, and particularly to a speech recognition device that recognizes a plurality of candidates and is adapted to perform the recognition more accurately.
音声認識装置では一般に入力音声パターンと辞
書部に格納された複数の標準音声パターンを比較
照合し、照合結果から最も似ている（距離が小さ
い）と判定したものを認識結果としている。
Generally, a speech recognition device compares and matches an input speech pattern with a plurality of standard speech patterns stored in a dictionary section, and uses the one determined to be the most similar (smallest distance) from the comparison results as the recognition result.
しかし、このような方法のみでは、例えば単語
の認識の場合におけるように、標準音声パターン
に登録されていないようなものに対する入力音声
が入力された場合や、発声不良により曖昧な入力
音声になつたものに対しても、何らかの認識結果
を出力しなければならないので、このようなとき
最も似ているという認識結果を出力してもこれは
間違つた結果を出力することになる。 However, this method alone cannot be used in cases where the input speech is not registered in the standard speech pattern, such as in the case of word recognition, or when the input speech is ambiguous due to poor pronunciation. Since it is necessary to output some kind of recognition result for objects, even if the recognition result of the most similarity is output in such a case, it will output the wrong result.
そのため、従来の音声認識装置では、照合結果
の値である照合距離にある判定距離を設定し、こ
の設定された判定距離よりも認識結果の照合距離
が大きいものは、辞書部に格納されている標準パ
ターンとは異なつた入力音声が入つたものとして
認識結果は正しくないと判定し、リジエクトを行
つている。また曖昧な発声により複数の標準パタ
ーンと似てしまい、正しく識別できないような入
力に対しては、最も照合距離の小さい第１候補
と、次に照合距離の小さい第２候補の照合距離の
差にある判定距離を設け、第２候補より第１候補
の距離差が判定距離よりも大きくなければ、入力
は曖昧な発声であつたとしてリジエクトさせる方
法がある。 Therefore, in conventional speech recognition devices, a judgment distance is set in the matching distance that is the value of the matching result, and if the matching distance of the recognition result is larger than this set judgment distance, it is stored in the dictionary section. The recognition result is determined to be incorrect because the input voice differs from the standard pattern and is rejected. In addition, for inputs that resemble multiple standard patterns due to ambiguous utterances and cannot be correctly identified, the difference in matching distance between the first candidate with the shortest matching distance and the second candidate with the next smallest matching distance will be used. There is a method in which a certain determination distance is set, and if the distance difference between the first candidate and the second candidate is not greater than the determination distance, the input is rejected as an ambiguous utterance.
しかしこの候補の距離差を比較する場合には、
判定要素は距離差のみであつて距離そのものが含
まれていないため、どの程度標準パターンと異つ
ているかということには無関係であり、また類似
語（例えば「カナガワ」と「カナザワ」）が存在
する場合にはこの機能は使用できない等の欠点が
あつた。 However, when comparing the distance differences between these candidates,
Since the determining factor is only the distance difference and does not include the distance itself, it is irrelevant to how much it differs from the standard pattern, and there are similar words (for example, "Kanagawa" and "Kanazawa"). In some cases, this function could not be used.
本発明の目的は、前記の如き欠点を改善するた
め、複数の候補が選択されたとき、第１候補と第
２候補の距離の比をチエツクして距離そのものの
フアクターを加えるとともに、その結果曖昧だと
思われる場合はこれらの候補が類似語か否かをチ
エツクし、類似語でなければ認識結果は曖昧であ
るとしてリジエクトさせ、類似語であれば第１候
補を認識結果とする（出力が１つの場合）ように
した音声認識装置を提供することである。
The purpose of the present invention is to improve the above-mentioned drawbacks by checking the ratio of the distance between the first candidate and the second candidate when a plurality of candidates are selected, adding the factor of the distance itself, and as a result, ambiguity. If it seems to be a similar word, check whether these candidates are similar words or not, and if they are not similar words, the recognition result is ambiguous and rejected. If it is a similar word, the first candidate is used as the recognition result (the output is In one case, it is an object of the present invention to provide a speech recognition device that does the following.
前記の如き目的を達成するために、本発明の音
声認識装置では、辞書部に格納された標準音声パ
ターンと入力音声パターンを比較照合して照合距
離の最も小さいものを第１候補とし次に小さいも
のを第２候補として認識する音声認識装置におい
て、前記第１候補と第２候補の各標準パターンの
照合距離を求める標準パターン照合手段と、第１
候補と第２候補間の照合距離が類似語判定距離か
否かを判定して両候補が類似語であると認識した
とき認識候補を有効として出力し、類似語でない
と認識したとき両候補の照合距離の比を求めこの
比と曖昧判定用の基準値を比較する曖昧判定手段
を設け、この曖昧判定手段により前記第１候補と
第２候補間照合距離を類似語判定距離と比較して
類似語であると判定したとき認識候補を有効とし
て出力し、逆に類似語でないと判定したとき前記
第１候補と第２候補との比を計算し、曖昧判定距
離比と比較して、比が大きければ曖昧でないと判
定して認識候補を有効として出力し、比が小さけ
れば曖昧であると判定して認識候補をリジエクト
するようにしたことを特徴とする。
In order to achieve the above object, the speech recognition device of the present invention compares and matches the standard speech pattern stored in the dictionary section and the input speech pattern, selects the one with the smallest matching distance as the first candidate, and selects the next smallest one as the first candidate. A speech recognition device that recognizes an object as a second candidate, comprising: standard pattern matching means for determining a matching distance between each standard pattern of the first candidate and the second candidate;
It is determined whether the matching distance between the candidate and the second candidate is the similar word determination distance, and when both candidates are recognized as similar words, the recognition candidate is output as valid, and when it is recognized that they are not similar words, the recognition candidate is output as valid. An ambiguity determination means is provided which calculates a ratio of matching distances and compares this ratio with a reference value for ambiguity determination, and this ambiguity determination means compares the matching distance between the first candidate and the second candidate with a similar word determination distance to determine similarity. When it is determined that it is a word, the recognition candidate is output as valid, and conversely, when it is determined that it is not a similar word, the ratio between the first candidate and the second candidate is calculated, and compared with the ambiguity judgment distance ratio, the recognition candidate is output. If the ratio is large, it is determined that the recognition candidate is not ambiguous and the recognition candidate is output as valid, and if the ratio is small, it is determined that the recognition candidate is ambiguous and the recognition candidate is rejected.
本発明を一実施例にもとづき詳述するに先立
ち、第１図および第２図により本発明の概略と、
従来の認識方式における欠点を説明する。
Before explaining the present invention in detail based on one embodiment, an outline of the present invention will be explained with reference to FIGS. 1 and 2, and
The drawbacks of conventional recognition methods will be explained.
第１図イにおいてP1，P2は標準パターンであ
り、このP1，P2の標準パターン間の照合距離を
dpとする。またia〜ieは入力パターンであり、第
１図イで示す各入力パターンと各標準パターン間
の距離は同ロの通りである。 In Figure 1 A, P 1 and P 2 are standard patterns, and the matching distance between the standard patterns of P 1 and P 2 is
Let it be dp. Further, ia to ie are input patterns, and the distances between each input pattern and each standard pattern shown in Fig. 1A are as shown in Fig. 1B.
ここでiaはP1，P2の直線に存在し、 da1＝１ da2＝２ の関係である。 Here, ia exists on the straight line between P 1 and P 2 , and the relationship is da 1 =1 da 2 =2.
このiaをP1であると判定できるとした場合、そ
の判定条件を距離の差で表すと
da2−da1≧１
また距離の比で表わすと
da2／da1≧２
となる。 If it is possible to determine that this ia is P 1 , then the determination condition is expressed as a distance difference: da 2 −da 1 ≧1, and expressed as a distance ratio: da 2 /da 1 ≧2.
すなわち、本発明では第１候補と第２候補の標
準パターンの照合距離が１より大であり、また第
１候補と第２候補との照合距離の比が２より大き
いときに判定可能であるとする。そして第１候補
と第２候補の標準パターンの照合距離が１未満で
あつてもこれらが類似語の場合には第１候補を判
定結果として出力する。 That is, in the present invention, determination is possible when the matching distance between the first candidate and the second candidate standard pattern is greater than 1, and the ratio of the matching distance between the first candidate and the second candidate is greater than 2. do. Even if the comparison distance between the first candidate and the second standard pattern is less than 1, if these are similar words, the first candidate is output as the determination result.
もし、従来のように、第１候補と第２候補との
距離の差だけで判定した場合には、ibやicの場合
にもP1が判定されることになるが、P1はibやicと
距離が相当離れておりP1と判定することはエラ
ーとなる。また距離の比だけで判定した場合に
は、idやieに示す如く、P1とP2からの距離に応じ
て、よりP2から離れた位置でないとP1であると
判定できないことが明らかである。（第１図にお
いてid，ieもP1と判定されるもの）これらの例か
ら明らかなように、照合距離の差のみで判定した
場合には、P1とPP2からの距離に関係なく判定基
準が一定であり、エラーが生ずる欠点があるが、
距離の比を判定する場合にはこのようなことはな
くなり、（勿論dd1，de1がある範囲内の距離にあ
る場合である。）ibやicは曖昧だとしてリジエク
トすることができる。 If, as in the past, the determination was based only on the difference in distance between the first and second candidates, P 1 would be determined even in the case of ib or ic, but P 1 would be determined based on the difference in distance between the first and second candidates. The distance from ic is quite far, and it would be an error to judge it as P 1 . Furthermore, when determining only based on the distance ratio, it is clear that it cannot be determined that it is P 1 unless it is located further away from P 2 , depending on the distance from P 1 and P 2 , as shown in id and ie. It is. (In Figure 1, id and ie are also judged as P 1 ) As is clear from these examples, when judgment is made only based on the difference in matching distance, the judgment is made regardless of the distance from P 1 and PP 2 . The disadvantage is that the standard is fixed and errors occur, but
This does not happen when determining the ratio of distances (of course, this is the case when dd 1 and de 1 are within a certain range of distances), and ib and ic can be rejected as ambiguous.
また、第１図のようにP1とP2が曖昧判定でき
る程離れていれば問題はないが、類似語の場合に
は、第２図に示す如く、標準パターンP3とP4が
接近してそれらの距離dpがdp＜１の場合がある。
この場合には、距離の差を１以上にとれるような
入力は不可能であり、また距離の比を２以上にと
るのも非常に狭い範囲に限られるため、リジエク
トが多くて使いにくいものとなる。いま第２図に
おいてdp＝0.5、df1＝0.5、df2＝0.75とすると、
距離の差も、距離の比も前記条件にあてはまらな
いため、ifがP3に非常に近い場合であつてもリジ
エクトされることになる。それ故、このような場
合に対処するため、あらかじめP3（第１候補）と
P4（第２候補）の間の距離を調べ、曖昧判定が可
能かどうかチエツクし、もし類似語であつて判定
不可能となつた場合にはリジエクトせずに第１候
補を有効とすることにより、類似語であつて曖昧
判定しにくいものに対してはそのまま第１候補を
認識結果として出力し、類似語でない場合だけ曖
昧判定を行うことになる。 Also, as shown in Figure 1, there is no problem if P 1 and P 2 are far enough apart to make an ambiguous judgment, but in the case of similar words, the standard patterns P 3 and P 4 are close together, as shown in Figure 2. In some cases, the distance dp between them is dp<1.
In this case, it is impossible to input a distance difference of 1 or more, and setting a distance ratio of 2 or more is limited to a very narrow range, so there are many rejects and it is difficult to use. Become. Now, in Figure 2, if dp = 0.5, df 1 = 0.5, df 2 = 0.75,
Since neither the difference in distance nor the ratio of distances meet the above conditions, it will be rejected even if if is very close to P3 . Therefore, in order to deal with such cases, P 3 (first candidate) and
Check the distance between P 4 (second candidate) and check whether an ambiguous judgment is possible. If it is a similar word and cannot be judged, do not reject and make the first candidate valid. Therefore, for words that are similar and difficult to make an ambiguous judgment, the first candidate is output as a recognition result, and ambiguity judgment is performed only when the words are not similar.
本発明の一実施例を第３図にもとづき説明す
る。 An embodiment of the present invention will be described based on FIG.
第３図において、１はマイクロフオン、２は音
声特徴抽出部であつて入力した音声信号よりその
特徴パラメータを抽出するもの、３は入力音声パ
ターン・バツフアであつて入力音声の特徴パラメ
ータを入力音声パターンとして一時保持するも
の、４は標準音声パターン辞書部であつて例えば
あらかじめ単語の標準音声パターンが格納されて
いるもの、５は第１パターン照合部であつて前記
入力音声パターン・バツフア３に保持された入力
音声の特徴パラメータを前記各標準音声パターン
辞書部４に格納された各標準音声特徴パラメータ
と照合してそれらの照合距離を算出するもの、６
は照合結果制御部であつて前記第１パターン照合
部５において入力音声と標準音声パターンとを順
次照合した結果照合距離の小さいものから複数個
認識候補として選出するもの、７は曖昧判定部で
あつて前記照合結果制御部６から伝達された第１
候補と第２候補情報を第２パターン照合部８に送
出したり、第２パターン照合部８から伝達された
第１候補と第２候補の標準パターンの照合距離が
類似語判定距離かどうかを判定して第１候補と第
２候補が類似語であると認識したとき認識候補が
有効であるとしてこれらを認識結果として出力
し、また類似語でないと認識したとき、第１候補
と第２候補の照合距離（入力音声に対するもの）
の比を計算し曖昧判定距離と比較して曖昧でない
と判定したとき認識候補を有効として認識結果を
出力し、曖昧であると判定した場合には認識候補
を無効としてリジエクトさせるもの、８は第２パ
ターン照合部であつて前記第１候補と第２候補と
の各標準パターンの照合距離を算出するものであ
る。 In FIG. 3, 1 is a microphone, 2 is a voice feature extractor which extracts the characteristic parameters from the input voice signal, and 3 is an input voice pattern buffer which extracts the characteristic parameters of the input voice from the input voice. 4 is a standard speech pattern dictionary section in which, for example, standard speech patterns of words are stored in advance; 5 is a first pattern matching section which is held in the input speech pattern buffer 3; 6, which compares the characteristic parameters of the input voice input with each standard voice characteristic parameter stored in each standard voice pattern dictionary section 4 and calculates a matching distance therebetween;
7 is a matching result control unit which selects a plurality of recognition candidates from the smallest matching distance as a result of sequentially matching input speech and standard speech patterns in the first pattern matching unit 5; 7 is an ambiguity determining unit; The first information transmitted from the verification result control unit 6
Sends the candidate and second candidate information to the second pattern matching unit 8, and determines whether the matching distance between the standard pattern of the first candidate and the second candidate transmitted from the second pattern matching unit 8 is a similar word determination distance. When it recognizes that the first and second candidates are similar words, the recognition candidates are considered valid and these are output as recognition results, and when it is recognized that they are not similar words, the first and second candidates are output as recognition results. Matching distance (relative to input audio)
The ratio is calculated and compared with the ambiguity determination distance, and when it is determined that it is not ambiguous, the recognition candidate is output as valid and the recognition result is output, and when it is determined that it is ambiguous, the recognition candidate is invalidated and rejected. This is a two-pattern matching unit that calculates the matching distance of each standard pattern between the first candidate and the second candidate.
本発明の動作について説明する。 The operation of the present invention will be explained.
(1) マイクロフオン１より入力された使用者の音
声（例えば単語）は、音声特徴抽出部２により
特徴パラメータが抽出されて照合用の入力音声
パターンに変換され、入力音声パターン・バツ
フア３に格納される。(1) The user's voice (for example, a word) inputted from the microphone 1 is extracted with characteristic parameters by the voice feature extraction unit 2, converted into an input voice pattern for verification, and stored in the input voice pattern buffer 3. be done.
(2) この入力音声パターンは、標準音声パターン
辞書部４に格納されている複数の標準音声パタ
ーンと順番に第１パターン照合部５と照合さ
れ、照合距離が算出される。(2) This input speech pattern is sequentially matched with a plurality of standard speech patterns stored in the standard speech pattern dictionary section 4 by the first pattern matching section 5, and a matching distance is calculated.
(3) この第１パターン照合部５から得られた照合
距離は順次照合結果制御部６に送出される。そ
して照合結果制御部６ではこの照合距離の小さ
いものから順に認識候補として選び出し、これ
を認識候補情報として曖昧判定部７に送出す
る。(3) The matching distances obtained from the first pattern matching section 5 are sequentially sent to the matching result control section 6. Then, the matching result control section 6 selects recognition candidates in descending order of matching distance, and sends them to the ambiguity determination section 7 as recognition candidate information.
(4) 曖昧判定部７ではこの認識候補情報により得
られた第１候補情報と第２候補情報を第２パタ
ーン照合部８に送出する。第２パターン照合部
８ではこれらにより標準音声パターン辞書部４
から第１候補と第２候補の各標準パターンを取
出してこれらの標準パターン間の照合距離を求
め、候補間照合距離として曖昧判定部７に送出
される。(4) The ambiguity determination unit 7 sends the first candidate information and second candidate information obtained from this recognition candidate information to the second pattern matching unit 8. In the second pattern matching unit 8, the standard speech pattern dictionary unit 4
The standard patterns of the first candidate and the second candidate are extracted from the standard patterns, the matching distance between these standard patterns is determined, and the result is sent to the ambiguity determination unit 7 as the matching distance between the candidates.
(5) これにもとづき曖昧判定部７では候補間照合
距離を、別途定められている類似語であるかど
うかを判定する類似語判定距離と比較する。も
し類似語判定距離より小さく類似語であると判
定したとき、認識候補が有効であるとして、こ
れら（場合によつては第１候補）を認識結果と
して出力する。逆に類似語でないと判断したと
き、第１候補と第２候補との比を計算し、これ
また別途定められている曖昧判定距離比と比較
して、比が大きく曖昧でないと判定できれば、
認識候補を有効として認識結果を送出する。し
かしこの比が小さくて曖昧であると判定した場
合にはこの認識候補は無効としてリジエクトさ
せることになる。(5) Based on this, the ambiguity determination unit 7 compares the inter-candidate matching distance with a separately determined similar word determination distance for determining whether the words are similar words. If it is determined that the distance is smaller than the similar word determination distance, it is determined that the recognition candidates are valid, and these (in some cases, the first candidate) are output as recognition results. On the other hand, when it is determined that the words are not similar, the ratio between the first candidate and the second candidate is calculated and compared with the ambiguity determination distance ratio, which is also separately determined.If the ratio is large and it is determined that the words are not ambiguous,
The recognition candidate is validated and the recognition result is sent. However, if this ratio is determined to be small and ambiguous, this recognition candidate is rejected as invalid.
なお前記数値はあくまでも１例にすぎず距離の
差や比等もふくめ、本発明がこれらに限定される
ものではない。 Note that the above numerical values are merely examples, and include distance differences, ratios, etc., and the present invention is not limited thereto.
本発明により、従来における第１候補と第２候
補の距離差で曖昧判定を行うものに比較して、第
１候補および第２候補の両者から大きく離れた場
合に曖昧と判定することができ、標準音声パター
ンに類似語が含まれている場合でも認識すること
が可能となる。そのため従来のように標準音声パ
ターンに類似語が含まれるとき曖昧判定の機能を
使用できない場合があつた欠点が、本発明では類
似語判定を行つて曖昧判定を制御できるので認識
効果を向上することができる。
According to the present invention, compared to conventional methods that perform ambiguity determination based on the distance difference between the first candidate and the second candidate, it is possible to determine ambiguity when the distance is far from both the first candidate and the second candidate, Even if the standard speech pattern contains similar words, it can be recognized. Therefore, in the past, the ambiguity judgment function could not be used when a similar word was included in the standard speech pattern, which was a drawback, but in the present invention, the ambiguity judgment can be controlled by performing similar word judgment, thereby improving the recognition effect. I can do it.
第１図は従来の問題点および本発明の概略説明
図、第２図は類似語における問題点説明図、第３
図は本発明の一実施例構成図である。
図中、１はマイクロフオン、２は音声特徴抽出
部、３は入力音声パターン・バツフア、４は標準
音声パターン辞書部、５は第１パターン照合部、
６は照合結果制御部、７は曖昧判定部、８は第２
パターン照合部を示す。
Fig. 1 is a schematic explanatory diagram of conventional problems and the present invention, Fig. 2 is an explanatory diagram of problems with similar words, and Fig. 3
The figure is a configuration diagram of an embodiment of the present invention. In the figure, 1 is a microphone, 2 is a voice feature extraction section, 3 is an input voice pattern buffer, 4 is a standard voice pattern dictionary section, 5 is a first pattern matching section,
6 is a matching result control section, 7 is an ambiguous judgment section, and 8 is a second
The pattern matching section is shown.
Claims (1)
音声パターンを比較照合して照合距離の最も小さ
いものを第１候補とし次に小さいものを第２候補
として認識する音声認識装置において、前記第１
候補と第２候補の各標準パターンの照合距離を求
める標準パターン照合手段と、第１候補と第２候
補間の照合距離が類似語判定距離か否かを判定し
て両候補が類似語であると認識したとき認識候補
を有効として出力し、類似語でないと認識したと
き両候補の照合距離の比を求めこの比と曖昧判定
用の基準値を比較する曖昧判定手段を設け、この
曖昧判定手段により前記第１候補と第２候補間照
合距離を類似語判定距離と比較して類似語である
と判定したとき認識候補を有効として出力し、逆
に類似語でないと判定したとき前記第１候補と第
２候補との比を計算し、曖昧判定距離比と比較し
て、比が大きければ曖昧でないと判定して認識候
補を有効として出力し、比が小さければ曖昧であ
ると判定して認識候補をリジエクトするようにし
たことを特徴とする音声認識装置。1. In a speech recognition device that compares and matches a standard speech pattern stored in a dictionary section and an input speech pattern, and recognizes the one with the smallest matching distance as a first candidate and the second smallest one as a second candidate, the first
A standard pattern matching means for calculating the matching distance between each standard pattern of the candidate and the second candidate, and determining whether the matching distance between the first candidate and the second candidate is a similar word determination distance to determine whether both candidates are similar words. ambiguity determination means is provided, which outputs the recognition candidate as valid when it is recognized as a valid word, and calculates the ratio of the matching distances of both candidates when it is recognized that they are not similar words, and compares this ratio with a reference value for ambiguity determination; The matching distance between the first candidate and the second candidate is compared with the similar word determination distance, and when it is determined that they are similar words, the recognition candidate is output as valid, and conversely, when it is determined that they are not similar words, the recognition candidate is outputted as valid. and the second candidate, and compare it with the ambiguity determination distance ratio. If the ratio is large, it is determined that it is not ambiguous and the recognition candidate is output as valid, and if the ratio is small, it is determined that it is ambiguous and it is recognized. A speech recognition device characterized by rejecting candidates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP59024123A JPS60168197A (en) | 1984-02-10 | 1984-02-10 | Voice recognition/decision system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP59024123A JPS60168197A (en) | 1984-02-10 | 1984-02-10 | Voice recognition/decision system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| JPS60168197A JPS60168197A (en) | 1985-08-31 |
| JPH0344316B2 true JPH0344316B2 (en) | 1991-07-05 |
Family
ID=12129528
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP59024123A Granted JPS60168197A (en) | 1984-02-10 | 1984-02-10 | Voice recognition/decision system |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JPS60168197A (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS63254498A (en) * | 1987-04-13 | 1988-10-21 | 日本電気株式会社 | Voice recognition responder |
| JP2003029783A (en) * | 2001-07-17 | 2003-01-31 | Oki Electric Ind Co Ltd | Voice recognition control system |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS59212900A (en) * | 1983-05-19 | 1984-12-01 | 株式会社日立製作所 | Voice recognition equipment |
-
1984
- 1984-02-10 JP JP59024123A patent/JPS60168197A/en active Granted
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS59212900A (en) * | 1983-05-19 | 1984-12-01 | 株式会社日立製作所 | Voice recognition equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| JPS60168197A (en) | 1985-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR100655491B1 (en) | Two stage utterance verification method and device of speech recognition system | |
| JP3284832B2 (en) | Speech recognition dialogue processing method and speech recognition dialogue device | |
| US6618702B1 (en) | Method of and device for phone-based speaker recognition | |
| JPS62217295A (en) | Voice recognition system | |
| EP0121248A1 (en) | Speaker verification system and process | |
| KR102655791B1 (en) | Speaker authentication method, learning method for speaker authentication and devices thereof | |
| KR20110070688A (en) | Apparatus and method using two phase utterance verification architecture for computation speed improvement of n-best recognition word | |
| US7181396B2 (en) | System and method for speech recognition utilizing a merged dictionary | |
| JP2996019B2 (en) | Voice recognition device | |
| JPH0344316B2 (en) | ||
| KR101229108B1 (en) | Apparatus for utterance verification based on word specific confidence threshold | |
| JPH1083195A (en) | Input language recognition device and input language recognizing method | |
| KR20020045960A (en) | Method for performance improvement of keyword detection in speech recognition | |
| JP3112037B2 (en) | Voice recognition device | |
| JPH11249688A (en) | Device and method for recognizing voice | |
| KR100952974B1 (en) | System and method for recognizing voice dealing with out-of-vocabulary words, and computer readable medium storing thereof | |
| JP2000214879A (en) | Adaptation method for voice recognition device | |
| Niimi et al. | Analysis of interactive strategy to recover from misrecognition of utterances including multiple information items | |
| KR100449912B1 (en) | Apparatus and method for detecting topic in speech recognition system | |
| KR100677224B1 (en) | Speech recognition method using anti-word model | |
| KR100281581B1 (en) | Korean Continuous Number Speech Recognition Using Simultaneous Articulation Model | |
| JP3100208B2 (en) | Voice recognition device | |
| KR20170103202A (en) | Word boundary cordinator apparatus for natural language talk voice recognition | |
| KR100382473B1 (en) | Speech recognition method | |
| KR20030001667A (en) | voice recognition method for joing the speaker adaptation and the rejection of error input |