JP2001337688A - Voice synthesizer, voice systhesizing method and its storage medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To synthesize a voice which expresses feelings representing a character strings such as pictographs. SOLUTION: A morpheme analysis module 203 detects a prescribed character string from a text information. A pictograph analysis module 207 consults a pictographic dictionary 208 to obtain a correction value for the metrical parameter of the detected character string and supply it to a metrical parameter production module 204. At the metrical parameter production module 204, the metrical parameter produced from the text information other than the prescribed character string is corrected with the correction value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech synthesizer for synthesizing speech information from character information, a speech synthesis method, and a storage medium.

[0002]

2. Description of the Related Art Generally, in a speech synthesizer for synthesizing speech from text, a text is input and a synthesized waveform is output. In such a speech synthesizer, a language analysis unit performs a language analysis of a text, and a sound processing unit performs selection of a speech unit and synthesis of a waveform.

[0003]

[Problems to be Solved by the Invention] The text described in the e-mail and the homepage includes "(^^)" and "(^ _ ^)"
May appear. Such a character string is called an emoji, an emoticon, or a smiley, and is one of the techniques for expressing the emotion of a text creator.

However, in the conventional speech synthesizer,
There is a problem that such a character string is determined as a simple character string, and it can only be uttered as a synthesized voice as it is or not to be uttered. In either case, there is a problem that emotions represented by character strings such as pictograms cannot be conveyed by synthetic speech.

The present invention has been made in view of the above problems, and provides a voice synthesizing apparatus, method, and storage medium capable of synthesizing voice expressing an emotion represented by a character string such as a pictogram. Aim.

[0006]

In order to achieve the object of the present invention, for example, a speech synthesizer of the present invention has the following arrangement. That is, it comprises a detecting means for detecting a predetermined character string from the character information, and a control means for controlling the prosody of voice information synthesized from the character information according to the predetermined character string.

[0007] In order to achieve the object of the present invention, for example, a speech synthesizer of the present invention further has the following configuration. That is, the apparatus further comprises a registration unit for registering the predetermined character string.

[0008] In order to achieve the object of the present invention, for example, the speech synthesizer of the present invention further has the following configuration. That is, the apparatus further includes a voice synthesizing unit that synthesizes voice information from the character information, and the voice synthesizing unit synthesizes voice information from character information other than the predetermined character string.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below in detail according to preferred embodiments with reference to the accompanying drawings.

[First Embodiment] FIG. 1 is a diagram showing a schematic configuration of a speech synthesizing apparatus according to this embodiment. The speech synthesizer according to the present embodiment is applicable to stationary information terminals such as personal computers, workstations, and car navigation systems, and portable information terminals such as mobile phones and portable personal computers.

In FIG. 1, 101 is an input unit, 102 is a speech synthesis unit, 103 is an output unit, 104 is a storage unit,
Reference numeral 105 denotes a network interface, and reference numeral 106 denotes a control unit.

The input unit 101 includes a character information input device such as a keyboard and a pointing device such as a mouse, and inputs character information and a user operation.

The voice synthesizing unit 102 includes an input unit 10
1. Speech information is synthesized from texts (consisting of languages such as Japanese and English) contained in an e-mail, a homepage, a sentence file, and the like input from the storage unit 104 and the interface 105.

The speech synthesis unit 102 of the present embodiment
It can be configured by hardware or software. In the present embodiment, a case where the speech synthesis unit 102 is configured by software will be described. In this case, a control program for implementing the speech synthesis unit 102 is stored in a storage medium such as the storage unit 104. This control program is read by the control unit 106 and processed according to the processing procedure described later.

The output unit 103 includes a speaker or the like, and outputs a synthesized voice generated by the voice synthesis unit 102.

The storage unit 104 includes a semiconductor memory, a magnetic recording medium, an optical recording medium, a hard disk, and the like, and stores various control programs for controlling functions of the speech synthesizer. The storage unit 104 stores an e-mail, a homepage, a text file, and the like created by using the input unit 101, an e-mail received by the interface 105,
Store homepages, text files, etc.

The interface 105 communicates with other information terminals connected to a network such as a telephone network, a mobile communication network, the Internet, a satellite communication network, and receives an e-mail, a home page, a text file, and the like. ,
The voice information synthesized by the voice synthesis unit 102 is transmitted to an information terminal specified in advance.

The control unit 106 includes a central processing unit (CP)
U). The control unit 106 reads various control programs stored in the storage device 104 and controls the functions of the speech synthesizer.

FIG. 2 is a diagram showing a functional configuration of the voice synthesizing unit 102 in the present embodiment.

In the figure, 201 is a text input module, 202 is a reading module, 203 is a morphological analysis module, 204 is a prosody parameter generation module, 205 is a phoneme segment selection module, 206 is a waveform editing module, 207 is a pictogram analysis module, 208 is an emoticon dictionary, 209 is a phoneme dictionary, 210 is a word dictionary,
211 is a grammar rule, and 212 is a pictogram registration module.

The text input module 201 manages input of text included in an e-mail, a homepage, a text file and the like designated by the user. Texts contained in e-mails, homepages, text files, etc. are input to the input unit 101, the storage unit 104 or the interface 105
Supplied from The input text is stored in a memory managed by the control unit 106.

The reading module 202 manages reading of the text stored in the memory. The read module 202 reads character information in units of one sentence from the text stored in the memory, and supplies the read character information to the morphological analysis module 203.

The morphological analysis module 203 uses the word dictionary 210 and the grammar rules 211 to morphologically analyze and read the character information supplied from the reading module 202.
Detects prosodic information such as accent position and accent strength, and presumes emoticons or emoticons from the end of the sentence of this character information and before and after punctuation marks (“,”, “,”, “;”, “:”, etc.) Detected character string The detected prosody information is transmitted to the prosody parameter generation module 20.
4 is supplied to the pictographic analysis module 207.

The word dictionary 210 is a database that holds prosody information such as reading of each word, accent position, and accent strength. The grammar rule 211 is a database holding grammatical connection rules. Note that the word dictionary 21
0, grammar rules 211 are stored in the storage unit 104.

The pictograph analysis module 207 converts the character string detected by the morphological analysis module 203 into the pictograph dictionary 2.
08 is determined. If the character string is registered, the correction value of the prosody parameter corresponding to the character string is obtained, and the obtained correction value is supplied to the prosody parameter generation module 204.

The pictograph dictionary 208 is a database that stores a plurality of pictographs or emoticons in association with the type of emotion and the correction value of the prosodic parameter. Emoji dictionary 20
FIG. 4 shows an example of pictograms, emotion types, and correction values of prosodic parameters held by 8. For example, the emoji "(laughs)"
Represents the emotion of "fun". By controlling the prosody of the voice information based on the correction value of the prosody parameter corresponding to this emotion, it is possible to generate voice information expressing the emotion of "fun". The relationship between emotions such as “happy”, “fun”, and “sadness” and the corrected values of prosodic parameters, and “(＾＾)”, “(laughs)”, “(・ _
.;)) And the emotions are set by the pictogram registration module 212. The pictograph dictionary 20
8 is stored in the storage unit 104.

The prosody parameter generation module 204 includes:
Based on the prosody information supplied from the morphological analysis module 203, the pitch frequency (representing the pitch of the voice), the phoneme time length (representing the length of the voice), the sound power (representing the strength of the voice), and the pause length (representing the pause) Prosody parameters such as an interval) are generated, and these prosody parameters are converted to the pictogram analysis module 2
Correction is performed using the correction value acquired in step 07.

The speech unit selection module 205 selects a plurality of speech units from the speech unit dictionary 209. The phoneme segment dictionary 209 is stored in the storage unit 104.

The waveform editing module 206 edits and connects a plurality of speech units obtained by the phoneme unit selection module 205 based on the prosody parameters obtained by the prosody parameter generation module 204. The waveform editing module 206 edits and connects a plurality of speech units using the PSOLA method (pitch-synchronized waveform superposition method) or the like, and generates one speech waveform data. The audio waveform data thus obtained is supplied to the output unit 103 or the interface 105. When supplied to the output unit 103, the signal is output from a speaker, and when supplied to the interface 105, the signal is transmitted to a desired information terminal connected to a network.

The pictogram registration module 212 controls the procedure for resetting the relation between the emotion and the correction value of the prosody parameter, the relation between the pictogram and the emotion, and registering a new pictogram and a new type. When the user requests registration of a new pictogram, the pictogram registration module 212 displays a registration screen shown in FIG.

In FIG. 5, reference numeral 501 denotes an item for inputting a new pictogram. The user inputs a new pictogram to the item 501 using the input unit 201. In FIG. 5, “(＾
o ＾) ”. Reference numeral 502 denotes an item for inputting an emotion corresponding to the pictograph input in the item 501. The user uses the input unit 201 to select a desired emotion from pre-registered emotions. FIG. 5 shows an example of selecting “fun”. 503 is a decision button, and 504 is a cancel button. When the enter button 503 is pressed, the module 212 associates the pictograph input in the item 501 with the emotion selected in the item 502 and the correction value of the prosodic parameter set for this emotion, and sets the pictograph dictionary 2
Register at 08.

On the other hand, when the user requests registration of a new emotion, pictogram registration module 212 displays a registration screen shown in FIG.

In FIG. 6, reference numeral 601 denotes an item for inputting a new emotion. The user inputs a new emotion to the item 601 using the input unit 201. 602 is an item for adjusting the prosody parameter corresponding to the pitch of the voice, 603 is an item for adjusting the prosody parameter corresponding to the length of the voice, 604 is an item for adjusting the prosody parameter corresponding to the voice intensity, 6
05 is an item for adjusting the prosody parameter corresponding to the pause interval.

Reference numeral 606 denotes an item for inputting a sample.
Is a play button that supports sample playback. When the press of the play button 607 is detected, the speech synthesis unit 1
02 synthesizes voice information from the sample input to the item 606, and sets the prosodic parameter of the voice information to the item 602.
The correction is made with the correction value of the prosodic parameter corresponding to the value input to. The corrected audio information is output to the output unit 103
Output from the speaker.

Reference numeral 608 denotes an enter button, and 609 denotes a cancel button. When the pressing of the enter button 608 is detected, the module 212 registers the emotion input in the item 601 and the correction value of the prosodic parameter corresponding to the values input in the items 602 to 605 in the pictograph dictionary 208 in association with each other.

FIG. 3 is a flowchart for explaining the processing procedure of the speech synthesis unit 102 in the present embodiment.

The text input module 201 inputs a text included in an e-mail, a homepage, a text file, and the like designated by the user, and stores the text in a memory (step S301).

The reading module 202 reads character information in units of one sentence from the text held in the memory, and reads the read character information into the morphological analysis module 20.
Supply 3

The morphological analysis module 203 uses the word dictionary 210 and the grammatical rules 211 to morphologically analyze the character information read in step S302, and detects prosody information such as reading, accent position, and accent strength. At the same time, a character string presumed to be a pictogram is detected from before and after the end of the sentence of this character information and punctuation marks (“,”, “,”, “;”, “:”, etc.) (step S30)
3). For example, if morphological analysis is performed for “Today's good weather (＾＾).”, The character string “(＾＾)” starts at the end of the sentence.
Is detected. Also, for example, when the morphological analysis of “Today is fine weather (＾＾), but tomorrow is rainy (cry)” is performed, the character strings “(＾＾)” and “(cry ) "Is detected.

The prosody parameter generation module 204 includes:
A prosodic parameter of the character information subjected to morphological analysis in step S303 is generated (step S304). However, when the morphological analysis module 203 detects a character string presumed to be a pictogram or a face character, it generates a prosodic parameter of character information excluding the character string. The prosody parameters generated in step S303 are pitch frequency, phoneme time length, sound power, pause length, and the like.

If a character string is detected in step S303, this character string is supplied to pictographic analysis module 207 (step S305). On the other hand, if no character string is detected in step S303, step S308
Execute the processing of

The pictographic analysis module 207 refers to the pictographic dictionary 208 to obtain a correction value of the prosody parameter corresponding to the character string detected in step S303, and supplies the obtained correction value to the prosody parameter generation module 204 ( Step S306). When a plurality of character strings are detected from one character information, a correction value of each character string is obtained.

The prosodic parameter generation module 204
The prosody parameter generated in step S304 is corrected using the correction value obtained in step S306 (step S307).

When a plurality of character strings are detected from one piece of character information, the character information is divided into a plurality of phrases, and the prosodic parameters of each phrase are corrected with the correction value of each character string.
For example, in the case of “Today is good weather (＾＾), but tomorrow is rainy (crying).” In the case of “correct today is good weather using the correction value corresponding to the character string“ (＾＾) ”. The prosody parameter of "But it's raining" is corrected using the correction value corresponding to the character string "(cry)". With this configuration, it is possible to express one piece of character information with more emotion.

The phoneme segment selection module 205 selects a plurality of speech segments from the phoneme segment dictionary 209 (step S).
308).

The waveform editing module 206 executes step S
Prosody parameters generated in step 304 or step S3
Based on the prosodic parameters modified in step 07, the plurality of speech units selected in step S308 are edited and connected (step S309).

By performing such processing, the emotion expressed by the pictogram can be reflected in the prosody of the synthesized speech obtained from the character information excluding the pictogram. The synthesized speech obtained by the waveform editing module 206 is supplied to the output unit 103 or the interface 105.

The reading module 202 determines whether or not there is character information that has not been read yet (step S310). If there is, the next character information is read, and the read character information is read by the morphological analysis module 2.
03 (step S302). On the other hand, if the file does not exist, the process ends.

As described above, according to the present embodiment,
When synthesizing voice information from character information including a character string such as a pictogram, the prosody of the voice information can be controlled according to the type of the character string. With this configuration, it is possible to express a text including a character string such as an emoticon with a synthetic voice in an emotionally rich manner.

[Second Embodiment] In the first embodiment,
The speech synthesizing apparatus that does not modify the prosody of speech information synthesized from character information when a character string not registered in the pictograph dictionary 208 is detected has been described.

In this embodiment, the speech synthesis unit 102
Is changed, and even if a character string not registered in the pictograph dictionary 208 is detected, the correction value of the prosodic parameter is determined from the characters and symbols constituting the character string, and is synthesized from the character information. A speech synthesizer that corrects the prosody of the speech information will be described.

FIG. 7 shows a functional configuration of the speech synthesizing unit 102 in the present embodiment. Note that the same numbers are assigned to modules having the same functions as those in FIG. 70
Reference numeral 1 denotes a pictographic part dictionary, and 702, a pictographic analysis module.

The parts dictionary 701 is a database that stores predetermined characters and symbols (hereinafter, referred to as parts) in association with correction values of prosody parameters. Parts dictionary 7
An example of 01 is shown in FIG. The parts dictionary 701 is stored in the storage unit 104.

The pictograph analysis module 702 converts the character string detected by the morphological analysis module 203 into the pictograph dictionary 2.
08 is determined. If the character string is registered, the correction value of the prosody parameter corresponding to the character string is obtained. If the character string is not registered, the correction value of the prosody parameter is obtained with reference to the parts dictionary 701.

FIG. 8 is a flowchart for explaining the processing procedure of the speech synthesis unit 102 in the present embodiment.

In the figure, each processing of steps S801 to S805 is performed in step S301 in FIG.
Since the processing is the same as the processing in steps S305 to S305, the description is omitted here.

If a character string estimated to be a pictogram is detected in step S804, this character string is supplied to pictogram analysis module 702 (step S805). on the other hand,
If no character string is detected in step S804, the process of step S812 is performed.

The pictograph analysis module 702 refers to the pictograph dictionary 208 and determines whether or not the character string detected in step S804 is registered in the pictograph dictionary 208 (step S806). If it is registered, obtain the modified value of the prosodic parameter corresponding to the character string,
Prosody parameter generation module 204
(Step S807). The prosody parameter generation module 204 corrects the prosody parameter generated in step S804 using the correction value acquired in step S810 (step S808).

On the other hand, if the character string detected in step S804 is not registered in pictogram dictionary 208, the parts constituting this character string are detected (step S80).
9). The pictographic analysis module 702 refers to the pictographic parts dictionary 701, acquires a correction value of the prosody parameter corresponding to each part, and supplies the obtained correction value to the prosody parameter generation module 204 (step S810).
The prosody parameter generation module 204 determines in step S8
Each time a correction value is obtained in step 10, the prosody parameter generated in step S804 is corrected (step S81).
1).

The phoneme segment selection module 205 selects a plurality of speech segments from the phoneme segment dictionary 209 (step S).
812).

The waveform editing module 206 determines in step S
Prosodic parameters generated in step 804, step S808
Based on the prosody parameters modified in step S811 or the prosody parameters modified in step S811
The plurality of speech units selected in step 7 are edited and connected (step S813).

By performing such processing, when a pre-registered pictograph is detected, the emotion expressed by the pictograph can be reflected in the prosody of the synthesized speech obtained from the character information excluding the pictograph. Becomes Further, even when a pre-registered pictogram is not detected, it is possible to change the prosody of the synthesized speech by the correction value corresponding to the pre-registered part.

The reading module 202 determines whether or not there is character information that has not been read yet (step S814). If there is, the next character information is read, and the read character information is read by the morphological analysis module 2.
03 (step S802). On the other hand, if the file does not exist, the process ends.

As described above, according to the present embodiment,
Even when a character string that is not registered in the pictograph dictionary 208 is detected, it is possible to control the prosody of the speech information synthesized from the character information according to the parts constituting the character string.

[Third Embodiment] In the second embodiment,
The speech synthesizing apparatus that controls the prosody of the speech information synthesized according to the parts forming the character string even when the character string not registered in the pictograph dictionary 208 is detected has been described.

In this embodiment, the speech synthesis unit 102
Is changed to detect a character string that is not registered in the pictograph dictionary 208, the closest (similar) pictograph to this character string is determined using the dynamic programming (hereinafter, DP) method. A speech synthesizing device that detects from the pictograph dictionary 208 and controls the prosody of the speech information synthesized from the character information by using the detected correction value of the pictograph will be described.

FIG. 1 shows the DP method used in this embodiment.
Explanation will be made using 0. FIG. 10 is a pictogram plane used in performing DP, in which an input pattern (a character string detected by the morphological analysis module 203) is set as a horizontal axis, and a standard pattern (a pictogram in the pictogram dictionary 208) is set as a vertical axis.

The XY plane is set as a comparison point of each pattern character on the X axis and the Y axis, and a scale for evaluating the difference between the pattern characters is d (x, y). Note that d (x, y) is a scale for evaluating the time between x and y. The value of d (x, y) can be obtained by using the table shown in FIG. As shown in the figure, for example, d (“＾”, “_”) = 2.
0. Consider a path that travels from the lower left to the upper right end point in the XY plane. However, the routes that can reach the grid point (x, y) come from right beside (x-1, y), come directly below (x, y-1), and come diagonally from the lower left (x-1, y-).
Only from the three points 1).

The partial sums of the minimum distance from the starting point to the above three points are D (x-1, y), D (x, y-1), D
Assuming that it is given by (x-1, y-1), the minimum distance to the grid point (x, y) is defined by the following equation.

D (x, y) = min (D (x-1, y) + d (x, y), D (x, y-1) +
d (x, y), D (x-1, y-1) + 2d (x, y)) At this time, D (end point) is the distance (difference) between the input pattern and the standard pattern. The smaller the distance is, the more similar the pattern is. Therefore, the distance between the input pattern and all the pictographs stored in the pictograph dictionary 208 is checked, and the pictograph with the shortest distance is selected.

The speech synthesizing unit 10 in the present embodiment
The processing procedure 2 will be described with reference to FIGS. FIG.
The flowchart shown in FIG. 2 corresponds to steps S809 to S809 in FIG.
This is an example in which S811 is changed. Step S8 in FIG.
Steps other than steps 09 to S811 are processed in the same manner as the above-described procedure, and a description thereof will be omitted.

The pictograph analysis module 702 refers to the pictograph dictionary 208 and determines whether or not the character string detected in step S804 is registered in the pictograph dictionary 208 (step S806). If the character string detected in step S804 is not registered in the pictograph dictionary 208, the processing in step S901 is started.

First, an input pattern (this pictogram in the case where a pictogram is included in a character string subjected to morphological analysis by the morphological analysis module 203) is stored in a variable e1 (step S901). On the other hand, the maximum real number (∞ for the speech synthesizer) that can be handled by the speech synthesizer according to the present embodiment is stored in the variable k ′ as NULL and e ′ is initialized (step S902). In the following processing, k ′ is a variable that always contains the pictograph of the minimum distance and e ′ is the minimum distance.

The pictograph analysis module 207 extracts one pictograph notation from the pictograph dictionary 208 and stores it in the variable e2 (step S903). Next, the DP distance k between the variables e1 and e2 is determined by the above-described method (step S90).
5) If the value is smaller than the value stored in the variable k '(step S906), the value stored in the variable k is stored in the variable k', and the value stored in the variable e2 is stored in the variable e '. (Step S907). This distance calculation is performed for all pictographs stored in pictograph dictionary 208 (step S904), and the process exits the loop.

After calculating the distance between the input pattern and all the pictographs stored in pictograph dictionary 208, if the value stored in variable e 'is NULL (pictograph dictionary 2
08, the distance from all the pictographs stored in the image did not come out), and without modifying the prosodic parameters, step S8
12 is performed (step S908). If the value stored in the variable k ′ is large (greater than 5 in the example of FIG. 4), it is assumed that no pictogram similar to the input pattern is present in the pictogram dictionary 208, and the prosody parameter is not corrected and step S812 is performed. (Step S90)
9).

On the other hand, when k ′ ≦ 5, it is regarded that the pictogram is most similar to the pictogram stored in the variable e ′, and a correction value corresponding to the pictogram stored in the variable e ′ is obtained (step S910). ). The prosody parameter generation module 204 corrects the prosody parameter generated in step S804 using the correction value obtained in step S910 (step S911).

As described above, according to the present embodiment,
Even when a character string that is not registered in the pictograph dictionary 208 is detected, it is possible to control the prosody of the speech information synthesized from the character information according to the pictograph most similar to this character string.

[0078]

As described above, according to the present invention, it is possible to synthesize a voice expressing an emotion represented by a character string such as a pictogram.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a schematic configuration of a speech synthesis device according to an embodiment.

FIG. 2 shows a speech synthesis unit 10 according to the first embodiment.
FIG. 3 is a diagram illustrating a functional configuration of the second embodiment;

FIG. 3 is a speech synthesis unit 10 according to the first embodiment.
6 is a flowchart illustrating a processing procedure of No. 2;

FIG. 4 is a diagram illustrating a pictograph dictionary 208.

FIG. 5 is a diagram illustrating a registration screen for registering a new pictogram.

FIG. 6 is a diagram illustrating a registration screen for registering a new emotion.

FIG. 7 shows a speech synthesis unit 10 according to the second embodiment.
FIG. 3 is a diagram illustrating a functional configuration of the second embodiment;

FIG. 8 shows a speech synthesis unit 10 according to the second embodiment.
6 is a flowchart illustrating a processing procedure of No. 2;

FIG. 9 is a diagram illustrating a pictograph part dictionary 701.

FIG. 10 is a diagram illustrating dynamic programming.

FIG. 11 is a diagram showing a table for obtaining a scale for evaluating a difference between patterns.

FIG. 12 shows a speech synthesis unit 1 according to a third embodiment.
11 is a flowchart for explaining the processing procedure of No. 02.

Claims (15)

[Claims] 1. A method comprising: detecting means for detecting a predetermined character string from character information; and control means for controlling the prosody of voice information synthesized from the character information according to the predetermined character string. Speech synthesizer. 2. The speech synthesizer according to claim 1, wherein the predetermined character string is a character string expressing a predetermined emotion. 3. The speech synthesizer according to claim 1, wherein the predetermined character string is a pictogram or a face character. 4. The speech synthesizer according to claim 1, wherein said control means controls a prosody of said speech information based on prosody control information corresponding to said predetermined character string. 5. The control means comprises: voice pitch, voice intensity,
5. The speech synthesizer according to claim 1, wherein at least one of a voice length and a pause interval is controlled. 6. The speech synthesizer according to claim 1, further comprising a registration unit for registering said predetermined character string. 7. The apparatus according to claim 1, further comprising voice synthesis means for synthesizing voice information from said character information, wherein said voice synthesis means synthesizes voice information from character information excluding said predetermined character string. 7. The speech synthesizer according to any one of 1 to 6. 8. A method comprising: detecting a predetermined character string from character information; and controlling a prosody of voice information synthesized from the character information according to the predetermined character string. Speech synthesis method to be used. 9. The speech synthesis method according to claim 8, wherein the predetermined character string is a character string expressing a predetermined emotion. 10. The speech synthesis method according to claim 8, wherein the predetermined character string is a pictogram or a face character. 11. The speech synthesis method according to claim 8, wherein the control step controls a prosody of the speech information based on prosody control information corresponding to the predetermined character string. 12. The method according to claim 8, wherein the control step controls at least one of a pitch, a voice intensity, a voice length, and a pause interval. The speech synthesis method described in 1. 13. The speech synthesis method according to claim 8, further comprising a registration step of registering said predetermined character string. 14. A speech synthesizing step of synthesizing speech information from the character information, wherein the speech synthesizing step synthesizes speech information from character information excluding the predetermined character string. 14. The speech synthesis method according to any one of Items 8 to 13. 15. A computer-readable storage medium storing a program for executing the method according to any one of claims 8 to 14.