JP2003099078A - Method and device for reproducing synthesized voice - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method and a device for reproducing synthesized voice, which can transmit various additional information other than a synthesized voice to a user. SOLUTION: The device has a text analysis part 2 which extracts as sound source position definition information text information to be synthesized as a voice by analyzing a text and information needed to define the sound source position of a synthesized voice in a virtual sound space, a voice synthesis part 4 which synthesizes a voice of the text information extracted by the text analysis part 2, a sound source position definition part 6 which defines a virtual sound source position in the space with the sound source position definition information, a sound image location processing part 7 which locate the synthesized voice obtained through the voice synthesis processing of the voice synthesis part 4 at the virtual sound source position in the space defined by the sound source position definition part 6, and a stereoscopic image output part 8 which outputs the synthesized voice from the position located by the sound image location processing part 7.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a synthetic speech reproducing method and a synthetic speech reproducing apparatus for generating a speech signal from text and outputting it as synthetic speech.

[0002]

2. Description of the Related Art A voice synthesizing technique for generating a voice signal from a text and outputting it as a synthetic voice is widely used in the field of information technology for handling voice.

Generally, in the conventional speech synthesis, since the synthesized speech output is almost one-channel monaural,
For example, if the text to be speech-synthesized is a text that records a dialogue, a meeting, or the like of multiple speakers, it is difficult to identify the speaker with the synthesized speech obtained by performing the speech-synthesis processing on the text. There is a problem that it is difficult to grasp the situation on the spot such as the positional relationship of the.

Further, in the case where a plurality of types of text exist as texts to be voice-synthesized and the plurality of types of texts are subjected to voice-synthesizing processing and the voice-synthesis results are output in a list, one text is used. Even if there is a large amount of text, the text-to-speech processing is performed, and even if the results of the text-to-speech are output in a list, what text the synthesized speech is output just by listening to the output synthetic speech, Alternatively, there is a problem that it is difficult to know which part of the text the synthesized voice is output.

This is because, as described above, the conventional speech synthesis technology simply performs speech synthesis processing on text and outputs the synthesized speech as the synthesis result as monaural speech, so that data that can be provided to the user is not available. This is because it is only the synthetic voice for the text and there is no additional information other than that.

As one means for solving this problem, it is conceivable to output synthetic speech as stereophonic speech of a plurality of channels. Conventionally, there is a technique for generating multi-channel stereophonic sound. For example, JP-A-8-33169
The “sound information conversion device” (hereinafter, referred to as the first conventional technology) of 7 sets a stereo signal to have an arbitrary frequency band characteristic to give a sense of expansiveness of the reproduced sound of the original stereo signal, thereby increasing the high frequency range. This is to generate a more natural sound field by artificially adding frequency components and converting timbre and sound quality.

Further, in Japanese Patent Laid-Open No. 9-90963, "Sound Information Providing Device and Sound Information Selecting Method" (hereinafter referred to as "second prior art"), a plurality of sound information is arranged in a virtual sound space,
The sound information selecting means is provided so that the user can select the plurality of sound information.

[0008]

As described above, the first prior art described above is designed to give a sense of expansiveness of the reproduced sound of the original stereo signal. In the first prior art, It is not possible to solve the above-mentioned problems with synthetic speech.

Further, in the second conventional technique, as described above, a plurality of sound information are arranged in a virtual sound space, and sound information selecting means is provided so that the user can select the plurality of sound information. However, the second conventional technique cannot solve the above-mentioned problem with synthesized speech.

Therefore, according to the present invention, by adding position information in space to the synthesized voice and reproducing the synthesized voice, which portion of the text is output as the synthesized voice, and when a plurality of texts exist, which of the texts is output. It is an object of the present invention to provide a synthetic voice reproduction method and a synthetic voice reproduction device capable of transmitting various additional information other than synthetic voice to a user such as whether synthetic voice is output.

[0011]

In order to achieve the above-mentioned object, a synthetic speech reproduction method of the present invention analyzes a text input as a speech synthesis target, extracts a text to be speech-synthesized, and synthesizes it. Extracting the information necessary for defining the sound source position of the voice on the virtual sound space as sound source position definition information, and defining the sound source position in the virtual sound space using this sound source position definition information, Speech synthesis processing is performed on the text to be speech-synthesized, the synthesized speech obtained thereby is localized at a virtual sound source position in the defined sound space, and the synthesized speech is generated from the localized position. Is output.

In this synthetic speech reproduction method, when there are a plurality of content groups in the text to be speech-synthesized, a virtual sound source position in the sound space is defined for each content group, and each of the content groups is defined. The synthesized voice is output from the sound source position for a group of contents associated with the sound source position.

When defining the sound source position for each group of contents, if the group of contents exists in the order along the flow of the text sentence, the sound source position for each group of the content is the flow of the sentence. The positions are defined so that they change in the order along.

In the synthesized voice reproduction method, when the voice synthesis target text is a plurality of independent texts, a virtual sound source position in the sound space is defined for each text, The synthesized voice for the text associated with the sound source position is output from the sound source position.

When a plurality of independent texts can be classified by a certain standard, the virtual sound source position in the sound space is defined for each classification, and the sound source position is calculated from each sound source position. The synthesized voice for the text belonging to the category associated with is output.

In the synthesized voice reproduction method, when the text to be voice synthesized is text existing in a plurality of applications in an information processing system having a multitask function, a virtual sound source in the sound space. The position is defined for each application, and the synthesized sound is output from each sound source position for the text existing in the application associated with the sound source position.

Further, the synthesized speech reproducing apparatus of the present invention analyzes the text to be synthesized, extracts the text information to be synthesized, and defines the sound source position of the synthesized speech in the virtual sound space. A text analysis means for extracting information necessary for sound source position definition information, and a sound source position definition means for defining a virtual sound source position in a sound space using the sound source position definition information extracted by the text analysis means, The sound source position defining means defines a voice synthesizing means for performing a voice synthesizing process on the text information to be subjected to the voice synthesizing extracted by the text analyzing means, and a synthesized voice obtained by the voice synthesizing processing by the voice synthesizing means. Sound image localization means for locating a virtual sound source position in a sound space, and the synthesized voice from the sound source position localized by the sound image localization means. It has a configuration having a three-dimensional sound image output means for force.

In this synthesized speech reproducing apparatus, when a plurality of content groups exist in the text to be synthesized, the virtual sound source position in the sound space is defined for each content group, and The synthesized voice is output from the sound source position corresponding to the group of contents associated with the sound source position.

When defining the sound source position for each group of contents, if the group of contents exists in the order along the flow of the text sentence, the sound source position for each group of the content is the flow of the sentence. The positions are defined so that they change in the order along.

Further, in the above-mentioned synthesized speech reproducing apparatus, when the texts to be speech-synthesized are a plurality of independent texts, the virtual sound source position in the sound space is defined for each text, The synthesized voice for the text associated with the sound source position is output from the sound source position.

When a plurality of independent texts can be classified by a certain criterion, the virtual sound source position in the sound space is defined for each classification, and the sound source position is calculated from each sound source position. The synthesized voice for the text belonging to the category associated with is output.

In the synthesized voice reproduction device, when the text to be voice synthesized is text existing in a plurality of applications in an information processing system having a multitask function, a virtual sound source in the sound space. The position is defined for each application, and the synthesized sound is output from each sound source position for the text existing in the application associated with the sound source position.

As described above, according to the present invention, by adding spatial position information to the synthesized voice information and reproducing the synthesized voice information, various additional information other than the synthesized voice can be transmitted to the user, and the information that can be transmitted to the user. The amount can be increased.

That is, the content of the text is transmitted to the user by voice by simply synthesizing the voice and outputting the synthesized voice as the synthesized voice. However, according to the present invention, which portion of the text is synthesized. Various additional information other than the synthesized voice can be transmitted to the user, such as whether the voice is output, which text has a synthesized voice output when a plurality of texts are present, and the like.

Specifically, when there are a plurality of groups of contents in the text to be speech-synthesized, a sound source position is defined for each of the plurality of groups, and from each sound source position, the text corresponding to the sound source position is defined. I am trying to output synthetic speech.

For example, the sound source position of the synthesized voice corresponding to each speaker obtained by performing the voice synthesis processing on the text in which the minutes of the conference by a plurality of speakers are described is made different for each speaker. be able to. Thereby,
It becomes easy to understand which speaker uttered what and how the conference was held.

It is also possible to make the sound source position of the synthetic voice different for each paragraph of the text or for each group of contents corresponding to the transfer of text. When there is a group of contents along the flow of a sentence, such as a group of contents corresponding to this paragraph or Kiseki-Koi, change the sound source position for each group in the order along the flow of the sentence. By defining it at a certain position, it becomes easier for a user who listens to the synthesized voice to understand the flow of the sentence, and at which part of the text the synthesized voice is currently output from the sound source position where the synthesized voice is output. You can know intuitively.

Further, when the texts to be speech-synthesized are a plurality of independent texts, the sound source position is defined for each text, and the synthesized speech of the text corresponding to the sound source position is output from each sound source position. By doing so, it is possible to intuitively know which text of the synthesized voice is currently output from the sound source position where the synthesized voice is output.

When such a plurality of texts can be classified into several groups based on a certain criterion, the sound source position can be defined for each classification, whereby the synthesized speech is output. From the sound source position, it is possible to intuitively know which classification of the synthesized voice of the text currently being output.

Further, when the text to be speech-synthesized is the text existing in a plurality of applications in the information processing system having the multitask function, the sound source position can be defined for each application, whereby the sound source position can be defined. It is possible to intuitively know which application currently outputs the synthesized voice of the text from the sound source position where the synthesized voice is output.

As described above, according to the present invention, the position information is added to the synthesized voice and the synthesized voice is reproduced, so that which portion of the text is output as the synthesized voice, and when a plurality of texts are present, which text is synthesized. Various additional information other than the synthetic voice, such as whether or not the voice is output, can be transmitted to the user.

When it is desired to hear the synthesized voice of a certain text or the synthesized voice of a specific portion of the text again,
It becomes easy to specify the content to be heard again from the sound source position where the synthesized voice is output, and it becomes extremely convenient when it is desired to know the content of a certain text or a specific portion of the text in more detail.

[0033]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below. The contents explained in this embodiment are
It includes both the synthetic speech reproduction method and the synthetic speech reproduction apparatus of the present invention.

FIG. 1 is a diagram showing the configuration of a synthesized voice reproduction apparatus of the present invention. The operation of each component will be described later. A text data (hereinafter simply referred to as “text”) 1, a text analysis unit 2, a text analysis dictionary 3, a voice synthesis unit 4, a voice synthesis dictionary 5, a sound source position definition unit 6, a sound image localization processing unit 7, and a stereoscopic sound image output unit 8. It has a structure that has.

The text analysis unit 2 analyzes the text 1 by referring to the text analysis dictionary 3. For example, the morphological analysis processing necessary for performing speech synthesis and the structure of sentences,
It has functions such as grasping the logical flow of sentences, and the information necessary to define the text to be synthesized from the contents of the text and the sound source position of the synthesized speech in a virtual sound space (this Information is referred to as sound source position definition information), text information to be subjected to voice synthesis is passed to the voice synthesis unit 4, and sound source position definition information is passed to the sound source position definition unit 6. This is the basic operation of the text analysis unit 2, but in addition to this, it has various functions for realizing various specific examples shown below, which will be described each time.

The voice synthesizing unit 4 performs a voice synthesizing process using the voice synthesizing dictionary 5 on the text to be subjected to the voice synthesizing passed from the text analyzing unit 2. The voice synthesizing unit 4 uses a known voice synthesizing processing technique. Can be used.

The sound source position definition unit 6 uses the sound source position definition information passed from the text analysis unit 2 as a virtual sound source position in the sound space with the position of the synthesized voice obtained by the voice synthesis by the voice synthesis unit 4. Is defined in.

The sound image localization processing unit 7 localizes the synthesized voice from the voice synthesis unit 4 to the virtual sound source position defined by the sound source position definition unit 6, and the sound image localization processing unit 7 and the sound source position definition unit 6 Known techniques can be used.

The stereoscopic sound image output unit 8 outputs the synthesized voice from the voice synthesis unit 4 from the virtual sound source position defined by the sound source position definition unit 6.

The operation of the present invention having such a configuration will be described below with reference to some specific examples.

[Specific Example 1] This specific example 1 is an example in which text composed of contents uttered by a plurality of speakers, such as a conference proceedings and dialogue sentences, is subjected to voice synthesis processing and the synthesized voice is output.

For example, as shown in FIG. 2, speakers (speaker A1, speaker A2, and speaker A3) are speakers A1 and A, respectively.
Consider a case where a conference is held in the order of the speaker A3 and the speaker A3 in this order, and the minutes of the speakers A1, A2, and A3 exist as texts. At this time, in order to indicate the utterance content for each speaker in the text, for example, speaker A1: "...", speaker A2: "...", speaker A3: "... -", Each speaker and its utterance content are associated with each other.

In FIG. 1, the respective speakers A1 and A
The text 1 in which 2, 2 and its utterance contents are associated is given to the text analysis unit 2. Text analysis unit 2
Then, the utterance content of the speaker A1, the utterance content of the speaker A2, and the speaker A
For each of the three utterance contents, each speaker is grasped as one group, and each utterance content is passed to the voice synthesizing unit 4. Here, since there are three speakers, the number of speakers is three.
Information indicating that the person is a person and information indicating the arrangement positions of the speakers A1, A2, and A3 are passed to the sound source position definition unit 6 as the sound source position definition information.

The sound source position definition unit 6 uses the sound source position definition information passed from the text analysis unit 2 to define a virtual sound source position in the sound space. On the other hand, the voice synthesis unit 4 performs voice synthesis processing on the utterance contents corresponding to each speaker, and passes the synthesized voice to the sound image localization processing unit 7. The sound image localization processing unit 7 localizes the synthesized voices corresponding to the speakers A1, A2, and A3 from the voice synthesis unit 4 to the virtual sound source position defined by the sound source position definition unit 6.

As a result, in this case, the three-dimensional sound image output unit 8 outputs the synthesized voice corresponding to the utterance contents of the speakers A1, A2, A3 from the sound source position determined for each speaker.

For example, as shown in FIG. 3, the synthesized voice for the utterance content of the speaker A1 can be heard from the left end in the sound space 20, and the synthesized voice for the utterance content of the speaker A2 can be heard from the center, The synthesized speech for the utterance content of A3 is heard from the vicinity of the right end, and the synthesized speech for the utterance content of each of the speakers A1, A2, and A3 comes from different positions for each speaker.

As described above, since the synthesized voices for the utterance contents of a plurality of speakers are emitted from different positions for each speaker, the texts created by the utterances of a plurality of speakers such as conference proceedings and dialogue sentences can be generated. In the case of the synthesized speech obtained by the speech synthesis processing, the synthesized speech corresponding to the utterance content of each speaker can be heard while being distinguished for each speaker. Moreover, the position of the sound source corresponding to each speaker can be set to a position that reflects the original position of each speaker (see FIG. 2) to some extent, which makes it easier to grasp the situation at the place and facilitates meetings and conversations. You can get a sense of reality.

[Specific Example 2] In this specific example 2, when a text having a large amount of text is output as a synthesized voice, the text to be voice-synthesized is divided into groups of content along the flow of the text, and the content of the content is divided. The sound source position for each group is defined in a virtual sound space.

As a first example thereof (this will be referred to as 1
It is assumed that, for example, the text 1 including a set of some contents as shown in FIG. 4 is input to the text analysis unit 2. In the example of FIG. 4, a group of contents is a paragraph, and here, there are three paragraphs B1, B2, and B3. In addition, the flow of sentences is assumed to be in the order of paragraphs B1, B2, B3 as indicated by arrow a.

When the text 1 in which such paragraphs B1, B2, B3 exist is input to the text analysis unit 2, the texts corresponding to the respective paragraphs B1, B2, B3 are the texts to be voice-synthesized. And the information indicating that there are three paragraphs and the connection between the paragraphs B1, B2, and B3 are passed to the sound source position definition unit 6 as the sound source position definition information.

The sound source position definition unit 6 uses the sound source position definition information passed from the text analysis unit 2 to define a virtual sound source position in the sound space. On the other hand, the voice synthesizing unit 4 performs a voice synthesizing process on the text content of each of the paragraphs B1, B2 and B3, and passes the synthesized voice to the sound image localization processing unit 7. The sound image localization processing unit 7 outputs the synthesized speech corresponding to the respective paragraphs B1, B2, B3 from the speech synthesis unit 4 to the sound source position definition unit 6
Localize to the virtual sound source position defined in.

As a result, in this case, the stereophonic sound image output unit 8 outputs the synthesized voice corresponding to the text contents of the paragraphs B1, B2, B3 from different positions in the sound space for each paragraph.

For example, as shown in FIG. 5, the synthesized voice for the contents of paragraph B1 is heard near the left end in the sound space 20, the synthesized voice for the contents of paragraph B2 is heard near the center of the sound space 20, and paragraph B3 is used. The synthesized speech for the content of each paragraph is heard from near the right end on the sound space 20.
It comes from different positions on the zero.

When the entire text is divided into groups (paragraphs in this case) along the flow of the sentence in this way, the sound source positions for each of the divided groups (paragraphs) are shown in FIG. Thus, if the sound source position is switched in one direction (along arrow b) along the flow of the sentence, the sound source position is sequentially switched in one direction. Makes it easier to understand the contents of the text.

In this way, the synthesized voices for a plurality of paragraphs are emitted from different positions in the sound space for each paragraph, and the sound source positions are sequentially switched in one direction along the flow of the sentence. , Even if it is a synthetic speech obtained by performing speech synthesis processing on a large amount of text,
You can intuitively know which side of the text the synthesized voice is output.

Thus, when a certain portion of the text is to be browsed, it is possible to know which side of the entire text the portion is. Further, when it is desired to hear the contents of a part again later, it is possible to specify the position in the sound space from the sound source position of the part.

Although a paragraph has been taken as an example of a group of text contents here, it is not limited to a paragraph, and for example, a text in which there are several headings and a sentence exists for each heading, etc. Can be considered similarly.

As a second example (this is referred to as Specific Example 2-2), some logical elements forming a text are grouped into sentences in the same manner as in Specific Example 2-1, An example of defining a sound source position for each element as a unit in a virtual sound space will be described. For example,
If the text is a text that clearly expresses the kyōshū kyōsetsu, each part of the kyōshū kyōshō is used as an element, and the sound source position is defined for each element in a virtual sound space.

As an example of the text in which the succession / transfer is clearly expressed, a technical paper or the like can be considered. in this case,
The text analysis unit 2 reads the technical paper as the text 1 to be speech-synthesized, performs a text analysis for grasping the structure of the technical paper, and forms an element (for example, a title, (Summary, preface, text, conclusion, etc.) is grasped, and the text contents corresponding to each element are passed to the voice synthesis unit 4,
Information on how many elements are configured and information indicating the connection between the respective elements are passed to the sound source position definition unit 6 as sound source position definition information. Here, it is assumed that the title, the abstract, and the preamble are one element C1, that the body is divided into two parts, the elements are C2 and C3, and the conclusion is the element C4.

The sound source position definition unit 6 uses the sound source position definition information passed from the text analysis unit 2 to determine a virtual sound source position in the sound space. On the other hand, the voice synthesizing unit 4 performs a voice synthesizing process on the text contents of each of the elements C1, C2, C3, and C4 forming the text, and passes the synthesized voice to the sound image localization processing unit 7. The sound image localization processing unit 7 converts the synthesized speech corresponding to the elements C1, C2, C3, C4 passed from the speech synthesis unit 4,
The sound source position is defined at the virtual sound source position determined by the sound source position definition unit 6.

As a result, in this case, the three-dimensional sound image output unit 8 outputs the synthesized voice obtained by voice-synthesizing the text contents corresponding to the elements C1, C2, C3, and C4 in the sound space for each element. It is output from different positions.

For example, as shown in FIG. 6, the synthesized voices corresponding to the respective elements C1, C2, C3, and C4 sequentially switch their sound source positions from the left end direction toward the right end direction in the sound space 20. I can hear it.

Also in this case, as in the case of the specific example 2 described above,
The sound source position for each unit (in this case, the respective elements C1, C2, C3, C4 corresponding to the locomotive transfer) is defined to switch in one direction (direction of arrow b) along the flow of the sentence. Then, as shown in FIG. 6, the sound source position is sequentially switched in one direction along the flow of the sentence, so that the user who hears it can easily understand the logic development.

As described above, since the synthetic voices corresponding to some elements constituting the text are emitted from different positions for each element, it is intuitive to know which side of the text the synthetic voice is currently output. Understand. This is particularly effective for texts in which the spontaneous transitions are clearly expressed, such as in technical papers.In this way, by defining the sound source position for each element that composes the text in the sound space, As in the case of 1 in Example 2, when it is desired to hear the contents of a part again later, the position in the sound space can be easily specified from the position of the sound of the part.

It should be noted that, in the first specific example 2 or the second specific example 2, as an example of the switching direction of the sound source position, the direction from the left end to the right end (or vice versa).
Although the example has been described in which the sound source positions are sequentially switched in one direction to, it is possible to define the sound source positions so as to draw a large loop in the sound space 20 when it is necessary to define a large number of sound source positions. In short, the sound source position may be defined so that the sound source is switched sequentially along the flow of the sentence.

[Specific Example 3] In Specific Example 3, when a plurality of independent texts exist, a sound source position is defined for each text, and a synthesized voice of the text corresponding to the sound source position is output from each sound source position. To do.
For example, although not shown, three texts (this is D1,
D2, D3), the contents of the respective texts D1, D2, D3 are passed to the voice synthesizing unit 4 for voice synthesizing corresponding to the respective texts, and the three texts D1, D2, D3 are The sound source position definition information indicating the existence is passed to the sound source position definition unit 6 to define a virtual sound source position.

As a result, in this case, the three-dimensional sound image output unit 8 outputs the synthesized voice for the texts D1, D2, D3 from different sound source positions for each text.

When a large number of such independent texts are present, the large number of texts are clustered according to their contents and the sound source position in the sound space is defined for each clustering result of the clustering processing. You can also

FIG. 7 shows the result of clustering performed by the clustering process. Here, four clusters E1, E2, E are obtained by the clustering process.
3 and E4 are obtained (in FIG. 7, black circles shown in the respective clusters E1, E2, E3, and E4 represent individual texts belonging to the cluster).

The text analysis unit 2 performs this clustering process. That is, in the text analysis unit 2, for each of many input texts 1, each text 1
Is analyzed, clustering is performed based on the analysis result, and the text for each cluster (here, clusters E1, E2, E3, E4) obtained by the analysis is passed to the speech synthesis unit 4, and a sound source such as the number of clusters is generated. The position definition information is passed to the sound source position definition unit 6.

The sound source position definition unit 6 determines a virtual sound source position in the sound space by using the sound source position definition information passed from the text analysis unit 2. On the other hand, the voice synthesizing unit 4 performs a voice synthesizing process on the respective text contents belonging to the clusters E1, E2, E3, E4, and passes the synthesized voice to the sound image localization processing unit 7. The sound image localization processing unit 7 localizes the synthesized speech of each text belonging to the clusters E1, E2, E3, E4 from the speech synthesis unit 4 to the virtual sound source position determined by the sound source position definition unit 6.

As a result, in this case, the three-dimensional sound image output unit 8 produces a different synthesized speech obtained by speech-synthesizing the text contents belonging to the clusters E1, E2, E3, E4 for each cluster in space. It is output from the position. For example, as shown in FIG. 8, the text belonging to the cluster E1 is heard from the rear of the central portion on the sound space 20, the text belonging to the cluster E2 is heard from the left end on the sound space 20, and the text belonging to the cluster E3 is heard. In the sound space 20, the sound is heard from the front of the center, and the text belonging to the cluster E4 is heard from the right end direction in the sound space 20, and so on.

As described above, since the synthesized voice of the text belonging to each cluster is emitted from a different position for each cluster, it becomes easy to grasp which cluster the text belongs to now. In addition, it is intuitively known which cluster the text synthesized voice is currently outputting.

Further, by defining the sound source position for each cluster in the sound space in this way, when it is desired to hear a text belonging to a certain cluster again, the position of the sound corresponding to the cluster is changed in the sound space. You can also specify the position with.

As a similar example to the third specific example, a synthetic voice for a large number of texts in which opinions such as a result of discussion and a letter of comment such as approval, disagreement, and neutrality are written on a sound space for each content of the opinion. You can also define the sound source position with.

From the contents of the large number of texts, it is assumed that the text analysis unit 2 also performs a process of determining whether the opinion is a supportive opinion, a dissenting opinion, or a neutral opinion, and classifying it. Each of the texts belonging to the group F1, the opposite opinion group F2, and the neutral opinion group F3 is passed to the speech synthesis unit 4, and in this case, a large number of input texts are classified into three categories: yes, no, and neutral. Information indicating that the sound source position is defined is passed to the sound source position definition unit 6 as sound source position definition information.

As a result, in this case, the three-dimensional sound image output unit 8 produces synthetic voices obtained by voice-synthesizing the texts respectively belonging to the opinion group F1, the opinion group F2, and the neutral opinion group F3. Are output from different positions in the sound space for each group of opinions.

For example, as shown in FIG. 9, the synthetic voice for the text belonging to the group F1 of the opinion is heard from the left end on the sound space 20, and the synthetic voice for the text belonging to the group F2 of the disagreement is the sound space 2.
0 is heard from the right end direction, and the synthesized voice for the text belonging to the neutral opinion group F3 is heard in the vicinity of the center in space, so that the synthesized voice for the text belonging to each group is different for each opinion group. Can be heard from the position.

In this way, since the synthesized voice of the texts belonging to each opinion is uttered from different positions for each opinion, it becomes easy to grasp which opinion the text belongs to now, and Now, it is possible to intuitively understand to which opinion the synthesized voice of the text is output.

Further, by defining the sound source position for each opinion in the sound space in this way, when it is desired to hear a text belonging to a certain opinion again, the sound position corresponding to the opinion is changed in the sound space. You can also specify the position with.

Concrete Example 4 Concrete Example 4 is an example in which, in an information processing system having a multitask function, sound sources are localized at different positions for each application.

As a first example (this will be referred to as Specific Example 1), when a plurality of windows and dialogs are opened in a GUI (Graphical User Interface) such as a personal computer (hereinafter referred to as a PC) , The sound source position is set to a different position in the sound space for each window or dialog.

For example, three windows G1, G2 and G3 as shown in FIG. 10 are opened on the screen 30 of the PC, and the synthesized voice for each text existing in each window G1, G2 and G3 is output. If
As shown in FIG. 11, the sound source position of the synthetic voice for the text existing in the window G1 is defined in the upper left direction in the sound space 20, and the sound source position of the synthetic voice for the text existing in the window G2 is true in the sound space 20. The sound source position is defined in the upward direction, and the sound source position of the synthesized voice with respect to the text existing in the window G3 is defined in the upper right direction in the sound space 20, so that the sound source position is different for each window.

As a result, it is possible to intuitively know from which output position of the synthesized voice the text of which window is the synthesized voice.

Further, as a second example of this specific example 4 (this will be referred to as specific example 2-2), for example, it has a scheduler function, an electronic mail transmitting / receiving function, a web news browsing function, and the like, and As shown in FIG. 12, in a portable information device 40 or the like capable of outputting output sound as stereophonic sound with headphones 41 or the like, respective applications H1 fulfilling these scheduler functions and respective applications H2 fulfilling e-mail transmission / reception functions. , It is also possible to define the sound source position for each application H3 that performs a function of browsing news on the web in the sound space 20 as shown in FIG.

In the example of FIG. 13, the sound source position of the synthesized voice for the text created by the application H1 of the scheduler function is defined near the left end in the sound space 20,
The sound source position of the synthetic voice for the text created by the application H2 of the transmission / reception function of the electronic mail is the sound space 2
0 is defined near the center, and the sound source position of the synthesized voice for the text acquired by the application H3 of the web news browsing function is defined near the right end in the sound space 20.

As a result, it is possible to intuitively know from which output position of each synthesized voice the synthesized speech for which text of the application is outputted.

It is possible to define the sound source position for each application as in the case of 1 of the fourth example or the second of the fourth example, and further, even if the same application is used, a plurality of positions obtained by the application can be obtained. It is also possible to localize the sound source for each text.

For example, in an application for sending and receiving electronic mail, whether the electronic mail is a business mail, a mail from a family member, a mail from a friend, or the like is divided into several groups. Divided into,
It is also possible to localize the sound source position for each group, so that it is possible to intuitively know what kind of partner the mail is from.

By the way, in this specific example 4, the text 1 of each application is converted into the text analysis unit 2.
When inputting into the text 1, the data that indicates in which application the text 1 exists is text 1.
If it is added as a header of, the text analysis unit 2 can identify the application by looking at the header portion and pass it to the sound source position definition unit 6 as sound source position definition information. The text existing in each application is passed to the speech synthesizer 4 together with the information for identifying each application.

Further, in the fourth specific example, when the application has a plurality of frames, the sound source position can be determined for each frame.

Furthermore, the present invention is not limited to the specific examples described above, and various modifications can be made without departing from the gist of the present invention.

Further, according to the present invention, a processing program in which a processing procedure for realizing the above-described present invention is described is created, and the processing program is recorded in a recording medium such as a floppy disk, an optical disk or a hard disk. The present invention also includes a recording medium in which the processing program is recorded. Further, the processing program may be obtained from the network.

[0094]

As described above, according to the present invention, the text input as the speech synthesis target is analyzed, the text to be speech-synthesized is extracted, and the information necessary for defining the sound source position of the synthesized speech is obtained. Extracted as sound source position definition information, using the sound source position definition information to define a virtual sound source position in a sound space, performing speech synthesis processing on the text to be speech-synthesized, and synthesizing it. Since the sound is localized at a virtual sound source position on the sound space defined by using the sound source position definition information, and the synthesized voice is output from the localized position, the amount of information that can be transmitted to the user. Can be increased. That is,
By simply synthesizing the text and outputting the synthesized speech as the synthesized speech, only the content of the text is audibly transmitted to the user. However, according to the present invention, the synthesized speech of any part of the text is output. It is possible to convey various additional information other than the synthesized voice to the text to the user, such as which synthesized voice of which text is output when there are a plurality of texts.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a synthetic voice reproduction device of the present invention.

FIG. 2 is a diagram for explaining a specific example 1 in the embodiment of the present invention, and is a diagram showing a plurality of speakers surrounding a round table.

FIG. 3 is a diagram illustrating an example in which a sound source position of a synthesized voice for a text corresponding to each speaker shown in FIG. 2 is defined for each speaker in a sound space.

FIG. 4 is a diagram showing an example of a text consisting of several paragraphs used for explaining specific example 1 of example 1 in the embodiment of the present invention.

5 is a diagram illustrating an example in which a sound source position of synthesized speech for each paragraph of the text of FIG. 4 is defined for each paragraph in a sound space.

FIG. 6 is a diagram for explaining the second specific example 2 of the embodiment of the present invention (the text in which the succession and rearrangement is clearly expressed) for explaining the specific example 2; It is a figure explaining the example defined for every element of.

FIG. 7 is a diagram showing an example in which a plurality of texts used for explaining a specific example 3 in the embodiment of the present invention are clustered and classified into some clusters.

8 is a diagram illustrating an example in which a sound source position of synthesized speech of a text belonging to each cluster obtained by the clustering process illustrated in FIG. 7 is defined for each cluster in a sound space.

[FIG. 9] As a similar example to Concrete example 4, the sound source position of the synthetic voice for a large number of texts in which opinions such as the result of discussions and comments such as approval, disagreement, and neutrality are written is defined for each content of the opinion in the sound space. It is a figure explaining the example which did.

FIG. 10 is a diagram for explaining specific example 1 of Embodiment 4 of the present invention, and is a diagram showing an example in which different windows are displayed on the screen for each application in the information processing system having the multitask function. is there.

11 is a diagram illustrating an example in which a sound source position of a synthesized voice of text corresponding to a plurality of applications illustrated in FIG. 10 is defined for each application in a sound space.

FIG. 12 is a diagram for explaining the second specific example 4 in the embodiment of the present invention, which is a diagram showing an example of a portable information device having a plurality of functions and capable of outputting an output sound as a stereo sound. is there.

13 is an example in which a sound source position of a synthesized voice of a text created or acquired by each application corresponding to each function in the portable information device as shown in FIG. 12 is defined for each application in the sound space. It is a figure explaining.

[Explanation of symbols]

1 Text data to be voice-synthesized 2 Text analysis unit 3 Text analysis dictionary 4 Voice synthesis unit 5 Voice synthesis dictionary 6 Sound source position definition unit 7 Sound image localization processing unit 8 Stereo sound image output unit A1, A2, A3 Speaker B1, B2 B3 Paragraphs C1, C2, C3, C4 Elements E1, E2, E3, E4 corresponding to the inscription transfer Clusters F1, F2, F3 obtained by clustering Approval, opposition, neutral opinion groups G1, G2, G3 Open on the screen Window H1 Application for scheduler function H2 Application for email sending / receiving function H3 web Application for news browsing function

Claims (12)

[Claims] 1. A sound source position, which is information necessary for analyzing a text input as a voice synthesis target, extracting a text to be voice synthesized, and defining a sound source position of synthesized voice in a virtual sound space. Extracted as definition information, the sound source position definition information is used to define a sound source position in the virtual sound space, and a voice synthesis process is performed on the text to be voice-synthesized. Is localized at a virtual sound source position on the defined sound space, and the synthesized voice is output from the localized position. 2. When a plurality of content groups are present in the text to be speech-synthesized, a virtual sound source position in the sound space is defined for each content group, and the sound source position The synthetic voice reproduction method according to claim 1, wherein a synthetic voice for a group of contents associated with a sound source position is output. 3. When defining a sound source position for each group of contents, if the group of contents exists in an order along the flow of a text sentence, the sound source position for each group of contents is 3. The synthetic voice reproduction method according to claim 2, wherein the position is defined so as to change in an order along the flow. 4. When the text to be speech-synthesized is a plurality of independent texts, a virtual sound source position in the sound space is defined for each text, and from each sound source position to that sound source position. 2. The synthetic voice reproduction method according to claim 1, wherein a synthetic voice for the text associated with is output. 5. When the plurality of independent texts can be classified by a certain criterion, a virtual sound source position in the sound space is defined for each classification, and the sound source positions are used to define the sound source. The synthetic voice reproduction method according to claim 4, wherein a synthetic voice for the text belonging to the classification associated with the position is output. 6. When the text to be speech-synthesized is text existing in a plurality of applications in an information processing system having a multitask function, a virtual sound source position in the sound space is set for each application. 2. The synthetic voice reproduction method according to claim 1, wherein the synthetic voice for the text existing on the application defined in the above and corresponding to the sound source position is output from each sound source position. 7. The sound source position definition information for analyzing the text to be synthesized, extracting the text information to be synthesized, and defining the sound source position of the synthesized speech in a virtual sound space. And a sound source position defining unit that defines a virtual sound source position in a sound space using the sound source position definition information extracted by the text analyzing unit, and the text extracting unit extracted by the text analyzing unit. A voice synthesizing means for synthesizing the text information to be voice-synthesized, and a synthesized voice obtained by the voice synthesizing processing by the voice synthesizing means, a virtual sound source in the sound space defined by the sound source position defining means. Sound image localization means for locating to a position, and stereoscopic sound image output means for outputting the synthesized voice from the sound source position localized by the sound image localization means, Synthesized speech reproducing device characterized in that it comprises. 8. When a plurality of groups of contents exist in the text to be speech-synthesized, a virtual sound source position in the sound space is defined for each unit of the contents, and the virtual sound source position is defined from each sound source position. 8. The synthetic speech reproduction device according to claim 7, wherein synthetic speech is output for a group of contents associated with a sound source position. 9. When defining a sound source position for each group of the contents, if the group of the contents exists in an order along the flow of a text sentence, the sound source position for each group of the contents is the sentence position of the sentence. 9. The synthesized speech reproducing apparatus according to claim 8, wherein the synthesized speech reproducing apparatus is defined at a position that changes in order along the flow. 10. When the text to be speech-synthesized is a plurality of independent texts, a virtual sound source position in the sound space is defined for each text, and from each sound source position to the sound source position. 8. The synthetic speech reproducing apparatus according to claim 7, wherein the synthetic speech for the text associated with is output. 11. When the plurality of independent texts can be classified based on a certain criterion, a virtual sound source position in the sound space is defined for each classification, and the sound source positions are used to define the sound source. 11. The synthetic speech reproduction device according to claim 10, wherein synthetic speech for text belonging to a classification associated with a position is output. 12. When the voice synthesis target text is text existing in a plurality of applications in an information processing system having a multitask function, a virtual sound source position in the sound space is set for each application. 8. The synthesized speech reproducing apparatus according to claim 7, wherein the synthesized speech reproducing apparatus outputs the synthesized speech corresponding to the text existing on the application defined by the respective sound source positions and associated with the sound source position.