WO2020136872A1

WO2020136872A1 - Atmosphere purifying device, atmosphere purifying method, and elevator

Info

Publication number: WO2020136872A1
Application number: PCT/JP2018/048478
Authority: WO
Inventors: 昌鷹馬場; 啓吾川島; 信秋田中
Original assignee: 三菱電機株式会社
Priority date: 2018-12-28
Filing date: 2018-12-28
Publication date: 2020-07-02
Also published as: JPWO2020136872A1

Abstract

An atmosphere purifying device (100) has an input unit (1), a feeling discrimination unit, an atmosphere estimation unit (8), and a stimulus output unit (11). The input unit (1) acquires a sound. The feeling discrimination unit discriminates the feeling of a person on the basis of a feature quantity derived from the feeling of the person included in a sound and boundary information for discriminating the feeling of the person from the feature quantity. The atmosphere estimation unit (8) estimates the degree of atmosphere in the periphery of a sound generation source on the basis of the result of discrimination of the person's feeling. The stimulus output unit (11) outputs a stimulus for purifying the atmosphere in accordance with the degree of atmosphere to the periphery of the sound generation source.

Description

Atmosphere cleaning device, atmosphere cleaning method and elevator

The present invention relates to an atmosphere cleaning device, an atmosphere cleaning method, and an elevator that detect the harshness of the atmosphere of a space where people are present and clean the harsh atmosphere.

There are differences depending on the degree of "bad atmosphere" or "unpleasant atmosphere" that appears in various scenes of human life and social activities, but in many cases, people live unconsciously or consciously There is. Repeating such a situation may cause stress to the parties, impair physical and mental health, and, in extreme cases, lead to some crime. Moreover, even if a direct crime does not occur, the social loss caused by the onset of psychosomatic disorder, which may be an extension of the "bad atmosphere", is large, and it can be said that this is a social problem to be dealt with. When the "bad atmosphere" becomes serious, it becomes a "bad atmosphere". The "harmful atmosphere" is, for example, a situation in which one of the parties is harmful to the other, for example, a situation in which an act of extortion, an injury, or a molester is performed, or a notice of harm is given as a preceding step. , The situation where some interaction occurs.

There is a crisis monitoring device that detects the critical state from such a "bad atmosphere" to the crime itself, and makes an emergency call to the guards. For example, when the voice signal inside the monitoring target is acquired and the emotion feature amount is extracted and compared with the emotion feature amount of the utterance voice prepared in advance in a critical situation, when the determined emotion is in a critical state, Emergency processing such as notifying a security guard according to a predetermined response method is performed (for example, see Patent Document 1).

JP 2005-346254 A

In the crisis monitoring device of Patent Document 1, emotions are estimated by analyzing the uttered voice of a person inside the monitoring target. In the case of a critical condition, emergency processing such as reporting to a guard outside the monitoring target is carried out, but there is no function to mitigate the critical condition on the spot, and from the reporting to the rush of the guard. There was a problem that the crisis situation continued for a long time or the crisis situation worsened.

The present invention has been made to solve the above-mentioned problems, and is to obtain an atmosphere cleaning device capable of cleaning a severe atmosphere.

An atmosphere cleaning device according to an aspect of the present invention is provided. The atmosphere cleaning device identifies a person's emotion based on an input unit for acquiring a sound, a feature amount derived from a person's emotion contained in the sound, and boundary information for identifying the person's emotion from the feature amount. The emotion identification unit, an atmosphere estimation unit that estimates the degree of the atmosphere around the sound source based on the identification result of the person's emotion identified by the emotion identification unit, and a stimulus that cleans the atmosphere according to the degree of the atmosphere And a stimulus output unit that outputs the sound around the sound source.

According to the present invention, the stimulus for cleaning the atmosphere is output according to the degree of the atmosphere, so that a bad atmosphere can be cleaned.

It is a block diagram which shows the structure of the atmosphere cleaning apparatus which concerns on Embodiment 1 of this invention. It is a figure which shows a "ring of emotions of Prutic". It is a figure for demonstrating the utterance voice (action sound) analysis processing which concerns on Embodiment 1 of this invention. It is a figure explaining the identification boundary information recorded on the 1st recording part for atmosphere detection which concerns on Embodiment 1 of this invention. FIGS. 5A and 5B are diagrams for explaining the identification boundary information for identifying an emotion from the feature amount vector for each emotion according to the first embodiment of the present invention. It is a figure explaining the identification boundary information recorded on the 2nd recording part for atmosphere detection which concerns on Embodiment 1 of this invention. The flow of the speech analysis processing according to the first embodiment of the present invention is applied to the block configuration of the atmosphere cleaning device shown in FIG. It is a figure which shows the example of emotion estimation by the 1st identification part which concerns on Embodiment 1 of this invention. The flow of action sound analysis processing according to Embodiment 1 of the present invention is applied to the block configuration of the atmosphere cleaning device in FIG. 1. FIGS. 10(a) and 10(b) are diagrams for explaining a method of separating an utterance voice and an action sound from an input voice according to the first embodiment of the present invention. It is a figure which shows the operation|movement which estimates an emotion for every predetermined estimation period in the atmosphere estimation part which concerns on Embodiment 1 of this invention. 12(a) to 12(c) are diagrams showing an operation for obtaining a transition of the atmosphere level from the appearance frequency of the estimated emotion in the atmosphere estimating unit according to the first embodiment of the present invention. It is a figure which shows an example of the specific structure of the stimulation control part, the stimulation production|generation part, and the stimulation output part which concern on Embodiment 1 of this invention. It is a figure which shows an example of the structure for the output processing of the auditory stimulation which concerns on Embodiment 1 of this invention. It is a figure which shows an example of a structure for the output processing of the olfactory stimulus which concerns on Embodiment 1 of this invention. It is a figure which shows an example of the structure for the output processing of the visual stimulation which concerns on Embodiment 1 of this invention. It is a figure which shows an example of the structure for output processing of the electromagnetic wave stimulation which concerns on Embodiment 1 of this invention. It is a figure which shows an example of the structure for output processing of the charged particle stimulation which concerns on Embodiment 1 of this invention. It is a matrix diagram which shows an example of the stimulus control which concerns on Embodiment 1 of this invention. It is a figure which shows an example of the analysis method of the action sound which concerns on Embodiment 2 of this invention. It is a figure which shows an example of the analysis result of the action sound which concerns on Embodiment 2 of this invention. It is a functional block diagram at the time of applying the atmosphere cleaning device concerning Embodiment 3 of this invention to an elevator. It is a figure which shows an example of the serious situation when the atmosphere cleaning apparatus which concerns on Embodiment 3 of this invention is applied to an elevator. It is a functional block diagram at the time of applying the atmosphere cleaning apparatus concerning Embodiment 3 of this invention to an elevator, using a malodorous component as an olfactory stimulus, and performing a ventilation process. It is a figure which shows the structure of the hardware which an atmosphere cleaning has.

Embodiments will be described below with reference to the drawings. The following embodiments are merely examples, and various modifications can be made within the scope of the present invention.

Embodiment 1.
1 is a block diagram showing a configuration of an atmosphere cleaning device according to a first embodiment for carrying out the present invention. The atmosphere cleaning device 100 is a device that executes an atmosphere cleaning method. The atmosphere cleaning device 100 can acquire sound. For example, the sound is a person's uttered voice (hereinafter, referred to as “uttered voice”), a person's action sound (hereinafter, referred to as “action sound”), or a synthesized sound of the uttered voice and the action sound. The atmosphere cleaning device 100 detects a harsh atmosphere based on the sound and outputs a stimulus for cleaning the harsh atmosphere to the vicinity of the sound source.

In the atmosphere cleaning device 100 according to the present embodiment, the sound is input by the sound input unit 1 that is an input unit that acquires the utterance voice and the action sound of the target people, the utterance voice analysis unit 2, and the action sound analysis unit 3. The input side is configured. Further, in the atmosphere cleaning device 100, the first recording unit 4 in which the identification boundary information regarding the emotion feature amount of the emotional speech to be detected is recorded in advance, and the identification boundary information regarding the emotion feature amount of the emotional action sound is also recorded in advance. The second recording unit 5 is configured. The first recording unit 4 and the second recording unit 5 constitute a boundary information reference side for identification. In the atmosphere cleaning device 100, the first identification unit 6 that identifies the emotion of the uttered voice based on the identification boundary information that is previously recorded in the first recording unit 4, and the identification that is previously recorded in the second recording unit 5. A second identification unit 7 that identifies the emotion of the action sound based on the boundary information and an atmosphere estimation unit 8 are configured. The first identification section 6, the second identification section 7, and the atmosphere estimation section 8 constitute an atmosphere detection side. Based on the detection result of the atmosphere, the stimulus control unit 9 and the stimulus generation unit 10 generate a stimulus for cleaning the atmosphere, and the stimulus output unit 11 outputs the stimulus to people.

The sound input unit 1 is an input unit that acquires a sound. The sound acquired by the sound input unit 1 is at least one of an utterance voice and an action sound. The action sound is a sound that accompanies a person's action. For example, the action sound refers to a sound generated as a result of handling the object without paying attention to the object itself or others around the person when handling the object, and also includes a sound generated when a person taps a desk, a wall, or a floor. ..

The emotion identifying unit is based on the feature amount derived from the human emotion (hereinafter referred to as “emotion”) included in the sound acquired by the sound input unit 1 and the boundary information for identifying the emotion from the feature amount. , Estimate emotions. The emotion identifying unit also generates information indicating the estimated emotion type (for example, "anger" or "irritation"). The processing executed by the identification unit can be realized by the utterance voice analysis unit 2, the action sound analysis unit 3, the first identification unit 6, and the second identification unit 7.

Next, we will explain the system operation by setting the specific situation. For example, it is assumed that the people in the conference room to be monitored have, for some reason, been in an "awful atmosphere". More specifically, the parties X, Y, and Z make up the people, and the party X is agitated and makes an inappropriate act of denying the character with a very intimidating attitude toward the party Y. Suppose Needless to say, the situation is extremely unbearable for the party Y, but also for the party Z who is present at that moment, the party Y feels a “terrible atmosphere” and becomes extremely uncomfortable.

Fig. 2 is a diagram showing the "Pultic's emotional circle". Under such a situation, the utterance voice and the action sound uttered by the party X are, for example, “anger”, “anger”, “frustration”, “hate”, “disgust” when viewed in “Pulch's emotional ring”. It is considered that the emotional characteristics such as “” are included. On the other hand, from a different point of view, it is said that “anger is the inside out of fear”, and the reason why the party X has become agitated is “fear”, “worry”, “anxiety” with respect to the party Y or a third party. It is also possible that the feeling of is a distant cause. For example, the party X feels that his or her thoughts and values have been denied, or feels that things do not go according to his or her own wishes, and fears that his or her social evaluation will decline. It may be. Therefore, the utterance voice and the action sound uttered by the party X may include emotional features such as “fear”, “worry”, and “anxiety”.

In such a situation, if, for example, the party X himself has self-control and can suppress the excitement himself, the "dangerous atmosphere" will not reach a critical situation in the first place. Further, if the party Z can restrain the party X to calm down and lead to a calm situation, similarly, it is possible to avoid reaching a critical situation. However, it is unavoidable to avoid the involvement of the party X, often concerned with the excitement of the party X directed to Z himself.

Therefore, the atmosphere cleaning device 100 according to the present invention analyzes the emotional characteristic amount of the utterance voice and the action sound acquired from the people including the agitated party X, and determines “anger”, “anger”, “irritability”, and “hate”. ”, “Aversion”, “Horror”, “Worry”, “Anxiety”, etc. Then, the atmosphere cleaning device 100 outputs to the people a stimulus for cleaning the atmosphere based on the detection result, so that the "dangerous atmosphere" does not reach a critical state. It works to cleanse (relax) the emotions you have made. However, the atmosphere cleaning device 100 may be operated so as not to reach from a slight "unpleasant atmosphere" to a serious "severe atmosphere" as a preceding step.

Also, explain another specific situation setting. For example, it is assumed that the target of monitoring is an elevator car, and there are parties X, Y, and Z in the elevator car. Described below is a case where the party X, of the parties X, Y, and Z, has an offense for some reason. For example, the party X expects the party Y to apologize for the reasons such as "the shoulder and the baggage of the party Y hit the party X" and "the party Y gazed at the party X". Instead, they assume that they were "ignored," "laughed," "ridiculed," etc., and they are notified of some harm and end up in a critical state. Notifications of harm include, for example, "shibaku," "tsukuri," "kill," "buy in the soil," "sink in the sea," and the like, as well as actions that are enough to scare the general public. It can be said that the act of ringing falls under that level if the ordinary person is afraid enough. It is considered that the utterance voice and the action sound in such a notification of malicious intent include emotional features such as "anger", "anger", "irritability", "hate", and "disgust". Similarly to the above, from the interpretation of "anger is the inside out of fear", emotional features such as "fear", "worry", and "anxiety" may be included.

Even in such a situation, it is not always easy to expect the restraint of the party X and the comfort of the party Z, as in the example of the conference room described above. Therefore, by the action of the atmosphere cleaning device 100 according to the present invention, it is possible to prevent the "dangerous atmosphere" from reaching a critical state, or to change from a slight "unpleasant atmosphere" to a serious "dangerous atmosphere". Preventing it from reaching the bottom is a broad response to the needs of society.

　As a situation setting, I explained the case where there is a "dangerous atmosphere" in the conference room and elevator as described above. Harsh atmospheres also occur in conference rooms and other closed spaces such as elevators. For example, the closed space is a space where various vehicles such as railroads, automobiles, ships, and space stations cannot easily escape to the outside. The atmosphere cleaning device 100 can clean a harsh atmosphere even if a harsh atmosphere occurs in a closed space other than the conference room and the elevator.

Next, as individual operations performed by the atmosphere cleaning device 100, operations up to extracting emotion feature amounts from people's uttered voices and estimating emotions will be described. FIG. 3 is a diagram for explaining the utterance voice analysis process, and shows the flow of voice emotion recognition. When the uttered voice is input to the sound input unit 1, feature quantities such as the height (fundamental frequency) of the uttered voice and the loudness (power) of the uttered voice are extracted. This is generally called LLD (Low-Level Descriptions), and is divided into frames using a window function and LLD is calculated for each frame, as in the case of performing voice recognition or the like.

　The LLD obtained in this way is time-series data, so the length varies depending on the utterance content. Moreover, it is presumed that the emotion contained in the voice is not expressed by a value for each moment but as a tendency throughout the utterance. Therefore, statistical values such as average value, variance, slope, maximum value, minimum value, etc. are calculated from the LLD, which is the feature value of voice obtained as time-series data. If m statistics are calculated for each of n LLD time-series data, the input speech voice is converted into an n×m-dimensional feature vector. Since the feature amount vector is a vector in which a plurality of feature amounts are combined into one, it may be simply expressed as a feature amount.

The uttered voice analysis unit 2 recognizes the emotion contained in the uttered voice by using the feature amount vector thus obtained. Two types of expressions of emotions are often used: those expressed in categories such as “anger” and “joy” and those expressed in coordinate axes in the emotional space such as “comfort-discomfort” and “wake-sleep”. When the former is used, pattern recognition is performed to estimate the category to which the feature amount vector belongs, and various statistical classifiers such as linear discriminant analysis and support vector machine (SVM: Support Vector Machine) are used. In the latter case, the coordinate value is estimated from the feature vector, and a model such as linear regression, support vector regression (SVR), neural network, etc. is used.

As described above, since the latter part of the speech analysis unit 2 in Fig. 3 is concentrated on the problem of estimating emotions from one feature vector, the pattern recognition method that has been studied many times since then can be used. Therefore, there is not much room for improvement, and rather, what kind of feature amount is used greatly affects the performance. As described above, in addition to features such as the fundamental frequency and power that have been used for a long time, spectrum features such as mel-frequency cepstrum coefficients (MFCC: Mel-Frequency Cepstrum Coefficients) have been added. Furthermore, in order to obtain sufficient performance, the idea is to use all the feature quantities that are likely to be related, instead of carefully examining and selecting the necessary and sufficient feature quantities. It is known that the SVM, which is one of the commonly used classifiers, does not easily cause problems such as overlearning even with a high-dimensional feature amount vector and maintains high performance.

Also, the processing executed by the utterance voice analysis unit 2 shown in FIG. 3 is similarly applied to the processing executed by the action sound analysis unit 3, the second recording unit 5, and the second identification unit 7. However, since the utterance voice and the action sound have greatly different acoustic features, the feature amounts are naturally different.

For example, in the case of uttered speech, differences occur due to the delicate deformation of the vocal tract due to emotions, but there are individual differences, but "anger", "anger", "irritation", "hate", "disgust" It is often experienced that subtle differences in emotions such as, "fear", "worry", and "anxiety" can be distinguished in the sense of hearing, and the fact that a person can distinguish them in terms of the sense of hearing is some acoustic feature for that. Means that there is.

On the other hand, in the case of action sounds, emotions such as “anger”, “anger”, “frustration”, “hate”, and “disgust” require the object to have a place of emotion, and the feeling of consideration for others and things is The characteristics of sound are caused by fading and rough handling of things. However, it is unlikely that the behavioral sound has a more subtle difference in emotional expression than the spoken voice, and due to the difference in the strength and frequency of the mere impact sound, "anger", "irritability", and "calmness" are at most 3 It is appropriate to only distinguish between stages.

FIG. 4 is a diagram illustrating identification boundary information for atmosphere detection recorded in the first recording unit 4, and illustrates an example of identifying four types of emotions from emotion A to emotion D. Although not shown, the utterance voice analysis unit 2 calculates the feature amount vector of the emotions A to D from the utterance voices of the plurality of emotions A to D. As the sound input, labeled speech sounds from emotion A to emotion D are prepared. As described above, the characteristic amounts such as the fundamental frequency, power, and spectrum are extracted, and the statistics are calculated. The feature vector of is derived.

FIG. 5 is a diagram illustrating identification boundary information for identifying an emotion from the feature amount vector for each emotion. FIG. 5A shows emotion distribution, and FIG. 5B shows emotion classification. The vertical axis in FIGS. 5A and 5B indicates the feature amount M. The horizontal axes of FIGS. 5A and 5B represent the feature amount L. In FIG. 5, the features L and M are shown in a two-dimensional space for simplicity. Assuming that the feature vectors of emotions A to D are distributed as shown in FIG. 5A, as a result of performing multi-class classification by the one-to-other classification method of SVM, as shown in FIG. 5B. Different identification boundaries are found by learning.

That is, a boundary AB for identifying emotion A and emotion B, a boundary BC for identifying emotion B and emotion C, a boundary CD for identifying emotion C and emotion D, and a boundary DA for identifying emotion D and emotion A. Desired. As a result, emotion A is identified by boundaries AB and DA, emotion B is identified by boundaries BC and AB, emotion C is identified by boundaries CD and BC, and emotion D is identified by boundaries DA and CD. The boundary AB, the boundary BC, the boundary CD, and the boundary DA obtained here are recorded in the first recording unit 4 of FIG. 1 as identification boundary information and used for emotion identification. Emotions can be classified by the feature space divided by the dotted line based on the identification boundary information.

On the other hand, FIG. 6 is a diagram for explaining the identification boundary information for atmosphere detection recorded in the second recording unit 5, similarly illustrating the case of identifying four types of emotions from emotion A to emotion D. ing. The description is the same as that of FIGS. 4 and 5 except that the input is a labeled action sound, and the details are omitted. The boundary AB, the boundary BC, the boundary CD, and the boundary DA obtained in the same manner are recorded in the second recording unit 5 as identification boundary information and are used for identification.

FIG. 7 is a diagram in which the operation explanatory diagram of FIG. 3 is applied to the block configuration of FIG. In the uttered voice analysis unit 2, the uttered voice is extracted from the voice input signal, and then the feature amount vector is calculated. Then, in the first discriminator 6 configured by a statistical discriminator that captures the discrimination boundary information of emotions related to the utterance voice recorded in the first recording unit 4 in advance, the emotions included in the voice input (utterance voice) To estimate. Furthermore, the first identification unit 6 outputs the identification result to the atmosphere estimation unit 8.

FIG. 8 is a diagram showing an example of emotion estimation by the first identification unit 6. In FIG. 8, the feature amount vector calculated by the speech analysis unit 2 for a new sound input is indicated by a star. When the feature vector of the input voice is located at the star mark in FIG. 8 with respect to the feature space divided by the dotted line based on the identification boundary information from the first recording unit 4, the star mark indicates the boundary AB. Since it exists within the boundary BC, it is identified as emotion B. The identification result of the first identification unit 6 is output to the atmosphere estimation unit 8.

Also, FIG. 9 is a diagram in which the operation explanatory diagram of FIG. 3 is applied to the block configuration of FIG. The action sound analysis unit 3 extracts the action sound from the voice input signal and then calculates the feature amount vector thereof. Then, in the second discriminating unit 7 configured by a statistical discriminator that captures the discrimination boundary information of the emotion related to the action sound recorded in the second recording unit 5 in advance, the emotion included in the voice input (action sound) To estimate. Further, the second identification unit 7 outputs the identification result to the atmosphere estimation unit 8.

Note that, as a method of extracting the uttered voice described in FIG. 7 and the method of extracting the action sound described in FIG. 9, for example, a configuration in which the frequency bands are simply separated by a filter is convenient. Here, how the uttered voice analysis unit 2 extracts the uttered voice and how the action sound analysis unit 3 extracts the action sound will be described.

FIG. 10 is an explanatory diagram of a method of separating an utterance voice from an input voice and an action sound, and shows an example of an emotion utterance voice and an emotional action sound. FIG. 10A is an example of an emotional utterance voice, and shows an amplitude waveform of the emotional utterance voice of "anger" and its spectrogram. FIG. 10B is an example of an emotional action sound, which shows an action waveform (often a strong percussive sound) produced by operating an OA device in a situation where the emotion of “anger” is inherent, and its amplitude waveform and spectrogram. Shows. In FIGS. 10A and 10B, the horizontal axis represents time, the upper vertical axis represents amplitude, and the lower vertical axis represents frequency.

In Fig. 10, the lower spectrogram shows that the darker the black, the stronger the power, and the action sound side has a wide spread of the signal component. On the other hand, the uttered voice has a strong power distribution in the vicinity of several hundred Hz to 5 kHz, but is relatively weak in other regions. Therefore, it is conceivable to separate the uttered voice and the action sound by the low-pass filter and the high-pass filter with the boundary around 5 kHz. For example, the uttered voice analysis unit 2 determines a signal having a frequency of less than 5 kHz as uttered voice and extracts the uttered voice. The action sound analysis unit 3 determines a signal having a frequency of 5 kHz or higher as the action sound, and extracts the action sound. Further, the frequency of less than about 5 kHz is divided into the uttered voice, and the frequency of 5 kHz or more is divided into the action sound, and during the period when the action sound has the power equal to or higher than a predetermined level, the atmosphere estimation unit 8 sets the weight of emotion recognition on the utterance voice side. It is possible to reduce the recognition accuracy due to the overlapping of sounds by lowering the importance of the action sound side.

Next, the atmosphere estimation unit 8 will be described. The atmosphere estimation unit 8 estimates the degree of the atmosphere around the sound generation source based on the emotion identification result identified by the emotion identification unit. FIG. 11 is a diagram showing an operation of estimating an emotion in the atmosphere estimation unit 8 for each predetermined estimation period. In FIG. 11, the uppermost diagram shows the amplitude waveform of the uttered voice, the vertical axis shows the amplitude, and the horizontal axis shows the time. The first identification unit 6 estimates emotions by dividing the estimation period. It is possible to divide the estimation period without overlapping, but by providing the overlapping period as shown in FIG. 11, even if the emotional feature quantity appears prominently in the boundary portion, the period is not divided and is appropriate. You can estimate your emotions.

In the lower part of Fig. 11, the estimated emotions are indicated by circles at the tip of the arrow drawn from the amplitude waveform of the speech voice. Here, out of emotions A to D, A and B are presumed to be estimated by black circles. White circles indicate other emotions (including those not estimated). Further, the description of the portion of the figure that is based on the amplitude waveform of the uttered voice is the same even if it is replaced with the amplitude waveform of the action sound, and thus the description thereof is omitted.

Next, the atmosphere determination process will be explained. FIG. 12 is a diagram showing the operation of expressing the estimated emotion shown in FIG. 11 in a wider period and obtaining the transition of the atmosphere level from the estimated appearance frequency of the emotion in the atmosphere estimation unit 8. The histogram of the appearance frequency of the estimated emotion is shown for each appearance frequency determination period in which several estimated emotion periods are collected. Incidentally, the appearance frequency determination period may also be changed by providing an overlapping period for the same reason as the estimation period.

Further, the atmosphere estimation unit 8 counts the appearance frequency from the emotion A to the emotion D and other appearance frequencies as an example. However, when the atmosphere is digitized based on each emotion, the contribution degree varies depending on the emotion. There is a need to. For example, if you consider emotions such as "anger", "anger", "irritability", "hate", "dislike", etc., "rage" is the highest contribution to "a terrible atmosphere", and "anger", "Hate" follows, followed by "frustration" and "disgust" at relatively low levels. Further, if the frequency of “frustration” and “disgust” is extremely high, the degree of contribution to the “aggressive atmosphere” is high. Therefore, it is necessary to weight the emotions A to D and other appearance frequencies according to the contribution.

For example, the histogram of the appearance frequency of emotions shown in FIG. 12A becomes a histogram of the appearance frequency of emotions after weighting shown in FIG. 12B. In FIG. 12B, the weight is shown for each emotion as emotion A>emotion B>emotion C>emotion D, and the others are shown as zero.

FIG. 12C shows the transition of the mood level and the threshold value obtained by summing the histogram of the appearance frequency of emotions after weighting. The sum value is an atmosphere value. Further, the mood value may be expressed as a numerical value based on information indicating the type of emotion. In this way, the atmosphere estimation unit 8 calculates the atmosphere value for each appearance frequency determination period. One appearance frequency determination period corresponds to one circle as an atmosphere value (atmosphere level), and a transition of the atmosphere value is expressed by sequentially shifting the period. By setting a threshold for this atmosphere value, the severity of the atmosphere can be defined in stages.

In FIG. 12C, the threshold value is given as Th1 to Th3 as the degree of deterioration of the atmosphere increases. For example, when the threshold value exceeds Th3, it is determined that the “aggressive atmosphere” has reached the extreme level, and the values fall to Th2 and Th1. Therefore, the degree of “aggressive atmosphere” is weakening.

Next, the stimulus control unit 9, the stimulus generation unit 10, and the stimulus output unit 11 will be described. The stimulus output unit 11 outputs a stimulus for cleaning the atmosphere to the vicinity of the sound generation source according to the degree of the atmosphere. FIG. 13 is a diagram illustrating an example of a specific configuration of the stimulation control unit 9, the stimulation generation unit 10, and the stimulation output unit 11 based on the estimation result of the atmosphere estimation unit 8. As a stimulus for cleaning the "acute atmosphere", for example, an auditory stimulus, an olfactory stimulus, a visual stimulus, an electromagnetic wave stimulus, a charged fine particle stimulus, etc. are output individually or in combination. The stimulus control unit 9 receives the estimation result of the atmosphere estimation unit 8 and combines a plurality of stimuli, a main control unit 9a for instructing the content of each stimulus control, and a control for each stimulus based on the instruction of the main control unit 9a. The control unit for performing the above, that is, the auditory stimulation control unit 9b, the olfactory stimulation control unit 9c, the visual stimulation control unit 9d, the electromagnetic wave stimulation control unit 9e, and the charged particle stimulation control unit 9f.

The stimulus generation unit 10 generates a stimulus for each stimulus control unit 9b to 9f, that is, an auditory stimulus generation unit 10b, an olfactory stimulus generation unit 10c, a visual stimulus generation unit 10d, and an electromagnetic wave stimulus generation unit. 10e and a charged particle stimulus generation unit 10f. The stimulus output unit 11 outputs an stimulus in a form suitable for each stimulus generation unit 10b to 10f, that is, an auditory stimulus output unit 11b, an olfactory stimulus output unit 11c, and a visual stimulus output unit 11d. An electromagnetic wave stimulation output unit 11e and a charged particle stimulation output unit 11f are provided. Each stimulus output from the stimulus output unit 11 is output to people, that is, around the sound source, according to the estimation result of the atmosphere estimation unit 8. Hereinafter, the content of each stimulus will be described individually.

First, I will explain about auditory stimulation. Auditory stimulus is the sound that cleans a harsh atmosphere. The auditory stimulus is composed of sounds that are said to have parasympathetic nerve enhancement and relaxing effects as effects on humans, for example, sounds that are emitted in the natural environment, especially sounds that accompany unpredictable movements of water and the atmosphere. "Natural sound" is used as a stimulus. Specifically, the sound of waves returning to the beach, the sound of brooks, the sound of clear waterfalls, the sound of rain falling on leaves, and the sound of the branches and leaves of trees rubbing against the wind. Furthermore, in addition to these combinations, even those that overlap, such as the chirping of birds, the crying of insects, the crying of animals, or those reconstructed as sounds that make you feel the depth and breadth of natural space, It is used as a sound stimulus.

The acoustic characteristic of this "natural sound" is that it is accompanied by unpredictable movement as described above, and the power becomes weaker as the frequency becomes higher like pink noise, that is, the power is inversely proportional to the frequency f. The characteristic is that it has "/f fluctuation". Aside from the case where you concentrate your attention on a specific feature contained in "natural sound," "natural sound" is used in everyday life, as if you were in a natural space by listening to it. It is speculated that by reproducing the annoyance, worries, and the situation of withdrawal from stress, parasympathetic nerve enhancement and relaxing effects can be obtained.

　In addition, auditory stimulus may cause parasympathetic nerve enhancement or relaxing effect depending on the relationship with human memory. For example, when a melody sound that reminds of a heartwarming story or one's own experience is heard, it is presumed that a parasympathetic nerve is enhanced or a relaxing effect is generated, and the melody sound may be used as an auditory stimulus. Specifically, it can be nursery rhymes, songs, anime and drama music, etc., but it is desirable that they have universal support with little bias in personal taste. In addition, some classical music is said to have a high relaxing effect, and for example, performance sounds such as Ravel's composition "Pavane for the Dead Princess" may be used as auditory stimuli, but in that case as well It is desirable to have universal support with little bias in taste. In addition, as a temperamental characteristic, not the main temperament in modern music (the temperament that causes the chords to undulate because the scale is mathematically indivisible), but the pure temperament (the scale has a simple integer ratio relationship, It may be a musical composition or melody sound based on a temperament in which a chord does not swell.

Also, use a sound that impresses "laughter" or "smileyness" as a sound stimulus for enhancing the parasympathetic nerve or relaxing effect. The process of expressing "laughter" by sound stimulation is (1) reminiscent of a specific scene in the past by sound stimulation, (2) making you feel happy, (3) laughing (expression by physiological body reaction) ), the three stages, "laughter" is evoked. However, with respect to (1), the sound stimulation itself may be interesting, and it may be directly connected to the pleasant feeling of (2). As a study of "laughter" on the autonomic nervous system, observations based on changes in blood flow in the brain and heart rate variability were made, and after stimulating "laughter" the sympathetic nerve was temporarily enhanced, and then the parasympathetic nerve It has been reported to increase and show a tendency to relax.

Sound stimulus that evokes “laughter” is a sound effect that impresses punches (end of “laughter”) and blur (misguided behavior and actions made to induce “laughter”) in comedy programs and controls. Or theme music is equivalent. As a specific music piece, "Somebody Stole My Gal" played by Pee Wee Hunt, an American jazz musician, is used. Even if it is not this performance itself, a similar "laughing" evoking effect can be expected if it is a melody sound reminiscent of the Pee Wee Hunt performance. This song is the theme song of "Yoshimoto Shin Comedy", and the causal relationship with "laughter" is strongly remembered by those who have something to do with the Kansai cultural area, mainly people from the Kansai region or those who have ties to the Kansai area. It is estimated that By the way, it is possible that the acoustic features included in this song may cause everyone to laugh, but it has not been verified.

FIG. 14 is a diagram showing an example of a configuration for outputting the auditory stimulation described above, and more specifically showing the auditory stimulation system shown in FIG. The auditory stimulus output unit 11b outputs a natural sound composed of a sound generated by the movement of water or the atmosphere as an auditory stimulus for cleaning the atmosphere. Alternatively, the auditory stimulus output unit 11b outputs a music or melody sound as an auditory stimulus for cleaning the atmosphere. The auditory stimulation control unit 9b transmits to the auditory stimulation generation unit 10b an instruction to generate a sound for cleaning a bad atmosphere. The auditory stimulation generation unit 10b includes an audio signal reproduction unit 10b1 and an audio signal data storage unit 10b2. Audio signal data such as the aforementioned "natural sound, music, melody sound" is stored in the audio signal data storage unit 10b2. Based on the instruction of the auditory stimulation control unit 9b, the predetermined audio signal data stored in the audio signal data storage unit 10b2 is read by the audio signal reproducing unit 10b1 and reproduced as an audio signal to configure the auditory stimulation output unit 11b. It is output to the space as an auditory stimulus from the speaker 11b.

Next, I will explain olfactory stimulation. The olfactory stimulus is a scent that cleans a harsh atmosphere. As the olfactory stimulus, for example, a diluted plant-derived aromatic component, so-called essential oil (essential oil) or aroma oil (containing a synthetic component) is used as the stimulus. According to the Aroma Environment Association (AEAJ), essential oils are natural aroma substances extracted from plant flowers, leaves, peels, fruits, heartwood, roots, seeds, bark, resins, etc. Has scent and function.

Among them, what is used as an olfactory stimulus in the present invention has a sedative effect that calms the work of the nervous system and relaxes the work of the mind and body. Specifically, ylang-ylang (effect ingredient: linalool), orange sweet (same: limonene), chamomile roman (same: angelica acid), clary sage (same: linalyl acetate, linalool), grapefruit (same: limonene), Herbs and citrus fruits such as lavender (the same: linalyl acetate, linalool), bergamot (the same: limonene, linalool), peppermint (the same: menthol), pine (the same: pinene), cedar (the same: pinene), cypress (Same: Pinene), kuromoji (Same: Linalool) and other tree-based ones are used alone or in combination.

FIG. 15 is a diagram showing an example of a configuration for output processing the olfactory stimulus described above, and more specifically shows the olfactory stimulus system shown in FIG. The olfactory stimulus output unit 11c outputs a diluted essential oil or a diluted aroma oil as an olfactory stimulus for cleaning the atmosphere. The olfactory stimulus control unit 9c transmits a scent generation instruction to the olfactory stimulus generation unit 10c. The olfactory stimulus generation unit 10c includes a cartridge operation unit 10c1 and scent component cartridges 10c2 to 10c4. Although three cartridges are shown in FIG. 15 for simplification of description, the number is not limited to this. In the olfactory stimulus generator 10c, the scent component cartridges 10c2 to 10c4 hold different scent components. The scent component is generated alone or in a plurality of blends according to an operation signal from the cartridge operation unit 10c1 based on an instruction from the olfactory stimulation control unit 9c. The olfactory stimulus output unit 11c releases the generated scent component.

As the scent component cartridges 10c2 to 10c4, for example, a spray structure that holds a liquid obtained by diluting the essential oil that serves as the scent component and atomizes by using mechanical movement by a high-pressure gas or a piezo element can be considered. In addition, an ultrasonic atomization separation structure in which ultrasonic waves are applied to the diluted liquid to atomize it may be used. The olfactory stimulation output unit 11c includes a nozzle 11c1, a guide tube 11c2, a fan 11c3, and an output operation unit 11c4. The scent component atomized in the olfactory stimulus generation unit 10c is guided to the nozzle 11c1 through the guide tube 11c2 and is discharged. The olfactory stimulus emitted from the scent component cartridge is operated by the output operation unit 11c4 based on the instruction of the olfactory stimulus control unit 9c to change the direction of the nozzle 11c1 in order to guide it to the intended direction or to operate the air volume of the fan 11c3. To do

Note that the scent output by the olfactory stimulus output unit 11c may remain at the place where the scent was output. Therefore, the olfactory stimulus output unit 11c may output the deodorant component a predetermined time after outputting the scent. Thereby, the atmosphere cleaning device 100 can deodorize the scent.

Next, I will explain visual stimulation. A visual stimulus is at least one of information and light that cleans a hostile atmosphere. Image output or lighting control is used as a visual stimulus. The image outputs a natural landscape regardless of whether it is a moving image or a still image. It is assumed that the contents associated with the auditory stimulus and the olfactory stimulus can reinforce each other to obtain the enhancement of parasympathetic nerve and the relaxing effect. For example, in the case of "natural scenery" such as forests and mountain streams, there is a high affinity with "natural sounds" such as the sound of brooks, the sound of clear waterfalls, and the sound of the branches and leaves of trees rubbing against the wind. It is assumed that it has a high affinity with the scent of the system.

As for lighting control, strengthen the blue light that has been confirmed to have a high sedative effect, enhance the green light that reminds you of the freshness of sunlight through the trees, and use orange light that reminds you of warmth, security, and nostalgia. Controls the color tone such as strengthening. By also controlling the contents in association with the auditory and olfactory stimuli, it is possible to reinforce each other to obtain parasympathetic nerve enhancement and relaxing effects.

FIG. 16 is a diagram showing an example of a configuration for outputting the visual stimulus described above, and more specifically showing the visual stimulus system shown in FIG. In FIG. 16, the visual stimulus generation unit 10d and the visual stimulus output unit 11d shown in FIG. 13 are provided in each of two systems (image display type and lighting operation type), but the present invention is not limited to this and only one may be provided. The visual stimulus output unit 11d outputs an image, a natural landscape existing with the movement of water and the atmosphere, or illumination light as a visual stimulus for cleaning the atmosphere. The visual stimulus output unit 11d uses, as the illuminating light, illuminating light accompanied by a color tone change that individually intensifies the blue, green, and orange lights. The visual stimulus control unit 9d transmits a visual stimulus output instruction to the visual stimulus generation unit 10d. The visual stimulus generation unit 10d includes an image signal data storage unit 10d2, an image signal reproduction unit 10d1, an illumination control data storage unit 10d4, and an illumination drive unit 10d3. The image signal data storage unit 10d2 stores image information. The image signal reproduction unit 10d1 reads the image information based on the output instruction from the image information storage unit 173a. The image signal reproduction unit 10d1 outputs an image signal based on the read image information to the monitor 11d1. Thereby, the monitor 11d1 can display the image information. The monitor 11d1 is an example of the visual stimulus output unit 11d.

The lighting control data storage unit 10d4 stores control data for controlling the lighting 11d2. Based on the instruction from the visual stimulus control unit 9d, the illumination drive unit 10d3 controls the illumination control data storage unit 10d4 to read the stored illumination control data and output a drive signal to the illumination 11d2. Thereby, the illumination 11d2 can output light. The illumination 11d2 is, for example, an illumination having a function of controlling the color tone, and may emit light having a color shade corresponding to the input drive signal directly or indirectly by reflection. Further, the visual stimulus generation unit 10d may adjust the intensity of the light output by the illumination 11d2.

Next, I will explain electromagnetic wave stimulation. Electromagnetic stimulation is an electromagnetic wave that cleans a harsh atmosphere. As electromagnetic wave stimulation, a standing wave generated by propagating extra-long waves generated by lightning discharge in a spherical shell-shaped cavity between the earth and the ionosphere, approximately 7.83 Hz based on so-called Schumann resonance (resonance) An electromagnetic wave having a frequency is simulated and used as a stimulus. As an effect on humans, in contrast to the high-frequency electromagnetic waves radiated from innumerable electric devices in modern society, the ultra-long-wave electromagnetic waves that have continued to exist in the history of biological evolution have increased parasympathetic nerves. Regarding the relaxation effect and the relaxation effect, there are cases where attention is focused on the fact that it is close to the frequency of the electroencephalogram α wave.

FIG. 17 is a diagram showing an example of a configuration for output processing of the electromagnetic wave stimulation described above, and more specifically shows the electromagnetic wave stimulation system shown in FIG. The electromagnetic wave stimulus control unit 9e transmits an electromagnetic wave stimulus output instruction to the electromagnetic wave stimulus generation unit 10e. The electromagnetic wave stimulus generation unit 10e is composed of an electromagnetic wave drive unit 10e1 and an ultra long wave oscillation unit 10e2. The ultra-long wave oscillating unit 10e2 includes an electronic circuit that oscillates at a frequency of about 7.83 Hz, and oscillates based on a command from the electromagnetic wave driving unit 10e1 to output an ultra-long wave signal. The electromagnetic wave driving unit 10e1 drives the input ultra-long wave signal with a predetermined electric power, supplies a current based on the ultra-long wave signal to the antenna 11e, and outputs an electromagnetic wave stimulus. The antenna 11e is an example of the electromagnetic wave stimulation output unit 11e.

Next, I will explain the stimulation of charged particles. Charged particle stimuli are charged particles that clean a harsh atmosphere. As the charged particle stimulus, fine water particles negatively charged by the Leonard effect are used as the stimulus. Originally, water droplets are polarized, and negative charges are distributed on the surface, and positive charges tend to be distributed inside.Therefore, when water droplets collide mechanically and are finely divided, relatively small water particles are generated. Are negatively charged (generally also called negative or negative ions), and large water particles tend to be positively charged. Large water particles fall quickly due to the action of gravity, but small water particles float and become an air mass in which negative ions predominate. In the natural world, these negative ions are generated around waterfalls and mountain streams, and it has been reported that they have an effect on humans to enhance parasympathetic nerves and have a relaxing effect. Artificially, for example, ultrasonic atomization technology is used in humidifiers, etc., and the amount of atomization can be adjusted by controlling the size of the input to an ultrasonic transducer placed in water. The size of water particles can be adjusted by controlling the frequency.

　Although this charged particle stimulus may be used alone, it is possible to reinforce each other by enhancing parasympathetic nerves and relaxing effects by making the contents associated with auditory stimuli, olfactory stimuli, and visual stimuli. To be specific, "natural sounds" such as the sound of brooks and the sound of clear waterfalls, tree-based scents, landscape images of forests and mountain streams, and charged particles can be output in association with each other. The natural environment can be reproduced more realistically, and as a result, it becomes possible to obtain the parasympathetic nerve promotion and relaxing effect.

FIG. 18 is a diagram showing an example of a configuration for outputting the charged particle stimulus described above, and more specifically showing the charged particle stimulus system shown in FIG. The charged particle stimulation control unit 9f sends an output instruction to the charged particle stimulation generation unit 10f. The charged particle stimulus generation unit 10f includes an element drive unit 10f1 and an atomized particle water generation unit 10f2. The charged particle stimulation output unit 11f includes a nozzle 11f1, a guide tube 11f2, a fan 11f3, and an output operation unit 11f4. The particle water atomized in the charged particle stimulus generation unit 10f is guided to the nozzle 11f1 through the guide pipe 11f2 and is discharged. The charged particles discharged from the atomized particle water generation unit 10f2 are changed in the direction of the nozzle 11f1 in order to be guided to the intended direction by the operation of the output operation unit 11f4 based on the instruction of the charged particle stimulation control unit 9f, and the fan. Operate the air volume of 11f3. In addition, the guide tube 11f2 and the nozzle 11f1 through which the charged particles pass may be partially negatively charged to supplement the charge amount of the charged particles.

Here, since the main control unit 9a identifies the stimulus to be output with reference to the stimulus control matrix diagram, an example of the stimulus control matrix diagram will be described. FIG. 19 is a matrix diagram showing an example of stimulation control. Although FIG. 13 shows a configuration in which five categories of stimuli are output, FIG. 19 shows three categories for the sake of simplification of explanation, four types of stimuli K, K1 to K4, and stimuli L, L1 to L3. It is assumed that there are three types and three types of stimulation M, M1 to M3. It is shown that the larger the number of each stimulus, the stronger the action of cleaning the deterioration of the atmosphere. For example, when the stimulus K is an auditory stimulus (that is, sound), the stimulus K1 is a natural sound and the stimulus K2 is classical music.

Also, the stimulus control matrix diagram expresses the degree of the harsh atmosphere in five levels. In the description of FIG. 12, the case where there are three threshold values is illustrated, but it may be considered that the case where the threshold value is 20 is illustrated in the matrix diagram of the stimulation control illustrated in FIG. 19. For example, when the degree of a bad atmosphere is 1, the atmosphere value is equal to or more than the threshold Th1 and less than the threshold Th2. When the degree of the severe atmosphere is 2, the atmosphere value is equal to or more than the threshold Th2 and less than the threshold Th3. As described above, the stimulus control matrix diagram shows the relationship between the degree of a bad atmosphere and the stimulus to be output.

In FIG. 19, the horizontal direction shows the degree of deterioration of the atmosphere, and is divided into 5 stages from mild to severe, and each stage is displayed in 4 sections. It means that the more you move to the right, the worse the atmosphere becomes. In a region where the degree of atmosphere deterioration is mild, one of the stimuli K1, K2, L1, and L2 is used, and two stimuli are gradually used at the same time according to the deterioration, and further the stimuli M1 to M3 are added. By simultaneously using three stimuli, the stimulus control method is used so as to enhance the action of cleaning the atmosphere deterioration. For example, by providing the main controller 9a in FIG. 13 with a function of combining a plurality of such stimuli, it is possible to flexibly cope with a situation in which the atmosphere deteriorates gradually. Further, although FIG. 19 does not show the control of the intensity or amount of each stimulus, it goes without saying that such a control can stepwise enhance the cleaning action for the deterioration of the atmosphere, and it goes without saying that the above-mentioned olfactory stimulus and charging are performed. Particle stimulation is applicable.

In the atmosphere cleaning device according to the first embodiment configured as described above, as an environment surrounding people, an inappropriate situation from a slight “unpleasant atmosphere” to a serious “severe atmosphere” The acquired audio signal is detected by analysis, and a stimulus that has the effect of changing the emotions or emotions of people is output according to the degree of inadequacy of the atmosphere, so it is possible to prevent further deterioration of the atmosphere. At the same time, the effect of being able to prevent a critical situation is obtained.

Further, in the atmosphere cleaning device according to the first embodiment, a stimulus having an action of guiding people's emotions or emotions in a gentle direction is output according to the degree of inappropriateness of the atmosphere, so that the atmosphere is further improved. As a result, it is possible to prevent deterioration and prevent a critical situation.

Further, in the atmosphere cleaning device according to the first embodiment, water is used as an auditory stimulus having an effect of changing people's emotions or emotions or an effect of guiding the person in a gentle direction according to the degree of inappropriateness of the atmosphere. Or, it outputs a natural sound composed of sounds that accompany the movement of the atmosphere, so that it is possible to prevent further deterioration of the atmosphere without hindering the concentration of consciousness and to prevent a critical situation. The effect that can be obtained is obtained.

Moreover, in the atmosphere cleaning device according to the first embodiment, music is used as an auditory stimulus having an effect of changing people's emotions or emotions or an effect of leading to a gentle direction according to the degree of inappropriateness of the atmosphere. Or, by outputting a melody sound, it is possible to prevent further emotional deterioration by suppressing the emotional emotions caused by the parties remembering a heartwarming story or their own experience, and preventing a critical situation. The effect of being able to do is obtained.

Further, in the atmosphere cleaning device according to the first embodiment, as an auditory stimulus having an action of changing people's emotions or emotions or an action of leading to a gentle direction according to the degree of inappropriateness of the atmosphere, By using sound effects and theme music that impress or blur, laughter is aroused, and further deterioration of the atmosphere can be prevented, and it is possible to prevent a critical situation.

Further, in the atmosphere cleaning device according to the first embodiment, as an auditory stimulus having an effect of changing people's emotions or emotions or an effect of leading to a gentle direction according to the degree of inappropriateness of the atmosphere, The jazz musician Pee Wee Hunt performed "Somebody Stole My Gal" or its imitation performance, or the melody sound reminiscent of the Pee Wee Hunt performance evoked a laughter and prevented further deterioration of the atmosphere. At the same time, the effect of being able to prevent a critical situation is obtained.

Further, in the atmosphere cleaning device according to the first embodiment, the essential oil is used as an olfactory stimulus having an effect of changing people's emotions or emotions according to the degree of inappropriateness of the atmosphere, or an effect of leading to a gentle direction. Since it uses a dilution of (essential oil) or aroma oil (containing synthetic components), stimulation is rapidly transmitted to the limbic system, preventing further deterioration of the atmosphere in a short period of time, and in a critical situation It is possible to obtain the effect that it can be prevented from reaching.

In addition, in the atmosphere cleaning device according to the first embodiment, water is used as a visual stimulus having an action of changing people's emotions or emotions according to the degree of inadequacy of the atmosphere, or a action of guiding in a gentle direction. Or image output such as "natural scenery" that reminds the freshness of plants and trees that exist with the movement of the atmosphere, or enhance the blue light that has been confirmed to have a high sedative effect, reminiscent of the freshness of sunbeams Since we use lighting control with color tone changes such as intensifying the greenish light and intensifying the orangeish light that reminds us of warmth, security and nostalgia, we can transmit a large amount of information to the brain at once. In addition, it is possible to prevent the atmosphere from further deteriorating in a short period of time and to prevent a critical situation.

In addition, in the atmosphere cleaning device according to the first embodiment, Schumann is used as an electromagnetic wave stimulus having an effect of changing people's emotions or emotions according to the degree of inappropriateness of the atmosphere, or an effect of guiding in a gentle direction. Since an electromagnetic wave having a frequency of about 7.83 Hz based on resonance is used, it is possible to simultaneously transmit stimuli to a large number of people without directly using the five senses stimuli that are susceptible to disturbances, which further deteriorates the atmosphere. It is possible to prevent such a situation as well as prevent a critical situation.

Further, in the atmosphere cleaning device according to the first embodiment, as charged particle stimuli having an effect of changing people's emotions or emotions or an effect of leading to a gentle direction, depending on the degree of inappropriateness of the atmosphere, Since fine water particles negatively charged by the Leonard effect are used, it is possible to convey to the brain the freshness as if one were in the natural environment, so that it is possible to effectively prevent further deterioration of the atmosphere, The effect is that it is possible to prevent reaching a critical situation.

In addition, the atmosphere cleaning device according to the first embodiment detects an inappropriate situation from a slight “unpleasant atmosphere” surrounding people to a serious “dangerous atmosphere”, and mainly detects people. By outputting a stimulus that has the effect of leading to a calm state and cleaning the atmosphere, it is possible to clean up stressful inappropriate situations before the crisis and prevent deteriorating health and crime. The effect of being able to be obtained is obtained.

Embodiment 2.
Although the emotion feature amount of the action sound used in the action sound analysis unit has been described in the first embodiment, the emotion feature amount will be described more specifically in the second embodiment. FIG. 20 is a diagram showing an example of a behavior sound analysis method. In addition, FIG. 21 is a diagram illustrating an example of a result of action sound analysis. As mentioned above, as emotional behavior sounds, feelings of "anger", "anger", "irritability", "hate", "disgust", etc. make us less considerate of others and things, and make things violent. It is expected that the characteristics of the sound due to being handled, for example, peak components will appear strong and many. Therefore, an acoustic feature quantity suitable for detecting a peak component called a crest factor is used.

FIG. 20 shows a method of analyzing the acquired action sound, in which an analysis period of a predetermined length is set and the crest factor for each section is calculated by shifting the period by half. This method of determination is a general analysis method in the field of speech analysis, and each analysis period is called a “short-time frame”, and how much the adjacent sections are shifted is called a “frame cycle”. In the analysis in the frequency domain, it is common to apply some window function to the short time frame in order to avoid the inconvenience caused by the discontinuity at both ends of the frame. However, since the crest factor is a feature amount in the time domain and has no relation to discontinuity at both ends of the frame, the window function need not be used. The crest factor definition is shown in the lower part of FIG. The maximum value of the absolute value is obtained for each analysis period, that is, for each short-time frame, and is similarly divided by the effective value for each short-time frame to be called a crest factor. The effective value is obtained by calculating the square root of the root mean square for each short time frame. The crest factor is calculated from the equation (1). Here, the effective value is the root mean square obtained during the analysis period.

FIG. 21 shows, as an example of the result of analysis of action sounds, comparison of action sounds when a keyboard operation of a personal computer is performed during normal times and during frustration. 21, the vertical axis represents the number of frames in which the crest factor exceeds a predetermined value, and the horizontal axis represents the number of frames in which the peak exceeds a predetermined value. FIG. 21 shows the result of sampling the action sound during normal times and the action sound during irritation a plurality of times. The black circles indicate normal times, and the X marks indicate irritation. For example, based on the behavior sound during frustration as shown in FIG. 20, the number of analysis periods in which the crest factor exceeds a predetermined value (first feature amount) and the number of analysis periods in which the peak component exceeds a predetermined value (first Two feature quantities) are calculated. Similarly, a plurality of analysis periods in which the crest factor exceeds a predetermined value and a plurality of analysis periods in which the peak component exceeds the predetermined value are calculated based on a plurality of action sounds during frustration. The calculated results are X marks at the time of irritation.

From Figure 21, it can be seen that there is a bias in the distribution during normal times and during frustration. From this fact, by observing the frequency at which the crest factor and the peak value exceed the predetermined values, it is possible to discriminate the emotional behavior and the normal behavior sound.

In the atmosphere cleaning device according to the second embodiment configured as described above, when detecting the atmosphere of the place from the behavior sounds of people, the characteristics of the behavior sounds are the crest factor and the frequency at which the peak value exceeds the predetermined value. Since it is configured to be used, it is possible to improve the usability of emotion identification by action sound, effectively utilize action sound that tends to interfere with emotion identification by uttered voice, and improve the robustness of atmosphere detection. Is obtained.

Embodiment 3.
FIG. 22 is a functional block diagram when the atmosphere cleaning device is applied to an elevator. The present embodiment shows an embodiment in which the atmosphere cleaning device is applied to an elevator. The third embodiment will mainly describe matters different from the first embodiment, and the explanation of matters common to the first embodiment will be omitted.

In FIG. 22, the atmosphere cleaning device 200 is coupled to the elevator car 300, the sound input unit 201 acquires a sound generated in the elevator car 300, and the stimulus output unit 202 outputs a stimulus into the elevator car 300. It is configured to do. Further, although the inside of the atmosphere cleaning device 200 is simplified and described only in the main parts of the atmosphere estimation unit 203 and the stimulus generation unit 204, the configuration is actually the same as the block diagram of FIG. 1. There is a party X and a party Y, and the party X is unilaterally agitated with the party Y, and the atmosphere cleaning device 200 also applies a series of sounds to the elevator as described above. Atmosphere is detected by and the atmosphere is cleaned by outputting various stimuli.

FIG. 23 is a diagram showing an example of a serious situation when the atmosphere cleaning device is applied to an elevator. FIG. 22 shows an example in which the situation shown in FIG. 22 has progressed to a more serious situation. The party X has taken a posture of taking out the weapon and covering the party Y. Or, as a situation that is not so serious, the situation where the party X is about to hit and grab the chest of the party Y is about to hit. The party X generally has a posture of holding the party Y in a so-called “blunted” state where the front and rear are indistinguishable. At least, it is difficult to imagine a situation in which the party X on the offending side is below the party Y. In such a serious condition, the outputs of the various stimuli described in the first embodiment may not be able to clear the situation before a serious crime occurs. In the situation shown in FIG. 23, there is little time margin to wait until the parasympathetic nerve is enhanced or the relaxing effect is produced by the output of the stimulus. It is not enough to have an olfactory stimulus that can provide a sedative effect that quickly transmits limbic system to the limbic system, calms the nervous system, and relaxes the work of the mind and body.

Therefore, for a serious situation similar to the situation shown in FIG. 23, an olfactory stimulus as shown below is output. The configuration for outputting the olfactory stimulus other than the content of the stimulus is the same as in FIG. As an olfactory stimulus, a malodorous substance diluted to such an extent that it is not directly harmful to human health is used. Specifically, as a plant system, Rafflesia (a malodorous component: indole, amines, etc.), Shokdai Okonjak (the same: dimethyl trisulfide), Durian (the same: propanethiol, etc.), American Hornbill, Dead Force Arum, It uses the malodorous components emitted by sycamore, Hidnola africana, Ginnan (butyric acid, heptanoic acid), quince, Dokudami (decanoyl acetaldehyde). As an animal system, the odorous components emitted by stink bugs, camels, skunks, vultures, millipedes, and zorillas are used. In addition, organic sulfur compounds used as odorants for city gas and propane gas are used. In addition to the plants and animals and organic sulfur compounds exemplified here, it is possible to use as long as it is diluted to such an extent that the human sense of smell recognizes an offensive odor and is not directly harmful to human health. .. Thereby, in the elevator car 300, it is possible to prevent the party X from calming or crime. The fact that the party X can be sedated or the crime is prevented can be said to temporarily cleanse the harsh atmosphere.

In addition, in the output of the olfactory stimulus in this case, the control may be performed so as to output the olfactory stimulus mainly to the party X who is the aggressor. Specifically, the output direction of the olfactory stimulus is controlled using the nozzle 11c1 and the fan 11c3 shown in FIG. At this time, a camera image installed in the elevator car 300 may be used to specify the position of the party X and determine the output direction of the olfactory stimulus. In addition, taking into consideration that the average head length of the Japanese is about 20 cm, and that it is assumed that the party X is likely to hold the party Y in a serious situation, 25 cm or more from the floor surface is used. By controlling such as setting the output direction of the olfactory stimulus centering on the space, the olfactory stimulus can be intensively affected on the party X side.

FIG. 24 is a functional block diagram when an atmosphere cleaning device is applied to an elevator, a malodorous component is used as an olfactory stimulus, and ventilation processing is performed. As described above, when a malodorous substance is used as the olfactory stimulus, the malodor remaining in the elevator car 300 for a long time hinders the operation of the elevator. Therefore, as shown in FIG. 24, it is effective to interlock with a mechanism for forming an airflow toward the ceiling from a position close to the floor surface, for example, a position less than 25 cm from the floor surface for ventilation.

In FIG. 24, like FIG. 22 and the like, a part of the configuration of FIG. 1 is extracted and described, and the description of the overlapping parts will be omitted. In FIG. 24, a ventilation fan 407 is installed on the ceiling of the elevator car 300, a ventilation port 408 is installed near the floor surface of the elevator car 300, and the atmosphere cleaning device 400 is equipped with a ventilation fan control unit 406. The ventilation fan control unit 406 receives the control signal from the stimulation control unit 404. The ventilation fan control unit 406 controls the ventilation fan 407 according to the control signal.

The ventilation fan control unit 406 operates the ventilation fan 407 installed on the ceiling portion of the elevator car 300 to flow the air for ventilating the inside of the elevator car 300 from the ventilation port 408 near the floor toward the ventilation fan 407. Cause Then, the rising air and scent are discharged to the outside. Further, the atmosphere cleaning device 400 holds a deodorant component in addition to the malodorous component used as the olfactory stimulus, and the stimulation control unit 404 outputs the malodorous component to the elevator car 300, and after a predetermined time, the deodorant component. May be output.

In the elevator to which the atmosphere cleaning device according to the third embodiment configured as described above is applied, an olfactory interruption is given to the perpetrator's consciousness in a serious situation immediately before a serious crime occurs. As a result, the behavior of the perpetrator can be temporarily suspended, and crime prevention and safety in the closed space can be improved. If the application is to an elevator, it will be possible to secure the time required for an emergency response such as an emergency stop on the nearest floor or rushing security guards, and the effect of improving crime prevention and safety of the elevator can be obtained.

Further, in the elevator to which the atmosphere cleaning device according to the third embodiment is applied, as a stimulus for applying an olfactory interruption to the consciousness of the assailant side, a diluted malodorous component or an organic sulfur compound emitted by a plant or an animal. Since it is configured to use, it is possible to temporarily suspend the actions of the perpetrator and improve the crime prevention and safety of the closed space. If the application is to an elevator, it will be possible to secure the time required for an emergency response such as an emergency stop on the nearest floor or rushing security guards, and the effect of improving crime prevention and safety of the elevator can be obtained.

The features of the embodiments described above can be combined with each other as appropriate. Further, although the case where the atmosphere cleaning device is applied to the elevator has been described in the third embodiment, the present invention is not limited to this. For example, in the case of various vehicles such as railroads, automobiles, ships, and closed spaces such as space stations that cannot easily escape to the outside, the same applies by applying the atmosphere cleaning device described in the first to third embodiments. The effect of is obtained.

Here, the hardware configuration of the atmosphere cleaning device will be described. FIG. 25 is a diagram showing a hardware configuration of the atmosphere cleaning device. The atmosphere cleaning device 100 includes a processor 101, a volatile storage device 102, and a non-volatile storage device 103.

The processor 101 controls the entire atmosphere cleaning device 100. For example, the processor 101 is a CPU (Central Processing Unit), an FPGA (Field Programmable Gate Array), or the like. The processor 101 may be a multiprocessor. The atmosphere cleaning device 100 may be realized by a processing circuit, or may be realized by software, firmware or a combination thereof. The processing circuit may be a single circuit or a composite circuit.

The volatile storage device 102 is a main storage device of the atmosphere cleaning device 100. For example, the volatile storage device 102 is a RAM (Random Access Memory). The non-volatile storage device 103 is an auxiliary storage device of the atmosphere cleaning device 100. For example, the non-volatile storage device 103 is an SSD (Solid State Drive).

Utterance voice analysis unit 2, action sound analysis unit 3, first identification unit 6, second identification unit 7, first recording unit 4, second recording unit 5, atmosphere estimation unit 8, stimulus control unit 9, and stimulus generation unit. 10 can be realized by an information processing device. Further, the first recording unit 4 and the second recording unit 5 may be realized as a storage area secured in the volatile storage device 102 or the non-volatile storage device 103.

Of the sound input unit 1, the utterance voice analysis unit 2, the action sound analysis unit 3, the first identification unit 6, the second identification unit 7, the atmosphere estimation unit 8, the stimulation control unit 9, the stimulation generation unit 10, and the stimulation output unit 11. Some or all may be realized by the processor 101. A part or all of the uttered voice analysis unit 2, the action sound analysis unit 3, the first identification unit 6, and the second identification unit 7 may be realized by the processor 101.

Of the sound input unit 1, the utterance voice analysis unit 2, the action sound analysis unit 3, the first identification unit 6, the second identification unit 7, the atmosphere estimation unit 8, the stimulation control unit 9, the stimulation generation unit 10, and the stimulation output unit 11. Some or all may be realized as a module of a program executed by the processor 101. Part or all of the uttered voice analysis unit 2, the action sound analysis unit 3, the first identification unit 6, and the second identification unit 7 may be realized as a module of a program executed by the processor 101. For example, the program executed by the processor 101 is also called an atmosphere cleaning program.

1,201,401 sound input unit, 2 utterance voice analysis unit, 3 action sound analysis unit, 4 first recording unit, 5 second recording unit, 6 first identification unit, 7 second identification unit, 8,203 atmosphere estimation Section, 9,404 stimulation control section, 9a main control section, 10,204,405 stimulation generation section, 11,202,402 stimulation output section, 100,200,400 atmosphere cleaning apparatus, 101 processor, 102 volatile storage apparatus , 103 non-volatile storage device, 300 elevator car, 406 ventilation fan control unit, 407 ventilation fan, 408 ventilation port.

Claims

An input part to get the sound,
A feature amount derived from the emotion of the person included in the sound, and an emotion identification unit that identifies the emotion of the person based on boundary information for identifying the emotion of the person from the feature amount,
An atmosphere estimation unit that estimates the degree of the atmosphere around the source of the sound based on the identification result of the emotion of the person identified by the emotion identification unit,
A stimulus output unit that outputs a stimulus that purifies the atmosphere according to the degree of the atmosphere to the vicinity of the sound source,
Atmosphere cleaning device having.
The sound is a speech voice,
The emotion identifying unit uses the feature amount calculated from the uttered voice,
The atmosphere cleaning device according to claim 1.
The sound is a human action sound,
The emotion identifying unit uses the feature amount calculated from the action sound,
The atmosphere cleaning device according to claim 1.
The emotion identifying unit uses the number of analysis periods in which the crest factor calculated for each analysis period of the action sound exceeds a predetermined value as the first feature amount, and sets the action sound detected for each analysis period of the action sound to the action sound. The number of analysis periods in which the included peak component exceeds a predetermined value is the second feature amount, and the boundary information for identifying emotions from the relationship between the first feature amount and the second feature amount is determined, and the person's Identify emotions,
The atmosphere cleaning device according to claim 3.
The sound is a speech voice and a human action sound,
The emotion identifying unit uses the feature amount calculated from the uttered voice and the action sound,
The atmosphere cleaning device according to claim 1.
The stimulus output unit outputs a stimulus having an action of guiding the emotion of the person in a gentle direction,
The atmosphere cleaning device according to any one of claims 1 to 5.
The stimulus output unit outputs a natural sound composed of a sound generated by the movement of water or the atmosphere as an auditory stimulus for purifying the atmosphere,
The atmosphere cleaning device according to any one of claims 1 to 5.
The stimulus output unit outputs a music or melody sound as an auditory stimulus for purifying the atmosphere,
The atmosphere cleaning device according to any one of claims 1 to 5.
The stimulus output unit uses a sound effect or theme music that evokes a laugh as the music or the melody sound, or impresses a laugh.
The atmosphere cleaning device according to claim 8.
The stimulus output unit uses "Somebody Stole My Gal" or an imitation performance thereof performed by the American jazz musician Pee Wee Hunt as the musical composition, or "Somebody Story Sole" performed by the Pee Wee Hunt as the melody sound. The atmosphere cleaning device according to claim 8, wherein a melody sound reminiscent of "My Gal" is used.
The stimulus output unit outputs a dilution of essential oil or a dilution of aroma oil as an olfactory stimulus for purifying the atmosphere.
The atmosphere cleaning device according to any one of claims 1 to 5.
The stimulus output unit outputs an image as a visual stimulus that purifies the atmosphere.
The atmosphere cleaning device according to any one of claims 1 to 5.
The stimulus output unit uses, as the image, a natural landscape existing with the movement of water and the atmosphere,
The atmosphere cleaning device according to claim 12.
The stimulus output unit outputs illumination light as a visual stimulus that purifies the atmosphere.
The atmosphere cleaning device according to any one of claims 1 to 5.
The stimulus output unit uses, as the illumination light, illumination light with a color tone change that individually intensifies blue-based light, green-based light, and orange-based light,
The atmosphere cleaning device according to claim 14.
The stimulus output unit outputs an electromagnetic wave having a frequency of about 7.83 Hz based on Schumann resonance as an electromagnetic wave stimulus for cleaning the atmosphere.
The atmosphere cleaning device according to any one of claims 1 to 5.
The stimulus output unit outputs negatively charged water particles by the Leonard effect as charged particle stimuli that purify the atmosphere.
The atmosphere cleaning device according to any one of claims 1 to 5.
The stimulus output unit outputs a dilution of a malodorous component released by a plant or an animal as an olfactory stimulus for purifying the atmosphere, or a dilution of an organic sulfur compound,
The atmosphere cleaning device according to any one of claims 1 to 5.
An elevator comprising the atmosphere cleaning device according to any one of claims 1 to 18.
Get the sound,
A feature amount derived from the emotion of the person included in the sound, and identifies the emotion of the person based on the boundary information for identifying the emotion of the person from the feature amount,
Estimating the atmosphere around the source of the sound based on the identification result of the identified emotion of the person,
Outputting a stimulus for purifying the atmosphere in the vicinity of the source of the sound according to the degree of the atmosphere,
Atmosphere cleaning method.