JP3340581B2 - Text-to-speech device and window system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a text-to-speech method for reading out text information, and more particularly to a text-to-speech method for expressing a text structure having a plurality of character types and layouts and a text position on a display screen.

[0002]

2. Description of the Related Art With the development of information devices such as computers and communication terminals, information has been digitized. Such electronic information is easy to handle in processing, storing, transmitting, presenting, and the like. In addition, recent advances in speech synthesis technology have made it possible to read out digitized text information by voice. Against this background,
2. Description of the Related Art Text-to-speech reading is used in proofreading and reading work of texts, acquiring information on visually impaired persons, and in situations where reading characters is difficult. In particular, in a portable information terminal or the like, since the display screen is small and difficult to see, and it is difficult to see the display while moving, reading out text by voice is effective.

On the other hand, in recent years, a multi-window system has been widely used as a terminal control method for a personal computer or a workstation. In such a system, a pointing device such as a mouse is used to select a window or an icon, select a command by operating a menu, search a file, and the like.
By reading out the window name, icon name, command type, file name, and the like by voice, the operation result of the window system can be easily understood. In particular,
For visually impaired persons who cannot obtain or have limited visual information, such reading aloud by voice is indispensable for operating the window system.

Conventionally proposed text-to-speech methods include, for example, J. Allen, MS Hunnicutt and D.
Klatt, "From text to speech: The MITalk system"
(Cambridge University Press, 1987). This is a method of analyzing text and converting it into an audio signal using rules. In order to determine the type of a sentence in a text, a method of changing the sound quality of voice according to the type of a sentence has been devised. As such a method, for example, a “sentence reading device” (Japanese Patent Laid-Open No. 62-21906)
6) As described, when a symbol such as a parenthesis or a quotation mark in a sentence is detected, a method of changing a sound quality of a reading voice so that a quotation part or the like can be identified, or a “speech output device” As described in (JP-A-56-101247), there is a method of assigning a reading attribute to each character in advance, and synthesizing a speech in accordance with the reading attribute.

[0005]

By using a bitmap display or the like, additional information such as character type, size, and character decoration is given to text information, and the layout of the entire document is structured and visually read. It is becoming possible to present information easily. However, if the display screen is small or the display cannot be viewed, the digitized text with additional information assuming such a display cannot be displayed in a plurality of character types, sizes, and character decorations by the conventional text-to-speech method. It is difficult to understand the structure of a text including character attributes such as paragraphs, columns, titles, etc., only by reading a monotonous voice. In addition, the names of windows and menu command names in the multi-window system are displayed at various positions on the screen. When such a name or the like is read out by voice, there is a problem that the display position cannot be determined by the conventional reading method.

An object of the present invention is to provide a text-to-speech method for expressing the structure of a text having such a layout by using voice and sound. It is another object of the present invention to provide a text-to-speech method for expressing a position on a screen of a reading target in operation of a window system or the like using voice or sound.

[0007]

Means are provided for classifying character attributes such as font, underline, and size of characters in a text, and a method is provided for reading out a text using synthesized speech having a different sound quality for each classification. Was.

In addition, the apparatus has a sound generating means for changing the type, sound quality, size and the like of a sound and outputting the same, and means for classifying character attributes such as font, underline and size of a character in a text. A method is provided for outputting sounds with different sound quality at the same time as reading out voices.

The apparatus further includes a voice synthesizing means for synthesizing voice from text, a sound generating means for generating a sound in association with text characters, and a sound output means for outputting voice and sound. According to the present invention, there is provided a method of reading out a target by voice, outputting a sound, or reading out a sound by voice and outputting a sound depending on the character attribute of text assigned to the target.

[0010] Further, a sound image localization means capable of presenting sound so as to be heard from an arbitrary direction, and mapping the coordinates of the character currently read out to the entire text in a visual display on an acoustic space. A position in the acoustic space corresponding to a relative position of the currently read out character in the text with respect to the entire text,
A method is provided for locating a voice signal or a sound signal assigned to the character.

[0011]

According to the present invention, there is provided a method of classifying character attributes such as font, underline, and size of a character in a text, and providing a method of reading out a text using a synthesized voice having a different sound quality for each classification. The attribute can be presented in the type of voice.

Also, there is provided sound output means for changing the sound quality and size of the sound and outputting the same, and means for classifying character attributes such as font, underline and size of characters in the text. By providing a method of outputting a sound having a sound quality at the same time as a reading voice, the attribute of the character can be presented as a sound.

[0013] The apparatus further includes voice synthesis means for synthesizing voice from text, sound generation means for generating sound in association with text characters, and sound output means for outputting voice and sound. Depending on the character attributes of the text assigned to the target, the target is read aloud, output a sound, or provided a method of reading the sound aloud and outputting a sound. It becomes possible to present differences in character attributes of texts and the like.

Further, the relative position of the character in the currently read text with respect to the entire text is mapped to the acoustic space, and a voice or sound can be heard from a direction corresponding to the relative position of the currently read character with respect to the entire text. By providing such a presentation means, it becomes possible to present the relative position of the currently read out character to the entire text.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows an embodiment of a hardware configuration for executing the method of the present invention. FIG. 1 has the configuration of a typical computer system having audio devices for voice and sound output. That is, a CPU 101 that performs an operation, a RAM 102 that temporarily holds a program and data for the operation and sequentially rewrites the data,
ROM 103 for storing a system boot program and the like, I / O controller 10 for controlling system input / output
4. It has a disk device 105 such as a magnetic disk for holding system programs, application programs, data, and the like, and these perform instructions and data transfer via a system bus 106. The I / O controller 104 also includes a display device 107 such as a CRT display, a pointing device 108 such as a mouse and a touch panel, an input device 109 such as a keyboard, a communication port 110 for data transmission and reception, a microphone, and an acoustic device 111 such as a speaker. Control via 112.

The standard control of such hardware is as follows.
This is generally performed by software called an operating system. The algorithm of the text-to-speech method according to the present invention can be realized as software executed on the hardware and the operating system.

The software for executing the algorithm of the present method is stored in the ROM 103 or the disk device 105.
2 is executed by the CPU 101.
Also, digitized text stored in the disk device 105 in advance, an input device 109, a communication port 110
The voice and sound generated by the text-to-speech method for the text input through the audio device 111 are output through the audio device 111. At this time, the pointing device 108, the input device 109, the communication port 11
0 may be used, or digitized text or other data may be displayed using a display device 107 such as a CRT display. However, these devices are not essential for the text-to-speech method of the present invention.

This embodiment can be realized by using, for example, a personal computer or a workstation having an audio output device.

FIG. 2 shows an embodiment of a hardware configuration for realizing the method of the present invention when performing stereophonic sound output. In FIG. 2, a sound or a sound using the stereophonic sound generated by using the present invention is
1 is output. As the three-dimensional sound device, for example, a commercially available high-speed operation board for creating a three-dimensional sound field and a plurality of speakers can be used.

Hereinafter, the function of each processing block of the algorithm of the text-to-speech method according to the present invention will be described.

First, a processing portion for realizing text-to-speech in the present invention will be described. FIG. 3 is a diagram illustrating a configuration of a speech rule synthesis processing block that changes sound quality according to character attributes. In FIG. 3, the input text is subjected to language processing in a language processing block 302 by detecting a character to be read in a reading position detection processing block 301,
In the prosody generation processing block 303, the prosody such as intonation and accent is determined. The phoneme generation processing block 304 generates a phoneme sequence from the language processing result. On the other hand, a character attribute extraction processing block 305 extracts a character attribute assigned to the character at the text-to-speech position of the text data. 307, a speaker table 308 associated with a speaker, and a level table 309 associated with a character size, to create sound quality control data. The character attribute is given to each character in advance to the digitized text having the character attribute for display or printing, and can be easily extracted by using this attribute data. The acoustic parameter generation processing block 310 uses the segment data stored in the segment dictionary 311 to generate the prosody data generated by the prosody generation processing block 303, the phoneme data generated by the phoneme generation processing block 304, and the sound quality. Control processing block 3
Acoustic parameters are actually generated from the sound quality data created in step 06, and a signal generation processing block 312 generates and outputs an audio signal.

A procedure for synthesizing reproduced audio information from text and obtaining an audio signal is described in, for example, J. Allen, MS Hunnicut.
t and D. Klatt, "From text to speech: The MITalk s
ystem "(Cambridge University Press ,, 1987).

However, as a method of obtaining voice from text, for example, a word or sentence of a text is associated with a voice signal that utters the word or sentence, and a so-called voice signal is output for the appearing word. It is also possible to use a recording / reproducing method.

FIG. 4 is an example of the tone color table in FIG. In FIG. 4, in order to change the tone of the voice for each character attribute by changing the fundamental frequency of the voice or performing processing such as low-pass filtering for each character attribute such as underline or bold, Different processing methods are associated.

FIG. 5 is an example of the speaker table in FIG. In FIG. 5, different speakers are associated with each font, and when the font changes, the speaker is changed to express the difference in the font. The speaker index in FIG. 5 indicates the number of the speaker data as shown in FIG. 6 included in the segment dictionary in FIG. 3, and the sound quality control processing block in FIG.
A speaker corresponding to the extracted font is selected from the segment dictionary. Further, as shown in FIG. 7, different speakers may be associated with character attributes, and differences in character attributes may be represented by differences in speakers.

FIG. 8 is an example of the level table in FIG. In FIG. 8, a difference in character size is expressed by associating a character size with an output level so that sounds of different levels are output depending on the character size. Note that the output level does not necessarily need to be equal to the difference in character size, and a level matching the difference in audibility may be selected.

Next, a processing portion for generating a sound from text in the algorithm of the present invention will be described. FIG.
FIG. 3 is a diagram illustrating a configuration of a sound generation processing block that outputs a sound having a different timbre for each character attribute. 9, a character attribute extraction processing block 901 performs a reading position detection processing block 902 on an input text.
The character attribute is extracted from the input text in the same manner as in FIG. The sound quality control processing block 903 uses a sound table 904 associated with the character attributes in advance from the extracted character attributes and a level table 905 associated with the character size as shown in FIG. Create sound quality control data. The sound parameter generation processing block 906 selects a sound associated with the character attribute from the sound data 907 from the sound table 904, and generates a sound parameter whose level has been changed. The signal generation processing block 908 generates a sound signal from the acoustic parameters.

FIG. 10 is an example of the sound table in FIG. A different sound is associated with each character attribute, and a different sound is output for each character attribute to express a difference in the character attribute.

FIG. 11 is another example of the sound table in FIG. In this example, pure sounds, instrument sounds, band noises, and the like are associated as sounds, and the timbre can be changed by changing the fundamental frequency or the center frequency. In this case, for example, if a known input / output interface that allows the user to change the timbre is prepared, the timbre of each sound can be changed according to the user's preference.

Also, the sound quality processing method shown in FIG. 4, the speakers shown in FIGS. 5 and 7, the levels shown in FIG. 8, and the sounds shown in FIGS. 10 and 11 can be selected according to the user's preference.
You may make it possible to newly register.

Here, a description will be given of a process of generating and outputting voice and sound from input text, which realizes the algorithm of the text-to-speech method according to the present invention. FIG. 12 is an example of a sound and sound output control process. 3 and 9 for the input text.
The voice and sound generated by the voice rule synthesis processing block 1201 and the sound generation processing block 1202 as shown in (1) are synchronously controlled in a synchronization control processing block 1203, and are simultaneously output corresponding to each character. However, the speech rule synthesis processing block 1201 includes, for example, J. Allen, MS Hunnicutt and D. Klatt, "From
text to speech: The MITalk system "(Cambridge Uni
(versity Press, 1987), a procedure based on a conventional speech rule synthesis method that does not consider character attributes may be used.

FIG. 13 shows an example of a stereophonic sound generation process for realizing the text-to-speech method according to the present invention using the stereophonic sound generation means. If the digitized text expresses a visual change by changing the character display position in various ways, such as columns, sentence alignment, line spacing, and indentation, as in a structured document, a three-dimensional By using the sound generation means, it is possible to express such a change in the display position. In FIG. 13, a character coordinate extraction processing block 1 is applied to an input text.
In step 301, for the character at the reading position in the text detected by the reading position detection processing block 1302, the relative coordinates of the input digitized text in the display area of the currently read character are extracted. Since digitized texts with various character positions set for display or printing have information such as left and right margins, step widths, and line spacing in advance, character coordinates can be easily extracted by using this position data. Can be. The sound image control processing block 1303 performs control for mapping a sound image to a corresponding position in the three-dimensional sound space based on the extracted character coordinates. The three-dimensional sound generation processing block 1304 generates three-dimensional sound such that the sound or sound generated in the sound / sound output control processing block 1305 shown in FIG. 12 is localized at a position in the sound space corresponding to the read-out character position. I do. The signal generation processing block 1306 generates and outputs a three-dimensional sound or a sound signal that can be heard from a three-dimensional spatial position. The sound image control and the stereophonic sound generation method are realized using, for example, the spatial sound image localization technology described in Komiyama, “Sound Image Control Technology” (Journal of the Institute of Television Engineers of Japan, Vol. 46, No. 9, 1076-1079, 1992). It is possible.

The text input means for executing the processing algorithm for realizing the text-to-speech method according to the present invention includes the digitized text previously stored in the disk device 105 described in FIG. A known method of loading text input via the input device 109 and the communication port 110 can be used.
In addition, voices and sounds generated by this text-to-speech method, or voices and sounds using three-dimensional sound,
Output can be performed through the acoustic device 111 or the stereophonic device 201 using a known method.

FIG. 14 shows an example of an electronic text document having character attributes, columns, and the like. Character attributes such as character size, font, italics, and underline differ depending on the title, chapter title, and text. In addition, the position of the text is set by column or paragraph division. When a table or a figure is included, the text is usually arranged so as to avoid the table or the figure. Therefore, the arrangement of the text differs depending on such conditions. The figure shows that the position of (x1, y1) is currently read out in the display coordinate system. The sound output is localized at the position in the acoustic space corresponding to the reading position (x1, y1) using the method described with reference to FIG. 13, so that the structure of the table and the position of the item can be determined. Become.

FIG. 15 shows the correspondence between the characters of the document shown in FIG. 14 and the character attributes. Based on such correspondence, the character attribute of the character to be read is extracted in the character attribute extraction processing block in FIGS.

FIG. 16 shows an example in which a display position is associated with an acoustic space. The figure shows a display space 1601 such as a CRT display device and a user 1602 in a bird's-eye view from obliquely behind. Text 16 to be read out in display space 1601
The coordinates (x1, y1) of 03 are associated with the coordinates (u, v) 1605 of the acoustic space 1604, and the sound or sound cracked to the character is localized on the acoustic space.

Here, if the acoustic space is arranged so as to completely overlap the display space, the position of the object in the display space and the position of the sound image in the acoustic space completely match. When the acoustic space is made larger than the display space, an effect of relatively improving the resolution of the sound image localization can be obtained. Also, as shown in FIG. 17, for example, as shown in FIG. 17, the center of the user's both ears is set as the origin 1701 and the origin and the coordinates of the reading position 1703 on the display space 1702 are connected by a straight line. This can be done by assigning the coordinates of the sound image 1705 corresponding to the object in the acoustic space 1704 to the equidistant position on the extension. Note that the mapping to the acoustic space may be modified so as to match the auditory sense according to the characteristics of the auditory sense. In FIG. 17, a two-dimensional space has been described as an example, but it goes without saying that this association can be easily extended to a three-dimensional space.

FIG. 18 is an example of a table of digitized text. In the case of a table, the characters, the number of lines, the number of columns, and the like in the table are read out in a predetermined order. At this time, the sound quality or the speaker of the voice is changed or the sound is output according to the character attribute. Further, by localizing the sound output to the position in the acoustic space corresponding to the reading position (x2, y2) using the method described with reference to FIG. 13, it is possible to determine the structure of the table and the position of the item. It becomes possible. The same method as in the case of the document in FIG. 14 can be used for associating the display position with the acoustic space.

FIG. 19 shows the correspondence between the characters and the character attributes in the table shown in FIG. Based on such correspondence, the character attribute of the character to be read is extracted in the character attribute extraction processing block in FIGS.

FIG. 20 is an example of a digitized text diagram. In the case of a figure as well, a character in the figure or a voice or the like previously associated as a shape of the figure is read out in a predetermined order. Is output. Further, the output of the sound is read at the reading position (x3, y) using the method described with reference to FIG.
By localizing to the position in the acoustic space corresponding to 3), it is possible to determine the structure of the figure and the position of the text in the figure. The same method as in the case of the document in FIG. 14 can be used for associating the display position with the acoustic space.

FIG. 21 shows the correspondence between the characters and the character attributes in the diagram shown in FIG. Based on such correspondence, the character attribute of the character to be read is extracted in the character attribute extraction processing block in FIGS.

At the time of reading, a portion where the reading position changes rapidly, such as a line change, a column change, or a page change appearing in the middle of the text, is detected, and a specific sound is output at that time. In this case, it is possible to determine the change of each row, column, page, and the like based on the type of sound. Also, when reading out the next column beyond the ruled line when reading out the table, if a specific sound is output, the movement of the column can be determined according to the type of sound.

FIG. 22 shows an example in which the difference between pages of a document is expressed by stereophonic sound. As shown in FIG. 22, the arrangement of one page of the document is associated with the X-Y coordinate and the number of pages is associated with the Z coordinate, and the position where the character 2201 to be read is heard by the page where the character 2201 is to be read is moved back and forth, thereby reading out the document at present. The page can be determined. In the example shown,
The sound image is localized so that two pages are more than one page and three pages are more than two pages.

Further, as shown in FIG.
The position where the sound image can be heard can be adjusted by detecting the position of the sound image with the position sensor 2302 and moving the sound image in accordance with the movement of the position of the user with respect to the position of the character 2303 to be read. At that time, the moving distance of the user and the moving distance of the sound image may be intentionally changed. For example, if the moving distance of the sound image is relatively larger than the moving distance of the user, the user can be heard from near by a loud sound when the user performs an operation such as leaning out, so that the user is alerted. be able to.

FIG. 24 shows an example of reading aloud by the window system. By outputting voices and sounds in response to system operations such as menu operations and commands, files, windows and icons of the window system, and movement of the icons, the operation results of the window system can be easily understood. Here, 2401, 2402, 24
03, 2404, and 2405 indicate a window, an icon, a menu, a clock, and a cursor, respectively. 24
Reference numeral 06 denotes a file manager that searches for a file or executes an application by selecting a folder or a file with a cursor or the like. Also, 24
07 represents a window name. Now menu 24
The “save” command in 03 is selected. At this time,
By locating the voice or sound corresponding to the text assigned to the command at the position in the acoustic space corresponding to the coordinate position (x4, y4) in the display space of the command, the position or the content of the command is determined. can do.

FIG. 25 shows an example of correspondence between the components of the window system and voices and sounds. For example, when a window is displayed, a melody associated with the window is output, and when the window is selected, the window name is read out simultaneously with the output of the melody. As for the menu, a click sound is output when the cursor is moved on the menu, and the command name at the position where the cursor is stopped is read out when the cursor is stopped. For a clock, the sound of a wall clock is output when the cursor stops on the clock, and the time is read out when selected. In the file manager, a percussion sound is read out when a cursor passes through the file, and a file name is read out when the cursor stops. For the folder, the sound is changed between when the cursor passes and when the folder is selected. Such association makes it possible to determine the contents, positions, and states of the components of the window system.

FIG. 26 is a diagram of World-Wide-Web (an information service using hypertext technology developed by the European Institute for Particle Physics (CEERN)) and Gopher (a computer network resource developed at the University of Minnesota). This is an example in which hypertext having a hierarchical structure of text is expressed by stereophonic sound, as in a menu-type system for searching for. In the hypertext, a call point called an anchor is prepared on the text. For example, when the anchor is selected with a mouse or the like, another text, a figure, a moving image, a sound, and the like previously associated with the anchor are displayed. Called and displayed. The association can be performed not only on the same computer but also by linking to various access points via a network. In FIG. 26, 2601 denotes the uppermost hypertext, and 2602 denotes an anchor on the hypertext. When the user selects the anchor 2602, a new text, figure, moving image, sound, or the like is normally displayed on the screen. Reference numeral 2603 denotes a second hypertext newly displayed in this manner. 2604 is an anchor on the second hypertext 2603, 260
5 represents the third hypertext opened by the selection of the anchor. When reading such digitized text, using the present invention, by changing the method of reading and the sound to be given in the text and the anchor,
The presence of the anchor can be determined. In addition, the hypertext hierarchy is associated with the depth direction of the acoustic space, and when a user selects an anchor, a newly opened speech of the hypertext, a sound added to the text, or a sound linked to the anchor. For example, by setting the XY coordinates of the anchor on the hypertext as the center of the new hypertext or sound on the XY plane, so that the hypertext can be heard from the near side in the Z-axis direction corresponding to the depth, Can be determined.

[0049]

According to the present invention, by reading out a text using a synthesized voice having a different sound quality for each classification of a character attribute such as font, underline, and size of the character in the text, the character attribute can be presented in the type of voice. Became.

Further, by outputting a sound having a different sound quality for each character attribute classification such as font, underline, and size of a character in the text simultaneously with the reading voice,
Character attributes can now be presented as sounds.

Depending on the position and state of the target, the character attribute of the text assigned to the target, and the like, the target can be read aloud by voice, output a sound, or read aloud by sound and output a sound to obtain the position and the target of the target. It is now possible to present the state, the character attribute difference of the text assigned to the object, and the like.

Further, the relative position of the character in the currently read text with respect to the whole text is mapped to the acoustic space, and the voice or sound can be heard from the direction corresponding to the relative position of the currently read character with respect to the whole text. By providing the presenting means, it is possible to present the relative position of the currently read out character to the entire text.

[0053]

[Brief description of the drawings]

FIG. 1 shows an embodiment of a hardware configuration for realizing the method of the present invention.

FIG. 2 is a diagram illustrating an example of a hardware configuration when performing stereophonic sound output.

FIG. 3 is a diagram illustrating an example of a speech rule synthesis process.

FIG. 4 is a diagram illustrating an example of a tone color table.

FIG. 5 is a diagram illustrating an example of a speaker table.

FIG. 6 is a diagram illustrating an example of speaker data.

FIG. 7 is a diagram illustrating an example of a speaker table when a different speaker is associated with each character attribute.

FIG. 8 is a diagram illustrating an example of a level table.

FIG. 9 is a diagram illustrating an example of a sound generation process.

FIG. 10 is a diagram illustrating an example of a sound table.

FIG. 11 is another example of a sound table.

FIG. 12 is a diagram illustrating an example of audio and sound output control processing.

FIG. 13 is a diagram illustrating an example of stereophonic sound generation processing for generating stereophonic sound.

FIG. 14 is a diagram illustrating an example of an digitized text document having character attributes, columns, and the like.

FIG. 15 is a diagram illustrating an example showing correspondence between characters of a document and character attributes.

FIG. 16 is a diagram illustrating an example in which a sound image at a character position in a display space is localized in an acoustic space.

FIG. 17 is a diagram illustrating an example of associating a display space with an acoustic space.

FIG. 18 is a diagram illustrating an example of a table of digitized text.

FIG. 19 is a diagram illustrating an example showing correspondence between characters in a table and character attributes.

FIG. 20 is a diagram illustrating an example of a digitized text diagram.

FIG. 21 is a diagram illustrating an example showing correspondence between characters in the figure and character attributes.

FIG. 22 is a diagram illustrating an example in which a difference between pages of a document is expressed by stereophonic sound.

FIG. 23 is a diagram illustrating an example in which a difference in user position is expressed by stereophonic sound.

FIG. 24 is a diagram illustrating a reading example of a window system.

FIG. 25 is a diagram illustrating an example of audio / sound correspondence of window system components.

FIG. 26 is a diagram illustrating an example in which hypertext is expressed by stereophonic sound.

[Explanation of symbols]

106: system bus, 112: bus, 1601: display space, 1602: user, 1603: characters to be read out,
1604: acoustic space, 160: coordinates, 1701: origin,
1702: display space, 1703: reading position, 170
4 acoustic space, 1705 sound image, 2301 user, 2
302 ... position sensor, 2303 ... character to read out, 24
01 window, 2402 icon, 2403 menu, 2404 clock, 2405 cursor, 240
6 File manager, 2406 Window name, 2
601, 2603, 2605... Hypertext, 26
02, 2604 ... anchor.

──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Hideaki Kikuchi 1-280 Higashi Koikekubo, Kokubunji-shi, Tokyo Inside the Central Research Laboratory, Hitachi, Ltd. (56) References JP-A-7-72875 (JP, A) JP-A-5-197517 (JP, A) JP-A-5-80958 (JP, A) JP-A-5-81294 (JP, A) JP-A-6-259216 (JP, A) JP-A-7-129356 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06F 3/16 G06F 3 / 00 G10L 11/00 G10L 13/06

Claims (4)

(57) [Claims] And means for reading out a text composed of a plurality of characters by voice, and layout information associated with the text to be read out.
From the text data of the character data read out above
Means for extracting position relative to the whole, and mapping the sound to a position corresponding to the position in acoustic space
A sound image localization device, the voice of the reading means in the sound image localization position is localized to
Means for generating stereophonic sound as described above . 2. A has a hierarchical structure, other text included in another layer, FIG, video, or sound and a voice hypertext including during call <br/> beauty out points and the associated text means for reading out, the sound image localization means capable of outputting audibly the audio from a plurality of locations, and reading how the body, the sound image localization means so made different and reading how the call point A text-to-speech apparatus, comprising: means for controlling. 3. A text-to-speech apparatus according to claim 2 , wherein said sound image localization means has means for outputting an acoustic signal having a depth, and wherein the hierarchy of said hypertext is made to correspond to said depth. 4. A window system capable of displaying a plurality of windows on a screen and displaying information in each window, means for reading out text by voice, sound output means for outputting sound, and said screen A cursor that can be moved on the screen, and that can select an object displayed on the screen, and each window has an area in which a command name is displayed. The sound output means outputs a sound when the cursor passes through the area where the command name is displayed. The text-to-speech means outputs the sound when the movement of the cursor is stopped in the area where the command name is displayed. In
A window system characterized by reading out the command name.