JPH09265299A - Text reading device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To freely control the reading operation in the text reading device. SOLUTION: In a display input device 2, a transparent touch panel (an input device 12) is integrally constituted on the screen of a display (a display device 10). Using the device 2, an operator performs a tracing operation of the device 12 along the character columns displayed on the device 10 by a pointer such as a finger. Input signals 14 obtained from the device 12 based on the tracing operation provide position data 50, moving speed data 52, pressure data 54 of the pointer at an input signal processing section 42 and these data are made to correspond to the parameters including a reading corresponding clause, an uttering speed and a volume (or an accent strength). A rule synthesis section 8 reflects the parameters representing the uttering speed and the volume against the specified clause and conducts the synthesis so that synthesized voices which reflect the thinking of the operator and easily understood, are easily obtained.

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 device having a machine interface.

[0002]

2. Description of the Related Art Conventionally, a text-to-speech apparatus has been used for editing work for detecting and correcting errors such as typographical errors and omissions in texts in business such as printing and publishing. For example, Japanese Unexamined Patent Publication No. 59-127148 discloses a device for reading out text for checking and correcting input text.

The conventional device is operated by using an input device such as a keyboard and a mouse provided separately from the display device. For example, in addition to inputting characters, you can use the keyboard or the mouse to set conditions such as the speech speed of the voice output from the device, the volume and strength of the voice, skip reading of sentences, backtracking, and repeating. Is operated to move the cursor on the screen to a predetermined position. In other words, the specified text range is basically unilaterally read under the above constant conditions, and it is necessary to operate the input device such as the keyboard separately in order to stop or change the reading conditions. There is.

Further, the prosodic characteristics of a sentence are variously changed depending on the word structure, sentence pattern, and the like. Therefore, the rule synthesis method is suitable for reading aloud with natural voice. The rule synthesizing method is an advanced synthesizing method for expressing the prosodic features of a sentence by rules. By the way, a voice can be captured by being decomposed into three components, a timbre, a height of a fundamental frequency, and an intensity. That is, the voice has a frequency spectrum composed of the sound wave component of the fundamental frequency and its multiplied frequency, and the timbre depends on the envelope shape of this frequency spectrum. The pitch is this fundamental frequency and determines the pitch of the pitch. The time variation of the pitch is represented by the pitch pattern. The intensity is captured as the amplitude of the voice.

[0005]

However, since the synthetic speech according to the rule-based synthesis method is generally less intelligible than the natural speech, the user has a great deal of difficulty in listening to the synthetic speech unilaterally read by the device for a long time. There was a problem that it required concentration and suffering. In order to improve the intelligibility of the synthesized voice, it is desirable that the user can arbitrarily start / stop the reading operation and change the condition. However, it is difficult to operate these with the above-mentioned conventional input device, or a high degree of skill is required. Therefore, there is a problem that it is extremely difficult for a visually impaired user or a user who is unfamiliar with computer operation to operate the text-to-speech device.

The present invention solves the above problems, provides a text-to-speech device having a man-machine interface that can be easily operated, and realizes work efficiency in conventional applications such as text proofreading support. It is an object of the present invention to provide a device for assisting reading or the like of a visually impaired person who has difficulty reading the displayed text.

[0007]

A text-to-speech device according to the present invention includes display input means for displaying text data as a character string on a screen and outputting position information representing a contact point of a pointing object on the screen. Trace specifying means for specifying the character string portion traced by the pointing object based on the position information, voice synthesizing means for synthesizing the voice corresponding to the character string portion based on the text data, and the synthesized voice. And an audio output unit for outputting.

According to the present invention, when the character string displayed on the screen of the display input means is traced, the display input means detects the traced position and outputs it as position information. The position information is associated with the text data, and the traced portion of the character string is voice-synthesized and output as voice. The pointing object tracing the screen is, for example, a finger of the operator or a pointer such as a pen. The display input means detects the position pointed by the pointing object by, for example, contact pressure received from a finger, electrostatic interaction, optical interaction, or the like. The display input means can be realized using, for example, a touch panel. The act of directly acting on the target character string, which is "tracing," is a natural act that we do every day, and the operator can freely control the device.

In the text-to-speech apparatus according to the present invention, the voice synthesizing unit determines a moving speed of the contact point based on the position information, and a speech utterance speed of the voice according to the moving speed. And an utterance speed control means for controlling the utterance speed. According to the present invention, the operator can naturally obtain the utterance speed desired by himself by adjusting the moving speed of the pointing object, and the intelligibility of the output voice is enhanced.

In the text-to-speech device according to the present invention, the display input means outputs pressure information representing the pressure received from the pointing object, and the voice synthesis means controls the volume of the voice according to the pressure signal. It has a volume control means. According to the present invention, the display input means can detect the pressure received from the pointing object. The operator can control the strength of the synthesized voice by adjusting the pressure on the screen by the pointing object, and the intelligibility of the output voice is enhanced.

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION A text-to-speech device according to an embodiment of the present invention will be described below with reference to the drawings. This text-to-speech device has a display input device in which a display device for displaying text and an input device for detecting pointing are integrated, and when a character string displayed on the device is traced with a finger or the like, the tracing operation is performed. It has a basic function of reading out the character string of the part indicated by voice synthesis. FIG. 1 is a principle diagram for explaining this basic function.
Here, a transparent touch panel as a display input device is integrally configured on the screen of the display, and the screen is shown. The character string is displayed on multiple lines on the display screen. When the line with the character string is traced horizontally rightward while the finger 1 is in contact with the transparent touch panel on the display screen, the device outputs the synthesized voice corresponding to the traced character string. The specific configuration of the device that realizes this function will be described below.

FIG. 2 is a block diagram of the text-to-speech device according to this embodiment. The present apparatus includes four blocks of a display input device 2, a reading interface driver 4, a language processing unit 6, and a rule synthesizing unit 8.

The display input device 2 is a display device 10 for displaying the text data output from the reading interface driver 4, and an input device 1 configured integrally with the display device 10 for detecting a contact point on the screen by the pointing object.
2 is provided. Specifically, this apparatus uses a display having a transparent touch panel on the screen as the display input device 2. That is, the display is the display device 10 and the touch panel is the input device 12.
The touch panel outputs an input signal 14 representing position information of a point touched by a pointing object (pointer) such as a finger or a pen and contact pressure information thereof. The display input device 2 may be a device using a tablet instead of the touch panel as an input device. Incidentally, this tablet is a device called a stylus that detects a position pointed by an input pen by an electromagnetic method or an optical method.

The reading interface driver 4 is located between the display input device 2, the language processing unit 6 and the rule synthesizing unit 8 and performs a group of processes such as data conversion and calculation among them. The basic data in this device is the text data stored in the sentence buffer 20. This text data is input from an input device such as a keyboard (not shown) or read from a recording medium in which electronic text data is recorded. The text data in the text buffer 20 is used by the language processing unit 6 and the display processing unit 22.

The language processing unit 6 linguistically analyzes the Japanese text from the sentence buffer 20 and, based on the information representing the "reading" of the Japanese text, the boundary of the phrase, the position of the accent, and the pause necessary for the speech synthesis processing. A phonological symbol string to which prosodic information such as the position of vowel and vowel devoicing is added is generated. This phoneme symbol string is stored in the phoneme symbol string buffer 24.

The rule synthesizing unit 8 performs rule synthesizing of voices based on the parameters output from the reading interface driver 4. Details will be described later.

Next, the reading interface driver 4 will be described. The display processing unit 22 performs output processing for the display input device 2. That is, the display processing unit 22 reads text data corresponding to, for example, one screen of the display device 10 from the text buffer 20, converts the text data into text screen data 26, and outputs the text screen data 26 to the display device 10. The display processing unit 22
Outputs display range information 28 indicating which part of the sentence buffer 20 is displayed on the display 10 to the phonological symbol output processing unit 40, and the phonological symbol output processing unit 40, which will be described later, performs the trace identification processing. It may be performed using information.

A position determiner 44 in the input signal processing section 42,
The speed determiner 46 and the pressure determiner 48 are the display input device 2
In response to the input signal 14 from, the contact point position data 50, the moving speed data 52 and the pressure data 54 on the display 10 are obtained and output. That is,
The coordinates and pressure value of the contact point on the screen are extracted from the position information and the contact pressure information included in the input signal 14, and the moving speed is calculated by the time change of the coordinates. The position determiner 44 may output the coordinates of the pointer itself as the position data 50, but in the present device, the contact of the pointer with the screen is detected from the contact pressure information, and while the pointer is continuously in contact, Only the displacement component in the row direction (horizontal direction) is extracted from the position information, and this is output as position data 50. That is, even if the movement of the pointer has a vertical displacement component, the vertical coordinate of the position data 50 remains fixed and does not change. This is because it is difficult for humans to accurately trace a fine character string without deviating from it, and this function is particularly effective when a visually handicapped person or the elderly operates. The vertical coordinate is changed by temporarily moving the pointer away from the screen and touching another line. The input signal processing unit 42
May notify the operator of the deviation from the line of the pointer by an alarm or by fading out a synthetic voice generated from the traced character string. When a button is provided on the screen and an operation is performed by double-clicking the button, as will be described later, the position data 50, the pressure data 54, etc. output from the input signal processing unit 42 are not shown in the text reading device. It is passed to the control unit and used for determining whether or not a button has been operated.

The position data 50 is input to the phoneme symbol output processing section 40. The phonological symbol output processing unit 40 identifies, from the position data 50, which position in the text data given to the sentence buffer 20 is the character string part traced (traced) by the pointer on the display 10. To do. At this time, the phonological symbol output processing unit 40 obtains information such as the display range information 28 from the display processing unit 22 or specifies the display range given to the display processing unit 22 from the text reading device control unit (not shown), if necessary. It is configured to obtain information. The phoneme symbol string corresponding to the traced phrase is fetched from the phoneme symbol string buffer 24 and output to the rule synthesizing unit 8. Further, in this device, the phonological symbol output processing unit 40 outputs the timing signal 60 for moving the contact point from one phrase to the next phrase to the speech rate control unit 62.

When the pointer is released from the screen, the output of the position data 50 from the input signal processing section 42 is stopped. In response to this, the output of the phoneme symbol string from the phoneme symbol output processing unit 40 is also stopped, and the speech synthesis in the rule synthesizing unit 8 using this is also interrupted. By the way, a skipping operation is realized by moving a pointer tracing a line of a character string on a screen away from the screen and touching another line. Further, when the pointer is moved in the opposite direction, that is, leftward from a certain phrase to the preceding phrase, the phonological symbol output processing unit 40.
Outputs a phoneme symbol string of a certain phrase and subsequently a phoneme symbol string of the preceding phrase. That is, backward reading is performed in phrase units by moving the pointer in the backward direction. The reverse reading of the phrase unit is easier to understand the meaning than the reverse reading of the character unit, and the backward movement operation for re-reading is easy.

The speech speed control unit 62 simply associates the speech speed in the traced character string portion with the moving speed of the pointer at that position. However, it is not always easy to smoothly control the moving speed of the pointer, and usually, the moving speed is moderated regardless of the character string. In particular, when the pointer is strongly pressed against the screen, friction on the screen is increased, and a large speed change is likely to occur in a short cycle. If the speed irrelevant to the character string is directly reflected in the speech rate, the synthesized voice tends to be difficult to hear. Therefore, in the present apparatus, the phrase movement timing signal 60 is obtained from the phonological symbol processing unit 40, and the utterance speed is changed between phrases using this, and the utterance speed within the phrase is controlled to be constant. That is, the speech speed of the next phrase is determined according to the moving speed data 52 at the moment when the timing signal 60 is obtained,
The speech rate parameter is output to the rule synthesizing unit 8. Further, in order to cancel the influence of a large speed fluctuation of a short cycle which is offensive to the ears, it is also effective to use a speed averaged over an appropriate time width.

The volume controller 64 determines the volume of the phrase pointed to based on the pressure data 54, and outputs it to the rule synthesizer 8 as a volume parameter. In this device, the volume is adjusted by using the contact pressure information in the pointing signal 14, but a structure in which the strength of the accent in the phrase is adjusted in place of the volume is also possible.

The rule synthesizing unit 8 includes a phoneme symbol output processing unit 4
The phoneme symbol string from 0, the speech rate parameter from the speech rate changing unit 62, and the volume parameter from the volume control unit 64 are input. This rule synthesizing unit 8 is different from the conventional one in that its input includes a speech rate parameter and a volume parameter which can be arbitrarily changed by the operator by operating a pointer.

In the processing of a normal speech synthesizer, first, a phonological symbol string is generated by adding a phonological duration time based on a constant speech rate to the result of linguistic analysis of text data.
In other words, the speech rate is initially given as a parameter and is fixed.To change it, you need to operate the keyboard, etc., and freely change it at the right time while reading the text. Was difficult. Also, the variation range of the sound source amplitude pattern generated based on the phoneme symbol string is usually determined by a parameter representing the initially set sound volume, and like the speech rate, it is difficult to freely change this.

On the other hand, the rule synthesizing unit 8 of the present apparatus can freely change the phoneme duration for each phrase by using the speech rate parameter that can be changed for each phrase by the pointer operation. Next, based on the phonological symbol sequence, a sound source amplitude pattern that represents a change in voice intensity, a pitch pattern that represents a change in fundamental frequency, and a spectrum pattern (formant pattern) that is an envelope shape of a frequency spectrum that represents a timbre,
Each is generated based on a rule or the like, and these are synthesized to perform a process of producing a synthesized voice. At this time, in this apparatus, the fluctuation range of the sound source amplitude pattern can be freely changed by using the volume parameter that can be arbitrarily changed by the pointer operation.

As described above, according to this apparatus, the operator can obtain the voice output by an extremely simple operation of tracing the character string with the pointer. That is, even a visually impaired person or a person unfamiliar with machine operation can operate the apparatus. Also, the operator can freely adjust the speaking speed and volume (or accent strength) according to the degree of intelligibility, and can perform the backward movement operation and skip reading operation, so that a synthetic speech with high intelligibility can be obtained. Therefore, if this device is used to support the grammar correction work, its work efficiency is improved.

By the way, at present, it is not always possible to obtain 100% correct "reading" and "accent" by the rule composition described above. Therefore, this device has a phrase correction mode. In the phrase correction mode, the device presents pronunciation candidates by synthetic speech to the phrase for which the operator has detected “misreading” or “accent mistake” of the device, and the operator selects the correct pronunciation. This is a function that allows the device to learn this. In this device, the phrase correction mode is activated, for example, when the operator double-clicks a phrase on the screen for which the operator wants to correct the pronunciation of the device with the pointer. The device audibly informs the operator that the phrase modification mode has been activated. In addition, it may be displayed on the screen. The device sequentially presents the reading candidates for the specified phrase with synthetic speech. At this time, for example, the screen is roughly divided into three areas, and each area serves as a contact switch for outputting the synthesized voice of the next candidate, the current candidate, and the previous candidate. The contact switch is operated to select a candidate, and the device is made to sound. When the candidates are exhausted, the device notifies the operator by the screen and the synthesized voice respectively. When the operator finds a candidate that he / she intends, the candidate will be able to double-click the contact switch provided at an easy-to-understand position on the screen, for example, at the top, bottom, or corner of the screen. Is confirmed as the reading of. When the reading is confirmed,
In a similar manner, the accent determination routine is entered.
When the pronunciation or accent is corrected in this way, the content of the phoneme symbol string buffer 24 or the data of the language processing unit 6 is changed, and thereafter, the phrase is speech-synthesized with the corrected reading and accent.

As described above, this device can correct the pronunciation of the phrase as described above on the screen. This operation
It can be done with a simple operation using the pointer that was tracing the character string on the screen until then. That is, the operation of switching from the pointer to another input means, such as a keyboard or a mouse, is unnecessary, so that the work efficiency is good. In particular, it is very convenient for the visually impaired person not to have such a transition of operation, and the contact switch set on the screen in the phrase correction mode is also used by the operator. The position can be specified roughly by a large one, or it is provided at the end of the screen where the position can be easily grasped, and the operability is high compared to other input means such as a commonly used keyboard.

[0029]

According to the text-to-speech device of the present invention, the operation of the device which has conventionally required operation from the keyboard or the mouse can be performed by a pointer such as a finger or a pen on the screen on which the read-out character string is displayed. Position, movement,
The pressure has an effect that it can be performed by a natural operation that is extremely intuitive. In particular, such an easy man-machine interface is effective for a visually impaired user or a user unfamiliar with computer operation. Further, the reading and accent errors that occur in voice synthesis can be corrected by operating the pointer on the screen. At this time, since candidates for reading and accent are presented in synthetic speech, even a general user who does not have linguistic or phonetic knowledge can easily make corrections by selecting from the candidates. be able to.

[Brief description of drawings]

FIG. 1 is a principle diagram illustrating basic functions of a text-to-speech device according to the present embodiment.

FIG. 2 is a block configuration diagram of a text-to-speech device according to the present embodiment.

[Explanation of symbols]

2 display input device, 4 reading interface driver, 6 language processing unit, 8 rule synthesizing unit, 10 display unit, 12 input unit, 14 input signal, 20 sentence buffer, 24 phoneme symbol string buffer, 40 phoneme symbol output processing unit, 42 Input signal processor, 44 Position determiner, 46
Speed judger, 48 Pressure judger, 50 Position data, 5
2 movement speed data, 54 pressure data, 62 speech speed control section, 64 volume control section.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G10L 3/00 G06F 15/20 568A

Claims (3)

[Claims] 1. Display input means for displaying text data as a character string on a screen and outputting position information indicating a contact point of the pointing object on the screen, and tracing by the pointing object based on the position information. A trace specifying means for specifying the character string portion, a voice synthesizing means for synthesizing the voice corresponding to the character string portion based on the text data, and a voice output means for outputting the synthesized voice. Characteristic text-to-speech device. 2. The text-to-speech device according to claim 1, wherein the voice synthesizing unit determines a moving speed of the contact point based on a change with time of the position information, and a speed detecting unit that determines the moving speed according to the moving speed. A text-to-speech device, comprising: a speech rate control means for controlling a speech rate of the voice. 3. The text-to-speech device according to claim 1, wherein the display input unit outputs pressure information indicating pressure received from the pointing object together with the position information, and the voice synthesis unit outputs the pressure signal to the pressure signal. A text-to-speech device having volume control means for controlling the volume of the voice according to the above.