JPH0375799A - Voice recognizing system - Google Patents

Abstract

PURPOSE:To improve a voice recognition rate by providing a standard pattern memory means, voice analyzing means, standard pattern selecting means, means for sequentially selecting the standard pattern selecting means, comparing means, and utterance inducing means. CONSTITUTION:The utterance inducing means 5 induces utterance with 'Which do you select? and waits for the utterance for the flavor name from a customer when coins are charged into a coil receiving means 7. The standard pattern nearest the inputted voice patterns is selected from the standard patterns stored in the standard pattern memory means 3 by the standard pattern selecting means A 9 or standard pattern selecting means B 10, by which the flavor name is recognized. The flavor name is recognized by the means 5 and an answer is confirmed. A control means 6 pours drinks into a cup and ejects the cup by using a drink ejecting means 8. The standard pattern nearest the voice pattern extracted by the voice analyzing means is selected by the comparing means in such a manner and, therefore, the voice recognition system having the high recognition rate is obtd.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a speech recognition system for recognizing word speech input by specific speakers and unspecified speakers and performing various processes using the speech. , for unspecified speakers.

2. Description of the Related Art Conventionally, a voice recognition system for a vending machine such as a vending machine for cup beverages (hereinafter simply referred to as a cup vending machine) first inputs information from a user through a microphone 1, as shown in FIG. The voice analyzed by the voice analysis means 2 is analyzed to extract a voice pattern using a BPF (Band Path Full) using a group of band pass filters.
ter) A method is used to sample the analysis results on the time and frequency axes and digitally process the intensity. The standard pattern storage means 8 stores, as standard patterns, audio patterns of a plurality of discrete words uttered by a large number of unspecified speakers, which are extracted by a similar method. However, the words stored as standard patterns are the name of the flavor sold at the cup vending machine (product name of coffee juice, etc.) and some response words (yes, no, hot, ice cream, etc.)
It is.

Then, the standard pattern selection means 4 selects the standard pattern closest to the input cover turn from among the standard patterns as DP (Dy
The DP matching method, which is translated as dynamic programming, is a method of mathematical programming proposed by Bellman in the United States in 1957, and is applied to the optimization of multi-stage decision processes. The method involves making certain decisions (controls) at each stage to transform the state, and evaluating the quality and quality of the process until reaching the goal.
The idea is to maximize/minimize the number of times for evaluating badness. Also, if the recognition system corresponds to a specific speaker,
The voice pattern of the recognized word uttered by a specific speaker is registered in the standard pattern storage means 3, and when dealing with unspecified speakers, a representative voice pattern of the recognized word uttered by an unspecified number of speakers is registered. The speech guiding means 5, which registers some of the patterns, is constituted by a speech synthesizing means, and prompts the user to make vocalizations in accordance with the control means 6, which will be described later. However, the flavor salts are clearly marked on the panel board on the front of the cup vending machine, and the user selects one flavor salt of their choice and speaks it out loud.

The control means 6 instructs the voice guidance means 5 to utter the guidance voice according to the process, and also determines whether the distance between the standard pattern selected by the standard pattern selection means 4 using the DP matching method and the input cover turn is rejected. If it is smaller than the threshold, it is determined that the input pattern has been recognized, and the subsequent operation of the cup vending machine is controlled based on the recognition result, and if it is larger than the reject threshold, it is determined to be rejected, and the pattern is selected again by the standard pattern selection means 4. The smaller the rejection threshold is, the higher the recognition rate will be, but the number of rejections will also increase, so it must be set to a reasonably high value. Coin receiving means for refund, 8
is a beverage dispensing means for pouring the selected flavor into a cup and discharging it. FIG. 5 shows the voice recognition possible period of the conventional voice recognition system for vending machines.

tl is a voice end confirmation period, and t2 is a pattern selection processing period. From FIG. 5, it can be seen that speech recognition is impossible during periods tl and t2.

Problems to be Solved by the Invention However, in the above method, the voice of the speaker cannot be recognized during the periods tl and t2. Since the standard pattern that satisfies the above criteria is recognized as the input pattern, it has the disadvantage that the recognition rate decreases.

The present invention solves the above-mentioned conventional problem H, eliminates the period t2 during which voice recognition is not possible, and selects the standard pattern that is closest to the input cover pattern among the standard patterns selected by the standard pattern selection means. The purpose is to provide a speech recognition system with a high recognition rate by selecting the following.

Means for Solving the Problems In order to achieve this object, the speech recognition system of the present invention includes a standard pattern storage means that stores a group of standard patterns of a plurality of discrete word sounds, and a standard pattern storage means that analyzes the speech of a speaker and generates speech patterns. a speech analysis means for extracting, a plurality of standard pattern selection means for selecting from the standard pattern group a standard pattern closest to the speech pattern extracted by the speech analysis means, and any standard among the plurality of standard pattern selection means. a selection means for sequentially selecting pattern selection means; a comparison means for selecting a standard pattern closest to the speech pattern extracted by the speech analysis means from among the standard patterns selected by the selected standard pattern selection means; and utterance guidance means for guiding the person to utter the words.

Effect With this configuration, by having a plurality of standard pattern selection means, when one standard pattern selection means is in the pattern selection processing period, the other standard pattern selection means performs pattern selection processing, thereby preventing voice recognition failure. A speech recognition system with a high recognition rate is achieved by eliminating the possible period t2 and by selecting the standard pattern closest to the input cover pattern among the standard patterns selected by a plurality of standard pattern selection means. This means that it can be realized.

EXAMPLE An example of the present invention will be described below with reference to the drawings.

In this embodiment, a voice recognition system for unspecified speakers is applied to a cup vending machine. However, among the structural requirements, those having the same configuration as the conventional example are given the same numbers and explanations are omitted. FIG. 1 shows a functional block diagram of a speech recognition system in an embodiment of the present invention. 9.10 are standard pattern selection means A and standard pattern selection means B, respectively, which select the standard pattern closest to the input cover turn from among the standard patterns as DP (Dynamic Pro).
This method recognizes selected speech using a matching method (gramlng). Reference numeral 11 denotes a selection means, which selects one of the plurality of standard pattern selection means 9 and 10.

Reference numeral 12 denotes a comparison means, which selects the standard pattern that is closest in distance to the input cover turn from among the standard patterns selected by the plurality of standard pattern selection means 9 and 10. FIG. 2 shows the voice recognition possible period of the voice recognition system for vending machines according to the embodiment of the present invention. m jl is a voice end confirmation period, and t2 and t2° are pattern selection processing periods. Second
As shown in the figure, the selection means 11 selects the standard pattern selection means 11 until the standard pattern selection means A9 enters the pattern selection processing period, and thereafter selects the standard pattern selection means BIO. Then, when the standard pattern selection means BIO enters the pattern selection processing period, it selects the standard pattern selection means A9 again, and thereafter repeats this series of operations.

Therefore, for example, if the speaker says, ``Um (voice input period)''
, coffee (voice input period)'', in the conventional voice recognition system for vending machines, the voice ``coffee'' is the pattern selection processing period of the standard pattern selection means A9 (the processing period for the voice ``um''). However, in the embodiment of the present invention, the m
It will be recognized by the quasi-pattern selection means BIO.

Next, the operation of the comparison means 12 will be explained using Tables 1 and 2.

Table 1 Table 2 Table 1 shows the standard patterns selected by the standard pattern selection means and the distances from the input cover turns.

From Table 1, it can be seen that the standard patterns selected by standard pattern selection means A and B are a and b, respectively, and the distances between the selected standard patterns and the incoming cover turns are L1eL2, respectively.Table 2 shows the magnitude of L1tL2. The relationship and the standard pattern selected by the comparing means 12 are shown. From Table 2, it can be seen that the comparison means 12 selects the standard pattern a when Ll<L2, L,=L2, and selects the standard pattern b when Ll<L2. therefore,
Among the standard patterns selected by the plurality of standard pattern selection means 9.10, the standard pattern closest to the voice pattern extracted by the voice analysis means 2 is selected. In this embodiment, the standard pattern selection means 9,
Although the case where there are two standard patterns selected in 10 has been described, it goes without saying that even if there are more than two standard patterns selected in 9, the standard pattern closest to the input audio pattern is selected in the same way.

The vending operation of the voice recognition system for a cup vending machine configured as described above will be explained using the flowchart shown in FIG. In FIG. 3, first, step 20
1, it is determined whether a coin has been inserted into the coin receiving means 7, and if a coin has been inserted, the process proceeds to step 202.In step 202, the voice guidance means 5 says "
"Welcome, what would you like?" and waits for the customer to say the flavor name. Step 2
In step 08, the standard pattern selection means A9 or the standard pattern selection means BIO selects the standard pattern closest to the input voice pattern from among the standard patterns stored in the standard pattern storage means 3 and creates a flavor name. recognize. In step 204, step 2
It is determined whether the recognition result in step 03 is appropriate or not, and if it is rejected, the process proceeds to step 205, where the voice guidance means 5 says "
Please answer again" and return to step 203. On the other hand, if the answer is not rejected, proceed to step 206.

In step 206, the subsequent operation is branched depending on the flavor recognized in step 203, but in this embodiment, it is assumed that coffee has been recognized, and the operation when another flavor name is recognized is the same as in the case of coffee. Since it is similar, the explanation will be omitted. 0 Next step 20?
Then, the voice guidance means 5 confirms that "it's coffee," and waits for the customer's response.Then, in step 208, a yes or no response is recognized in the same manner as for the flavor name.

In step 209, it is determined whether the recognition result in step 208 is appropriate or not, and in the case of rejection, step 207
Otherwise, the process proceeds to step 210. In step 210, if the answer recognized in step 208 is yes, the process proceeds to step 211, and if no, the process returns to step 205.

In step 211, the control means 6 pours and transports coffee into a cup using the beverage transport means 8. and,
In step 212, if there is change, the change is refunded by the coin receiving means 7, and finally, in step 213
Then, the voice guidance means 5 utters "Thank you very much" and the series of operations ends.

As described above, according to this embodiment, by having a plurality of standard pattern selection means, when one standard pattern selection means enters the pattern selection processing period, the other standard pattern selection means enters the speech analysis result acquisition period. Therefore, voice recognition is now possible even during periods when voice recognition was previously impossible. For example, when a speaker is unsure about choosing a flavor, it is possible to recognize the voice by saying, ``Um, (flavor name).'' The probability increases. Also, among the standard patterns selected by the plurality of standard pattern selection means, the standard pattern closest to the speech pattern extracted by the speech analysis means is selected. Therefore, the recognition rate of the speech recognition system improves. At the same time, the number of rejections is reduced, and the speaker can have a smooth conversation, which is a great effect.

Effects of the Invention As described above, the speech recognition system of the present invention includes: a standard pattern storage means that stores a group of standard patterns of a plurality of discrete word sounds; a speech analysis means that analyzes a speaker's speech and extracts a speech pattern; a plurality of standard pattern selection means for selecting a standard pattern closest to the voice pattern extracted by the voice analysis means from the standard pattern group; and one of the standard pattern selection means among the plurality of standard pattern selection means is selected in order. a selection means for selecting a standard pattern that is closest to the speech pattern extracted by the speech analysis means from among the standard patterns selected by the selected standard pattern selection means; By providing a voice guidance means for guiding, when one standard pattern selection means is in the pattern selection processing period, the other standard pattern selection means enters the speech analysis result acquisition period, which conventionally made speech recognition impossible. Speech recognition is now possible even during a certain period of time, and since the comparison means selects the standard pattern closest to the speech pattern extracted by the speech analysis means from among the standard patterns selected by the plurality of standard pattern selection means, the recognition rate increases. This makes it possible to realize a speech recognition system with high performance.

[Brief explanation of drawings]

FIG. 1 is a functional block diagram of a speech recognition system according to an embodiment of the present invention, FIG. 2 is an explanatory diagram of the speech recognition possible period of the speech recognition system according to an embodiment of the present invention, and FIG. 3 is an embodiment of the present invention. FIG. 4 is a functional block diagram of the conventional speech recognition system, and FIG. 5 is an explanatory diagram of the speech recognition possible period of the conventional speech recognition system. 2. Voice analysis means, 8. Standard pattern storage means, 5. Vocal guidance means, 9. Standard pattern selection means A, 10. Standard pattern selection means B111.
Selection means, 12. - Comparison means.

Claims (1)

[Claims] a standard pattern storage means that stores a group of standard patterns of a plurality of discrete word sounds, a voice analysis means that analyzes the voice of a speaker and extracts a voice pattern, and a standard pattern that is closest to the voice pattern extracted by the voice analysis means. a plurality of standard pattern selection means for selecting from the standard pattern group; a selection means for sequentially selecting one of the standard pattern selection means from the plurality of standard pattern selection means; A speech recognition system comprising a comparing means for selecting, from standard patterns, a standard pattern closest to the speech pattern extracted by the speech analyzing means, and a speech guiding means for guiding a speaker to pronounce a word.