JP2008299135A - Speech synthesis device, speech synthesis method and program for speech synthesis - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a speech synthesis device by which a user, having listened synthesized speech, easily understands a positional relation with an object associated with the synthesized speech, a state of the object and an intention of utterance. <P>SOLUTION: A synthesis parameter determination section 21 determines a speech synthesis parameter according to position relation indicated by at least any of a direction C, distance r or a relative angle, or their combination, or environment around the object 01. A speech synthesis section 22 synthesizes speech which is related to the object 01, according to a synthesized speech parameter which is determined by the synthesis parameter determination section 21. The synthesis parameter determination section 21 determines the speech synthesis parameter by, for example, preparing a parameter determination table in which a value of the synthesized speech parameter corresponding to a value of each input parameter is registered, based on a predetermined control plan, and by referring to the parameter determination table according to the value of each input parameter. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a speech synthesis technology, and more particularly to a speech synthesis technology for synthesizing speech associated with an object that does not have a physical speech uttering organ such as a robot or a game character.

  For example, in Patent Document 1, when the utterance object and the viewer are both facing the same direction, the voice output when the speech synthesis faces each other is suppressed by passing the output of speech synthesis through the LPF and suppressing the high range. More obscuring techniques are described.

  Patent Document 2 discloses a technique for changing sound attributes such as pitch, length, and strength depending on how much the current reading position is when speaking the contents of a web page. Is described.

  Patent Document 3 describes a technique for generating a three-dimensional sound by converting a distance and an azimuth into an elevation angle and a horizontal angle in order to accurately grasp the distance to the guidance object.

  In general, a stereophonic technique is known in which the position of an object such as a musical instrument is made to correspond to the position information of stereophonic sound.

JP 2006-109295 A JP 2006-171544 A JP 2005-333621 A

  However, since many of the synthesized voices often do not directly correspond to physical phenomena such as mouth movements, the user may not be able to associate the synthesized voice with the object associated with the synthesized voice.

  For example, when a robot having a speech synthesis function is moving on a plane, the robot utters synthesized speech, and a user who is a human being on the same plane can hear the voice. However, since the robot does not have a mouth or outputs a synthesized voice from a speaker having a sound radiation characteristic different from that of a human mouth, there is a problem that it is difficult for the user to grasp position information from the voice of the robot.

  Further, there is a problem that the intention of information transmission becomes unclear if the synthesized speech is heard in the same way even if the positional relationship between the robot and the user is different. For example, if you want to express joy because the robot and the user are facing each other, and say you want to express loneliness because the robot and the user are far away, what do you want the robot to say? It becomes ambiguous.

  Such a problem is not limited to a robot, but may also occur in a synthesized voice associated with an object (object) whose location is not always fixed, such as a toy, a mobile phone, and a personal computer. Note that the synthesized speech associated with a virtual space object such as a video game character is associated with an object having a physical entity as described above because the object does not have a physical entity. It is more unclear than synthesized speech.

  Therefore, the present invention provides a speech synthesis that enables a user who has heard a synthesized speech to generate a synthesized speech that can easily recall the positional relationship with the object associated with the synthesized speech, the status of the object, and the intention of the utterance. An object is to provide a device, a speech synthesis method, and a speech synthesis program.

  A speech synthesizer according to the present invention is a speech synthesizer that synthesizes speech associated with an object, and as speech associated with the object in accordance with the positional relationship between the user who is the object observing object and the object. A synthesis parameter determining unit that changes a speech synthesis parameter indicating what kind of synthesized speech is generated, and a synthesized speech as a speech associated with the object according to the speech synthesis parameter changed by the synthesis parameter determining unit And a speech synthesizer for generation.

  Further, the synthesis parameter determination unit changes the speech synthesis parameter according to the positional relationship between the user and the object indicated by an angle formed by a line connecting the user and the object and a predetermined direction reference determined according to a predetermined setting method. You may let them.

  Further, the composite parameter determination unit is configured to determine a position direction between the user and the object indicated by an angle between a line connecting the user and the object and a predetermined direction reference determined according to a predetermined setting method, and between the user and the object. The speech synthesis parameter may be changed according to the positional relationship between the user and the object indicated by any one of the distance, the relative angle between the front direction of the user and the front direction of the object, or a combination thereof.

  In addition, the speech synthesizer includes a synthesis parameter determination unit that changes a synthesis parameter indicating what kind of synthesized speech is generated as speech associated with the object according to an environment around the object, and the synthesis parameter A speech synthesis unit that generates synthesized speech as speech associated with the object according to the speech synthesis parameter changed by the determination unit may be provided.

  Also, the speech synthesizer generates any synthesized speech as speech associated with the object according to the positional relationship between the user who is the object observing object and the object and the environment surrounding the object. A synthesis parameter determination unit that changes a synthesis parameter that indicates whether to perform, and a voice synthesis unit that generates a synthesized speech as a voice associated with the object according to the voice synthesis parameter changed by the synthesis parameter determination unit May be.

  In addition, the synthesis parameter determination unit determines at least the brightness of the surroundings, the volume of the surrounding sound, whether there is another object in the vicinity, other objects that are in contact with or possessed by the object The speech synthesis parameter may be changed according to the environment around the object indicated by any of the above types.

  In addition, the synthesis parameter determination unit may change any of voice quality, speech speed, and voice volume as a voice synthesis parameter.

  Further, the synthesis parameter determination unit determines the detail level of the content spoken by the object according to the positional relationship between the user and the object, and the speech synthesis unit determines the detail level determined by the synthesis parameter determination unit. An utterance text editing unit that summarizes text in which the utterance content of the object is shown, and a text speech synthesis unit that generates synthesized speech using the text summarized by the utterance text editing unit as utterance content may be included.

  Further, the synthesis parameter determination unit determines the level of detail of the content spoken by the object according to the positional relationship between the user and the object, and the speech synthesis unit provides the level of detail determined by the synthesis parameter determination unit. An utterance text generation unit that generates a text in which a given item is expressed as utterance content, and a text-to-speech synthesis unit that generates a synthesized speech with the text generated by the utterance text generation unit as utterance content Good.

  Further, the synthesized parameter determination unit may change the synthesized speech parameter on the assumption that the synthesized speech is output from a position independent of the object.

  Further, the synthesized parameter determination unit may change the synthesized speech parameter on the assumption that synthesized speech is output from the vicinity of the user.

  Moreover, you may use the direction which the user is facing as a direction reference | standard.

  Moreover, you may use a user's gaze direction as a direction reference | standard.

  Further, the synthesis parameter determination unit may change the voice quality depending on whether or not the object is close to the front direction of the user.

  Further, the synthesis parameter determination unit may change the volume of the sound depending on whether or not the object is close to the front direction of the user.

  Further, the synthesized parameter determination unit may change the synthesized voice parameter according to the environment surrounding the object specified by the positional relationship expressed in the polar coordinate system by using the absolute direction as the direction reference.

  Further, the synthesized parameter determination unit may change the synthesized speech parameter according to the surrounding environment only when the positional relationship satisfies a predetermined condition.

  Further, the synthesized parameter determination unit may change the synthesized speech parameter more drastically when the object associated with the synthesized speech is a virtual object that exists only in the virtual space.

  Further, the synthesis parameter determination unit may change the level of detail depending on whether or not the object is close to the front direction of the user.

  Further, the synthesis parameter determination unit may change the level of detail and the speech rate according to whether or not the object is close to the front direction of the user.

  The speech synthesis method according to the present invention is a speech synthesis method for synthesizing speech associated with an object, and corresponds to the object according to a positional relationship between a user who observes the object and the object. A synthesized speech as speech associated with the object is generated according to the changed speech synthesis parameter according to the changed speech synthesis parameter. To do.

  In addition, the speech synthesis method includes a position direction of the user and the object indicated by an angle between a line connecting the user and the object and a predetermined direction reference determined according to a predetermined setting method, and a distance between the user and the object. The voice synthesis parameter may be changed according to the positional relationship between the user and the object indicated by any one or a combination of relative angles between the front direction of the user and the front direction of the object.

  In the speech synthesis method, a speech synthesis parameter may be changed according to an environment around the object, and a synthesized speech as speech associated with the object may be generated according to the changed speech synthesis parameter.

  Further, the speech synthesis method determines the level of detail of the content spoken by the object according to the positional relationship between the user and the object, and summarizes the text indicating the content of speech of the object according to the determined level of detail. Then, a synthesized speech having the summarized text as utterance content may be generated.

  Further, the speech synthesis method determines the level of detail of the content spoken by the object according to the positional relationship between the user and the object, and the text in which the given item is expressed as the speech content according to the determined level of detail. And a synthesized speech having the generated text as the utterance content may be generated.

  A speech synthesis program according to the present invention is a speech synthesis program for synthesizing speech associated with an object. The computer synthesizes a speech according to the positional relationship between the user who is the subject who observes the object and the object. As a speech associated with the object, a parameter determination process for changing a speech synthesis parameter indicating what kind of synthesized speech is generated as a speech associated with the object, and a speech associated with the object according to the changed speech synthesis parameter A synthesis process for generating synthesized speech is executed.

  In addition, the speech synthesis program causes the computer to determine the position direction of the user and the object indicated by an angle between a line connecting the user and the object and a predetermined direction reference determined according to a predetermined setting method in the parameter determination process. The speech synthesis parameters are changed according to the positional relationship between the user and the object indicated by the distance between the user and the object, the relative angle between the front direction of the user and the front direction of the object, or a combination thereof. Also good.

  Further, the speech synthesis program causes the computer to change a speech synthesis parameter indicating what kind of synthesized speech is generated as the speech associated with the object according to the environment surrounding the object, In accordance with the changed speech synthesis parameter, speech synthesis processing for generating synthesized speech as speech associated with the object may be executed.

  Further, the speech synthesis program causes the computer to determine the level of detail of the content spoken by the object in accordance with the positional relationship between the user and the object in the parameter determination process, and to determine the level of detail determined in the speech synthesis process. In accordance with the above, the text in which the utterance content of the object is shown may be summarized to generate a synthesized speech using the summarized text as the utterance content.

  Further, the speech synthesis program causes the computer to determine the level of detail of the content spoken by the object in accordance with the positional relationship between the user and the object in the parameter determination process, and to determine the level of detail determined in the speech synthesis process. Accordingly, a text in which a given item is expressed as utterance content may be generated, and a synthesized speech having the generated text as the utterance content may be generated.

  The first effect is that the synthesized parameter determination unit changes the parameters of the synthesized speech in accordance with the positional relationship between the user and the object or the surrounding environment, so that the object associated with the generated synthesized speech is accurately expressed. This makes it easier for the user to understand the connection between the synthesized speech and the object.

  The second effect is that the user can be presented with an easy-to-understand understanding of the positional relationship and situation with the object associated with the generated synthesized speech and the intention of information transmission by the synthesized speech.

  The third effect is that, even if the object associated with the synthesized speech is not directly in the user's field of view, the positional relationship with the object, the situation, It is possible to recall the intention of information transmission.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a configuration example of a speech synthesizer according to the first embodiment of the present invention. As shown in FIG. 1, the speech synthesizer according to the present embodiment includes a direction input unit 11, a distance input unit 12, a synthesis parameter determination unit 21, and a speech synthesis unit 22.

  Here, the positional relationship in the present embodiment is determined as follows. FIG. 2 is an explanatory diagram defining the positional relationship between the object 01 and the user 02 in the present embodiment. Here, the user 02 is a subject who observes the object 01. The object 01 is an object associated with the voice synthesized by the voice synthesizer as a generation source, and here, a robot having an entity is assumed. The objects having such entities include not only robots but also toys and mobile phones, and these are called entity objects. In the example illustrated in FIG. 2, the synthesized voice associated with the object 01 is output from the same position as the object 01 by the synthesized voice output unit 03 such as a speaker device included in the object 01. The voice synthesizer is realized, for example, as a voice output device for a real object, in a form incorporated in the object 01, or in a form incorporated in a speaker control device in a room displaying the real object. .

  As shown in FIG. 2, in this embodiment, the positional relationship between the object 01 and the user 02 is defined by a direction C and a distance r. The direction C is an angle formed by a direction reference C0 that is an arbitrary angle in a space defined in advance with respect to the user 02 and a direction in which the object 01 can be seen from the user 02. The distance r is information indicated by the distance between the user 02 and the object 01. FIG. 2 shows an example in which the direction that is the front of the user 02 is the direction reference 05.

  The direction input means 11 inputs the direction C. The distance input unit 12 inputs the distance r. The processing unit 13 inputs an ambient condition indicating an ambient environment as a condition to be considered when the object 01 emits synthesized speech.

  The composite parameter determination unit 21 associates the object 01 with the object 01 based on the direction C input by the direction input unit 11, the distance r input by the distance input unit 12, and the ambient condition input by the processing unit 13. A speech synthesis parameter for generating the synthesized speech is determined. Here, the speech synthesis parameter is a parameter indicating what kind of synthesized speech is generated. For example, it may be information directly indicating the value of an attribute included in the generated synthesized speech, or may be information indicating a characteristic categorized with respect to the attribute included in the synthesized speech.

  The speech synthesizer 22 synthesizes speech according to the speech synthesis parameter determined by the synthesis parameter determination unit 21.

  Next, the operation of the present embodiment will be described with reference to FIG. FIG. 3 is a flowchart showing an operation example of the speech synthesizer according to the present embodiment.

  First, the direction input unit 11, the distance input unit 12, and the ambient condition input unit 13 input the direction C, the distance r, and the ambient condition, respectively (step S101).

  The direction input unit 11 and the distance input unit 12 operate as follows, for example. By recognizing the positions of the object 01 and the user 02 from the image data captured by the camera and arranging them on the graph paper, the plane coordinates (x1, y1) of the object 01 and the plane coordinates (x2, y2) of the user 02 Get. Note that the coordinates may be specified using GPS or a gyroscope (gyro sensor) instead of shooting with the camera.

  At this time, the direction input means 11 may obtain the direction C by the following formula (1) using the angle C0 that the direction reference 05 makes with the x axis on the graph paper.

  C = arctan ((y1-y2) / (x1-x2))-C0 Formula (1)

  Moreover, the distance input means 12 should just obtain | require the distance r by the following formula | equation (2).

  Here, as the information indicating the positional relationship between the object 01 and the user 02, the direction C and the distance r based on the positional coordinates of both in the orthogonal coordinate system are used, but other coordinate systems and scales indicating the positional relationship are used. May be. Moreover, it can be similarly used not only on a plane but also at a higher order such as a solid angle. For example, as described later, a relative angle R between the front direction of the user 02 and the front direction of the object 01 may be used. The direction C indicates the position direction of the object 01 and the user 02 by the angle formed with the direction reference C0, while the relative angle R indicates the angle of orientation between the object 01 and the user 02. .

  In addition, the ambient condition input means 13 is in contact with or holds the object 01 as ambient conditions, ambient brightness, ambient sound power (sound volume), whether or not another object exists in the vicinity. Enter information indicating the type of other objects. As for the input of the ambient conditions, for example, the object 01 stores information indicating the possession status of an optical sensor such as a photodiode, a sound sensor such as a microphone, a contact sensor such as a microswitch or a pyroelectric element, a non-contact sensor, and other objects. It is realized by providing a holding status table and detecting inputs from these sensors or referring to the holding status table.

  Next, when the predetermined input parameters (here, the direction C, the distance r, and the ambient condition) are input by the direction input unit 11, the distance input unit 12, and the ambient condition input unit 13, the composite parameter determination unit 21 A speech synthesis parameter is determined according to the positional relationship between the object 01 and the user 02 indicated by the input parameter and the environment around the object 01 (step S102). Depending on the control policy of the synthesized speech parameter, only the direction C, the distance r, and only the ambient conditions may be input. Or the relative angle R by the front direction of the object 01 and the front direction of the user 02 may be input. Note that the synthesis parameter determination unit 21 performs speech synthesis according to the positional relationship between the object 01 and the user 02 indicated by at least one of the direction C, the distance r, the relative angle R, or a combination thereof, or the environment around the object 01. What is necessary is just to determine a parameter.

  The synthesis parameter determination unit 21 includes, for example, a parameter determination table in which values of synthesized speech parameters corresponding to the values of each input parameter are registered, and by referring to the parameter determination table according to the values of each input parameter, the synthesized speech is determined. The parameter may be determined.

  Then, the speech synthesizer 22 synthesizes and outputs the speech associated with the object 01 according to the synthesized speech parameter determined by the synthesis parameter determiner 21 (step S103).

  Next, a method for determining a synthesized speech parameter will be described with a specific example. 4 to 6 are explanatory diagrams showing examples of the parameter determination table. In these parameter determination tables, it is assumed that the closer the object 01 is to the front of the user 02, the more the user 02 is interested in the object 01, and control is performed so that the utterance content of the object 01 is clearly transmitted to the user 02. (Changed). On the other hand, assuming that the object 01 is less interested as it moves away from the front of the user 02, control is performed such that the consciousness of the user 02 is not disturbed by the utterance of the object 01. In addition, control is performed so that the atmosphere at the position where the object exists is transmitted in accordance with brightness, which is one of the ambient conditions of the object. A method of obtaining the direction C with reference to the front of the user 02 will be described later.

  FIG. 4 is an example of a parameter determination table for determining speech synthesis parameters indicating voice quality and speech rate based on direction C. In FIG. 4, for example, when the direction C (that is, the direction in which the object 01 is viewed from the user 02) when the front of the user 02 is the direction reference C0 is −45 to 45 degrees, the voice quality is “clear” ", And the speech rate is shown to be" normal ". For example, in the case of 45 to 90 degrees or −90 degrees to −45 degrees, it is indicated that the voice quality is “somewhat ambiguous” and the speech rate is “somewhat fast”. For example, in the case of 90 to 180 degrees or −180 to −90 degrees, it is indicated that the voice quality is “mogomogo (unclear)” and the speech rate is “fast”.

  FIG. 5 is an example of a parameter determination table for determining a speech synthesis parameter indicating power based on the distance r. In FIG. 5, for example, when the distance r between the object 01 and the user 02 is 0 to 50 cm, the power is “large”, when the distance r is 50 to 150 cm, the power is “medium”, and when the distance r exceeds 150 cm, the power is increased. Shown to be “small”.

  FIG. 6 is an example of a parameter determination table for determining a speech synthesis parameter indicating an utterance atmosphere based on ambient brightness. FIG. 6 shows, for example, a “joyful” atmosphere when the surrounding brightness is bright, and a “scary” atmosphere when the surrounding brightness is dark. The ambient condition can also be used as a limiting element according to the direction C and the distance r, for example, when the object 01 is far away from the user 02 and used to convey the state.

  For example, assuming that the direction reference C0, which is the line of sight of the user 02, is taken on the x axis of the coordinate axis, the position of the object is (x1, y1) = (20 cm, 40 cm), and the position of the user 02 is (x2, y2) = ( 0 cm, 0 cm), when the ambient conditions are observed to be bright, the direction C is calculated to be 60 degrees and the distance r = 44.7 cm. At this time, based on the parameter determination tables shown in FIGS. 4 to 6, the voice synthesis parameters are selected as “slightly ambiguous” voice quality, “slightly fast” speech speed, “small” power, “looks fun”. Is done. These speech synthesis parameters are sent to the speech synthesizer 22, and synthesized speech is generated by the corresponding parameters and output.

  As a result, it is assumed that the object 01 is present at a position of 60 ° diagonally and at a distance of 44.7 cm when viewed from the user 02, so that the user's interest is not as strong as near the front, and “somewhat ambiguous” and “somewhat fast” "Speech" with a power of "Small", but since the position of the robot is bright, a synthesized voice with characteristics according to the positional relationship between the user and the object is output so as to convey a "joyful atmosphere" The

  In this example, as an example of using the parameter determination table as shown in FIGS. 4 to 6 as means for obtaining the speech synthesis parameter from the direction C, the distance r, and the ambient conditions, the parameter determination table is used. It is not limited. In the above example, the input parameters described in each table are independent. However, a value obtained by combining the input parameters may be associated with a synthesized speech parameter. For example, it is possible to design such that the power is determined by a combination of the direction C and the distance r. Further, for example, it is possible to take a method of performing control according to the ambient conditions only when the distance to the user 02 is short or when the user is not in the field of view. Further, the parameters may be determined using the result of calculating the value of each input parameter regardless of the table reference.

  In addition, as an example of the ambient condition, the speech synthesis parameter is controlled according to the brightness. However, for example, the ambient sound power is input as the ambient condition, and the ambient sound power is associated with the ambient sound power. The sound power of the bass may be determined so as to be larger than the sound. Also, for example, as an ambient condition, whether there is another object in the vicinity of the object 01, whether there is another object in contact with or possessing the object 01, and if there is, the object's Input information (for example, type, weight, level, etc.), and make it a scary voice, a heavy voice, or a joyful voice according to the object information. Also good.

  Further, although voice quality, speech speed, power, and atmosphere are used as speech synthesis parameters, the present invention is not limited to these, and the speaker, pitch frequency, accent strength, and the like may be controlled.

  In the present embodiment, the speech synthesizer 22 may take forms such as recording / playback, codec such as CELP, text speech synthesis, etc. depending on the system to be applied. There are those that can change everything, and those that can only change the speech rate at most, such as a combination of recording and playback and speech rate conversion technology. Here, what can handle all parameters is used, but it is also possible to adopt a configuration in which parameters that cannot be handled are ignored.

  The synthesized speech parameter may be determined by a determination method as described below. FIG. 7 is an explanatory diagram showing an example of the position of the synthesized speech output unit 03 in this example. As shown in FIG. 7, in this example, the synthesized speech associated with the object 01 is emitted from another position regardless of the position of the object 01. Here, it is assumed that the data is output from sufficiently close to the user 02.

  In this example, since the synthesized speech is output from sufficiently close to the user 02, when there are a plurality of objects 01, the position where the synthesized speech of each object is output is sufficiently close or completely the same. Become.

  FIG. 8 is an explanatory diagram showing an example of a parameter determination table in this example. FIG. 8 shows an example of a parameter determination table for determining power based on the distance r. As illustrated in FIG. 8, the power of the synthesized speech may be increased when the object 01 is close as viewed from the user 02, and the synthesized speech power may be decreased when the object 01 is far away.

  As a result, the synthesized speech power can be increased when the object is close to the user, and the synthesized speech power can be decreased when the object is far away, thereby obtaining a more realistic synthesized speech. FIG. 9 is an explanatory diagram showing an example of the positional relationship between the object 01 and the user 02. As shown in FIG. 9, even when the object 01 exists outside the field of view of the user 02, it is possible to estimate a rough position from the characteristics of the corresponding synthesized speech.

  Further, instead of using the parameter determination table based on the distance r as shown in FIG. 8, it is also possible to obtain the power p (r) by the following equation (3) using the input distance r. Here, K is a negative constant.

p (r) = r ^K ... Formula (3)

  Here, by setting K = − (2/3), it is possible to obtain an effect in accordance with the sense.

  In addition, as a parameter determination method based on the distance r, the speech synthesis parameters are set according to the parameter determination table as shown in FIG. You may decide. FIG. 10 is an explanatory diagram showing another example of a parameter determination table for determining power based on the distance r. As shown in FIG. 10, it is also possible to perform control such that the power is increased so that information is transmitted at a distant place, and the power is reduced so as not to feel an agility at a close place.

  Here, only the example of controlling the power in correspondence with the distance r is shown, but the voice quality and the atmosphere may be controlled by combining the direction C and the ambient conditions as described above.

  Next, an example of setting the direction reference C0 is shown. First, an example in which the direction reference C0 is defined as an absolute direction is shown. FIG. 11 is an explanatory diagram showing the direction reference C0 in this example. As shown in FIG. 11, the direction reference C0 may be defined as an absolute direction in a room where the user 02 and the object 01 exist. In the example shown in FIG. 11, the direction parallel to the long side of the room from the position of the user 02 is defined as the direction reference C0.

  Accordingly, the direction C can be easily obtained by measuring the angle formed by the long side of the room and the line segment connecting the user 02 and the object 01. Also, by defining the direction reference C0 as an absolute direction, the positional relationship between the object 01 and the user 02 can be expressed in a polar coordinate system with the user position as the origin. Based on the positional relationship shown in the polar coordinate system, for example, when the object 01 is on the north side of the user 02, the expression is cold, or, for example, the place where the object 01 is located compared to the user 02 By performing control such as creating a scared atmosphere when the user is closer to the object, it is possible to convey the situation around the object and the intention of information transmission more clearly. This example is also an example in which the speech synthesis parameters are controlled according to the environment around the object 01 specified by the positional relationship between the user 02 and the object 01.

  Further, the direction reference C0 may be determined as follows. In this example, the direction of the eyes of the user 02 is added to the positional relationship between the object 01 and the user 02. FIG. 12 is an explanatory diagram showing a configuration example of the speech synthesizer and the direction reference C0 in this example. As shown in FIG. 12, the speech synthesizer may be communicably connected to the user visual field detection means 16, the user position detection means 15, and the object position detection means 14. In FIG. 12, it is assumed that the speech synthesizer is included in the object 01.

  The object position detection unit 14 is attached to the object 01 and detects the position of the object 01. The user position detection unit 15 is attached to the user 02 and detects the position of the user 02. The object position detection unit 14 and the user position detection unit 15 are realized by a GPS receiver, for example.

  Moreover, the user visual field detection means 16 is comprised from a camera apparatus and an image analysis means, for example, and detects the gaze direction of the user 01. Specifically, the camera device attached to the glasses worn by the user 02 captures the black eye position of the user 02, and the image analysis unit analyzes the image data to obtain a deviation from the reference value of the black eye position. Thus, the user's line-of-sight direction may be detected. Note that the direction in which the user's line of sight is directed may be based on, for example, the front direction of the face measured using a gyro sensor attached to the glasses. In addition, for example, the image analysis unit analyzes image data captured by a fixed camera device capable of capturing the user 02 from at least two directions, and detects the feature indicating the front of the face and the position of the black eye from the image data. The front direction of the face and the gaze direction with respect to the front direction of the face may be specified, and the actual gaze direction may be calculated from these directions and the position of the camera device that captured the image.

  The direction input unit 11 and the distance input unit 12 receive information indicating the object position detected by these detection units, information indicating the user position, and information indicating the user's line-of-sight direction, for example, via an input port or a communication network. Thus, the direction C and the distance r may be calculated. At this time, the direction input unit 11 defines the line-of-sight direction detected by the user visual field detection unit 16 as the direction reference C0. In such a case, the direction reference C0 depends on the direction in which the user is facing.

  Thereby, it is also possible to change the characteristics of the synthesized speech in response to the fact that the user 02 moves around and the position and line-of-sight direction change every moment. The user can change how information is received by changing his / her orientation and position as necessary.

  In addition to the line-of-sight direction, a more stable direction may be used by detecting a direction in which the face is facing (front direction), a direction in which the body is facing (front direction), and the like. As for the front direction of the face and the front direction of the body, for example, a wearing object (for example, glasses) that is worn corresponding to a part where the front of the face or the front of the body can be seen, such as the user's hands, both ears, and both eyes. Measurement using a gyro sensor attached to a microphone or an earphone microphone), or a feature indicating the face or the front of the body may be detected from image data from two directions and specified.

  FIG. 13 is an explanatory diagram illustrating an example of a parameter determination table in the present example. FIG. 13 shows an example of a parameter determination table for determining voice quality based on the direction C with the front direction of the user 02 as the direction reference C0. This example is based on the assumption that the distance r indicates that the distance between the object 01 and the user 02 is sufficiently close. As shown in FIG. 13, when the object 01 is present near the front as viewed from the user 02, that is, when the angle from the front is near 0 degrees, the voice quality is “clear”, and conversely, the side direction and the rear direction If it exists, the voice quality may be controlled to be a “whispering voice”.

  As a result, it is possible to expect the effect of attracting the user's attention and having the user 02 face the front by uttering the user whispering in the vicinity of the front. Further, an intermediate stage between clearness and whispering may be prepared. A similar effect may also be realized by changing the power as shown in FIG. 14 so that the consciousness is directed with a large amount of power in the side and rear directions. FIG. 14 shows an example of a parameter determination table for determining power based on the direction C with the front direction of the user 02 as the direction reference C0.

  When the object 01 has a front direction as well as the front direction of the user 02, the relative angle between the object 01 and the user 02 depends on the direction of each other (line of sight, face, front of the body). R may be added. When the object 01 does not manage the front direction of the object 01, the front direction of the object 01 may be detected using a method similar to the method of detecting the front direction of the user 02.

  Then, the direction input unit 11 sends information indicating the object position, information indicating the user position, information indicating the front direction of the user 02, and information indicating the front direction of the object 01 via an input port or a communication network. The direction C and the relative angle R may be obtained.

  In such a case, for example, the synthesis parameter determination unit 21 regards the object 01 as moving if it is close to the back of the user 02 and is directed in the same direction if the distance is close. The speech synthesis parameters may be controlled so as to have a pleasant atmosphere. Further, for example, if the distances are close to each other even if the distance is short, the atmosphere may be controlled so as to be indifferent. Further, it is possible to perform the above whisper control only when the object is facing the user. In this way, by controlling the synthesized speech parameters based on the positional relationship indicated by the direction C, the distance r, and the relative angle R, it is possible to utter with more realistic feeling.

  FIG. 15 is an explanatory diagram showing an example of a parameter determination table in this example. FIG. 15 shows an example of a parameter determination table for determining the pitch frequency height, inflection strength, and atmosphere (pattern) based on the direction C, the distance r, and the relative angle R. 15A is an example of a parameter determination table that is referred to when the distance r is smaller than 100 cm, and FIG. 15B is an example of a parameter determination table that is referred to when the distance r is 100 cm or more. However, the determination of the parameters combining the three elements is not limited to registering the table divided into two in this way. In FIG. 15, for example, the distance r is a short distance (smaller than 100 cm), the relative direction R is the front direction (0 to +90 degrees or 0 to −90 degrees), and the direction C is the front direction (0 to +45). In the case of (degree or 0 to -45 degrees), an example is shown in which the pitch frequency is increased and decreased, the inflection is strong, and the atmosphere is controlled so as to have a very pleasant atmosphere. Further, for example, the distance r is a long distance (100 cm or more), the relative direction R is the opposite direction (+90 to +180 degrees or −90 to −180 degrees) from the directly facing direction, and the direction C is the back direction (+90). ~ + 180 degrees or -90 to -180 degrees), an example is shown in which the pitch frequency is moderately high, the inflection is weak, and the atmosphere is steep (distant feeling). Has been.

  In the above example, the case where the object 01 is a real object has been described as an example. However, the present invention is applied even if the object 01 is a virtual object that exists only in a virtual space such as a video game character. It is possible. However, the object 01 has a concept of position on the virtual section and position information.

  In such a case, the virtual position of the user 02 and the direction reference C0 are set in the same virtual space as the object 01, and the direction C and the distance r between the object 01 and the user 02 in the virtual space can be obtained. That's fine. The direction reference C0 may be set to, for example, a certain absolute direction, the direction of the face of the user 02, or the traveling direction. Further, for example, it is also possible to obtain the relative angle R according to the front direction of each other, with the traveling direction of the object 01 or the user 02 as the front direction.

  The position of the object 01 and the position of the user 02 may or may not be presented to the user by a method such as projecting onto a video display. Further, the virtual position of the user 02 may be moved according to a button or mouse operation on the controller by the user 02. Note that the parameter determination method in this example is controlled so that, for example, the closer the object 01 is to the front of the user 02, the more clearly the utterance content of the object 01 is transmitted to the user 02, as in the method already shown. Alternatively, it may be controlled so that the atmosphere of the position where the object exists is transmitted in accordance with the brightness which is one of the ambient conditions of the object. Further, for example, in order to get a sense of distance, the sound power may be controlled to be large when the object 01 is close when viewed from the user 02 and to be small when the object 01 is far away. Further, the sound power may be controlled so that the sound power is increased when the object 01 is far from the user 02 and the sound power is decreased when the object 01 is close so that the utterance content can be sufficiently heard. Further, for example, based on the positional relationship shown in the polar coordinate system, the sound power may be controlled for purposes such as making it cold or using a fixedly located environment. Further, for example, when the object 01 is present in the vicinity other than the front, the voice quality may be controlled so as to be a whisper, or the sound power may be controlled so as to attract attention. Further, for example, based on the positional relationship in consideration of the relative angle R, it may be controlled to have a pleasant atmosphere or an indifferent atmosphere.

  In addition, when controlling the synthesized speech parameter of the synthesized speech associated with the virtual object, the parameter change range may be increased to change it more drastically.

  Note that the speech synthesizer in this example is realized by, for example, a CPU or the like that reads a program describing a speech synthesis function as one function of a program for operating a virtual object and operates according to the program.

  This makes it possible for the user to easily estimate the position in the virtual space and the intention of state transmission based on the characteristics (voice quality and power) of the synthesized speech regardless of whether or not the position is presented.

  As described above, according to the present embodiment, based on the positional relationship between the object 01 and the user 02 indicated by at least one of the direction C, the distance r, and the relative angle R, or a combination thereof, and the surrounding environment, By determining a synthesized speech parameter according to a predetermined determination method based on a predetermined control policy, an object associated with the synthesized speech generated depending on the control policy can be expressed more accurately. In addition, it is possible to present the positional relationship and situation with the object associated with the generated synthesized speech and the intention of information transmission by the synthesized speech in an easy-to-understand manner, and even if the object associated with the synthesized speech is Even if it is not directly in the user's field of view, the user can see that object as well as in the field of view. Positional relationship and circumstances, it is possible to further recall the intentions of the information transmission by synthesized speech.

Embodiment 2. FIG.
Next, a second embodiment of the present invention will be described. FIG. 16 is a block diagram illustrating a configuration example of a speech synthesizer according to the second embodiment. As shown in FIG. 16, in the speech synthesizer according to the present embodiment, the speech synthesizer 22 includes an utterance text editing unit 221 and a text speech synthesizer 222. In the speech synthesizer in the present embodiment, the synthesis parameter determination unit 21 determines at least the level of detail of the utterance content as the speech synthesis parameter.

  The utterance text editing unit 221 edits the input text (text indicating the utterance content) based on the level of detail determined by the synthesis parameter determination unit 21. Specifically, the utterance contents are summarized according to the level of detail determined by the synthesis parameter determination unit 21.

  The text-to-speech synthesis unit 222 generates synthesized speech with the content edited by the utterance text editing unit 221.

  Next, the operation of the present embodiment will be described with reference to FIG. FIG. 17 is a flowchart showing an operation example of the speech synthesizer according to the present embodiment. The operation in step S101 in FIG. 17 is the same as that in step S101 in the first embodiment shown in FIG.

  The composite parameter determination unit 21 receives predetermined input parameters (for example, direction C, distance r, relative angle R, ambient condition, etc.) from the direction input unit 11, the distance input unit 12, and the ambient condition input unit 13. Based on these input parameters, a synthesized speech parameter indicating at least the level of detail of the utterance content is determined (step S201).

  The composite parameter determination unit 21 determines, for example, a detail parameter that takes three values “all input”, “summarize only a little”, and “summarize almost” by referring to a parameter determination table as shown in FIG. May be. FIG. 18 is an example of a parameter determination table for determining a speech synthesis parameter indicating the degree of detail based on the direction C. In the parameter determination table shown in FIG. 18, when the object 01 is in front of the user 02, it is assumed that the user 02 is interested in the object 01 and the detailed information is transmitted. If it is facing, it is considered that there is little interest and it is controlled so that only the outline is transmitted.

  Next, the utterance text editing unit 221 edits (summarizes) the input text according to the detail level parameter (step S202). The summary of the text is, for example, as disclosed in a non-patent document “Inderjeet Mani (translated by Manabu Okumura),“ Automatic Translation ”, Kyoritsu Shuppan, June 2003, p. 14-21”. The analysis may be performed by generating an internal representation, transforming the internal representation into a summary representation according to the detail parameter, and returning the summary representation to a natural language.

  Then, the text-to-speech synthesizer 222 synthesizes and outputs the speech associated with the object 01 with the contents edited by the utterance text editing unit 221 (step S203).

  In the above example, the utterance text editing unit 221 for editing the text indicating the given utterance content is used. However, the present invention is not limited to this, and the utterance content is generated from the details and the utterance content as inputs. May be. For example, as the utterance content, only the route from the house and the time it takes are “bus / 15 minutes + train / 1 hour + waiting / 15 minutes + train / 30 minutes + bus / 30 minutes” and only an outline. When the level of detail for explanation is given, an utterance content “I used a bus and a train from home” may be generated. It should be noted that, along with the level of detail, a politeness such as “mass tone”, “a tone that is”, and “a tone” may be given, and the utterance content may be generated in a tone indicated by the politeness.

  Also, by giving a rough time length for the entire utterance, for example, when performing art collection collection guidance by voice synthesis, the picture in the direction that the user is facing will be explained over a long time, and the direction that is not suitable The picture can be used to finish explanation in a short time.

  In addition, not only text playback but also speech synthesis by recording and playback, only a part of the utterance target is skipped and played, or a headline etc. when one utterance target consists of multiple partial voices By selectively reproducing only the important partial voices, it is possible to present the intention of information transmission by synthetic voices in an easy-to-understand manner.

  In addition to controlling the degree of detail according to the direction C, for example, the same effect can be obtained by controlling the degree of detail according to the distance r as shown in FIG.

  Further, a value obtained by combining the input parameters may be associated with a synthesized speech parameter (detail level) without making the input parameters described in each table independent. For example, it is possible to design so that the degree of detail is determined by a combination of the direction C and the distance r. Further, parameters (for example, utterance time) may be determined using the result of calculating the value of each input parameter without referring to the table.

  FIG. 20 is an explanatory diagram showing an example of a parameter determination table in this example. FIG. 20 is an example of a parameter determination table for determining a speech synthesis parameter indicating the degree of detail based on the distance r and the direction C. In FIG. 20, for example, even if the distance r is a short distance (0 to 50 cm), if the direction C is the front direction (0 to +45 degrees or 0 to −45 degrees), it should be briefly and briefly spoken. An example is shown in which control is performed and, conversely, when the direction C is the back direction (+90 to +180 degrees or −90 to −180 degrees), control is performed so as to speak in detail. This is an example in which, even if a person who seems to be less interested is in the vicinity, control is performed so that the user can get into the ear for a long time to attract the person's interest.

  FIG. 21 is a flowchart showing another operation example of the speech synthesizer according to this embodiment. As shown in FIG. 21, the text-to-speech synthesizer 222 is the content edited (or generated) by the utterance text editor 221 and according to the speech synthesis parameters such as voice quality and power determined by the synthesis parameter determination unit 21. You may make it synthesize | combine the audio | voice of the object 01 (step S301). Other operations are the same as the operations shown in FIGS. 3 and 17.

  As described above, according to the present embodiment, by determining the level of detail of the utterance content based on the positional relationship between the object 01 and the user 02 and the surrounding conditions, the detailed level according to the level of interest of the user 02 is obtained. The contents can be output (spoken or reproduced), or only the summary can be output, and the intention of information transmission by synthesized speech can be recalled more.

  The present invention can be applied to a voice conversation in a robot or a toy, for example. Further, the present invention can be applied to a traffic system in which the position and situation change from time to time, a video game character in which an object does not have a substance, a virtual reality system, and the like. In particular, the present invention is suitable for application to these systems using text-to-speech synthesis, which generally has a high degree of freedom in setting parameters for speech synthesis.

It is a block diagram which shows the structural example of the speech synthesizer by 1st Embodiment. It is explanatory drawing which defined the positional relationship of the object 01 and the user 02. FIG. It is a flowchart which shows the operation example of the speech synthesizer by 1st Embodiment. It is explanatory drawing which shows the example of a parameter determination table. It is explanatory drawing which shows the example of a parameter determination table. It is explanatory drawing which shows the example of a parameter determination table. It is explanatory drawing which shows the example of the position of the synthetic | combination audio | voice output part 03. FIG. It is explanatory drawing which shows the example of a parameter determination table. It is explanatory drawing which shows the example of the positional relationship of the object 01 and the user 02. FIG. It is explanatory drawing which shows the example of a parameter determination table. It is explanatory drawing which shows the direction reference | standard C0. It is explanatory drawing which shows the structural example and direction reference | standard C0 of a speech synthesizer. It is explanatory drawing which shows the example of a parameter determination table. It is explanatory drawing which shows the example of a parameter determination table. It is explanatory drawing which shows the example of a parameter determination table. It is a block diagram which shows the structural example of the speech synthesizer by 2nd Embodiment. It is a flowchart which shows the operation example of the speech synthesizer by 2nd Embodiment. It is explanatory drawing which shows the example of a parameter determination table. It is explanatory drawing which shows the example of a parameter determination table. It is explanatory drawing which shows the example of a parameter determination table. It is a flowchart which shows the other operation example of the speech synthesizer by 2nd Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Direction input means 12 Distance input means 13 Ambient condition input means 14 Object position detection means 15 User position detection means 16 User visual field detection means 21 Synthesis parameter determination part 22 Speech synthesis part 221 Utterance text edit part 222 Text speech synthesis part

Claims (30)

A speech synthesizer that synthesizes speech associated with an object,
A synthesis parameter determination unit that changes a speech synthesis parameter indicating what kind of synthesized speech is generated as speech associated with the object according to a positional relationship between the user who is an object observing object and the object;
A speech synthesizer, comprising: a speech synthesizer that generates synthesized speech as speech associated with the object according to the speech synthesis parameter changed by the synthesis parameter determination unit. The synthesis parameter determination unit changes the speech synthesis parameter according to a positional relationship between the user and the object indicated by an angle formed by a line connecting the user and the object and a predetermined direction reference determined according to a predetermined setting method. Item 2. The speech synthesizer according to Item 1. The composite parameter determination unit is configured to determine a position direction of the user and the object indicated by an angle between a line connecting the user and the object and a predetermined direction reference determined according to a predetermined setting method, a distance between the user and the object, The speech synthesis parameter is changed according to a positional relationship between the user and the object indicated by any one or a combination of relative angles between the front direction of the user and the front direction of the object. Synthesizer. A speech synthesizer that synthesizes speech associated with an object,
A synthesis parameter determination unit that changes a synthesis parameter indicating what kind of synthesized speech is generated as speech associated with the object according to an environment around the object;
A speech synthesizer, comprising: a speech synthesizer that generates synthesized speech as speech associated with the object according to the speech synthesis parameter changed by the synthesis parameter determination unit. A speech synthesizer that synthesizes speech associated with an object,
A synthesis parameter indicating what kind of synthesized speech is generated as speech associated with the object according to the positional relationship between the user who is the object observing object and the object and the environment around the object A synthesis parameter determination unit to be changed;
A speech synthesizer, comprising: a speech synthesizer that generates synthesized speech as speech associated with the object according to the speech synthesis parameter changed by the synthesis parameter determination unit. The synthesis parameter determination unit determines at least the brightness of the surroundings, the volume of the surrounding sound, whether or not other objects exist in the vicinity, and the types of other objects that are in contact with or possessed by the objects. The speech synthesis apparatus according to claim 4, wherein speech synthesis parameters are changed according to an environment around the object indicated by any of the above. The speech synthesis device according to any one of claims 1 to 5, wherein the synthesis parameter determination unit changes any of voice quality, speech rate, and voice volume as a speech synthesis parameter. The synthesis parameter determination unit determines the level of detail of the content spoken by the object according to the positional relationship between the user and the object,
The speech synthesizer
An utterance text editing section that summarizes the text indicating the utterance content of the object according to the level of detail determined by the synthesis parameter determination section;
The speech synthesizer according to any one of claims 1 to 3, further comprising: a text speech synthesizer that generates synthesized speech having the text summarized by the utterance text editing unit as utterance content. The synthesis parameter determination unit determines the level of detail of the content spoken by the object according to the positional relationship between the user and the object,
The speech synthesizer
According to the level of detail determined by the synthesis parameter determination unit, an utterance text generation unit that generates text in which a given item is expressed as utterance content;
The speech synthesizer according to any one of claims 1 to 3, further comprising: a text speech synthesizer that generates a synthesized speech whose utterance content is the text generated by the uttered text generator. The speech synthesis apparatus according to any one of claims 1 to 4, wherein the synthesis parameter determination unit changes the synthesis speech parameter on the assumption that the synthesized speech is output from a position independent of the object. The speech synthesis apparatus according to claim 10, wherein the synthesis parameter determination unit changes the synthesis speech parameter on the assumption that synthesized speech is output from the vicinity of the user. The speech synthesis apparatus according to claim 2, wherein a direction in which the user is facing is used as the direction reference. The speech synthesizer according to claim 2 or 3, wherein a user's line-of-sight direction is used as the direction reference. The speech synthesis apparatus according to claim 3, wherein the synthesis parameter determination unit changes the voice quality according to whether or not the object is close to the front direction of the user. The speech synthesis apparatus according to claim 3, wherein the synthesis parameter determination unit changes the sound volume according to whether or not the object is close to the front direction of the user. The speech synthesis according to claim 3, wherein the synthesis parameter determination unit changes the synthesized speech parameter according to the environment surrounding the object specified by the positional relationship expressed in the polar coordinate system by using the absolute direction as the direction reference. apparatus. The speech synthesis apparatus according to claim 5, wherein the synthesis parameter determination unit changes the synthesized speech parameter according to the surrounding environment only when the positional relationship satisfies a predetermined condition. The speech synthesis apparatus according to claim 1, wherein the synthesis parameter determination unit changes the synthesis speech parameter more drastically when the object associated with the synthesized speech is a virtual object that exists only in the virtual space. The speech synthesis device according to claim 8 or 9, wherein the synthesis parameter determination unit changes the degree of detail according to whether or not the object is close to the front direction of the user. The speech synthesis device according to claim 8 or 9, wherein the synthesis parameter determination unit changes the level of detail and the speech rate according to whether or not the object is close to the front direction of the user. A speech synthesis method for synthesizing speech associated with an object,
According to the positional relationship between the user who observes the object and the object, the speech synthesis parameter indicating what kind of synthesized speech is generated as the speech associated with the object is changed,
A synthesized speech as a speech associated with the object is generated according to the changed speech synthesis parameter. Position direction of the user and the object indicated by an angle between a line connecting the user and the object and a predetermined direction reference determined in accordance with a predetermined setting method, a distance between the user and the object, a front direction of the user and the object The speech synthesis method according to claim 21, wherein the speech synthesis parameter is changed in accordance with a positional relationship between the user and the object indicated by any one or a combination of relative angles depending on the front direction. A speech synthesis method for synthesizing speech associated with an object,
Depending on the environment around the object, change the speech synthesis parameters,
A synthesized speech as a speech associated with the object is generated according to the changed speech synthesis parameter. According to the positional relationship between the user and the object, determine the level of detail of the content spoken by the object,
Summarizing text with the utterance content of the object according to the determined level of detail,
The speech synthesis method according to any one of claims 21 to 23, wherein a synthesized speech having the summarized text as an utterance content is generated. According to the positional relationship between the user and the object, determine the level of detail of the content spoken by the object,
According to the determined level of detail, generate a text in which a given item is expressed as utterance content,
The speech synthesis method according to any one of claims 21 to 23, wherein a synthesized speech having the generated text as an utterance content is generated. A speech synthesis program for synthesizing speech associated with an object,
On the computer,
A parameter determination process for changing a speech synthesis parameter indicating what kind of synthesized speech is generated as speech associated with the object according to a positional relationship between the user who is an object observing object and the object, and change A speech synthesis program for executing a synthesis process for generating synthesized speech as speech associated with the object according to the speech synthesis parameters that have been made. On the computer,
In the parameter determination process, the position and direction of the user and the object indicated by the angle between the line connecting the user and the object and a predetermined direction reference determined according to a predetermined setting method, the distance between the user and the object, the user 27. The speech synthesis program according to claim 26, wherein the speech synthesis parameter is changed according to a positional relationship between the user and the object indicated by any one or a combination of relative angles between the front direction of the object and the front direction of the object. A speech synthesis program for synthesizing speech associated with an object,
On the computer,
In accordance with the environment surrounding the object, according to a parameter determination process for changing a speech synthesis parameter indicating what kind of synthesized speech is generated as speech associated with the object, and according to the changed speech synthesis parameter, A speech synthesis program for executing speech synthesis processing for generating synthesized speech as speech associated with an object. On the computer,
In the parameter determination process, according to the positional relationship between the user and the object, the detail level of the content spoken by the object is determined,
29. The speech synthesis process, according to the determined level of detail, summarizes the text indicating the utterance content of the object, and generates synthesized speech using the summarized text as the utterance content. The speech synthesis method according to any one of the above. On the computer,
In the parameter determination process, according to the positional relationship between the user and the object, the detail level of the content spoken by the object is determined,
27. The speech synthesis process generates a text in which a given item is expressed as utterance content according to the determined level of detail, and generates a synthesized speech using the generated text as the utterance content. 28. The speech synthesis method according to any one of 28.

Images