JP2005309173A - Speech synthesis controller, method thereof and program thereof, and data generating device for speech synthesis - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a speech synthesis controller capable of easily synchronizing a synthesized speech with a signal to be synchronized such as a video signal at specified display time. <P>SOLUTION: The speech synthesis controller 1 acquires an utterance start position indicating the start position of speech synthesis of a text to be read aloud and utterance start time as time when utterance at the utterance start position is started by an utterance start timing acquisition part 13 when the text to be read aloud is uttered through speech synthesis, and outputs the utterance start timing obtained by the utterance start timing acquisition part 13 to a speech synthesizer 7 by a speech synthesis start control part 15 to control the start of speech synthesis by the speech synthesizer 7. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a speech synthesis control device for controlling a speech synthesizer so as to synthesize text to be read out and uttered as synthesized speech at a predetermined utterance timing, its method, its program, and control of the speech synthesizer The present invention relates to a speech synthesis data generation device that generates speech synthesis data.

Recently, a speech synthesizer has been developed that generates text to be read out as synthesized speech that is synthesized, and outputs the synthesized speech and a moving image in synchronization.
For example, this type of speech synthesizer analyzes the lip shape displayed in a moving image and its movement, acquires the lip movement start time and the duration of lip movement as lip information, and Which synchronizes the synthesized voice and the video (the movement of the lip shape) by synthesizing the estimated character in the text and synthesizing the estimated character to fit the lip shape to generate a synthesized speech ( A synchronization system between a moving image and a text / speech converter is known (for example, see Patent Document 1).

  In the synchronization system between the moving image and the text / speech converter described in Patent Document 1, the duration of each phoneme during speech synthesis is adjusted according to the lip shape, and the text and the moving image are referred to by referring to the lip shape. In order to synchronize with each other, the lip shape to be assigned to each phoneme is estimated from the phoneme-specific articulation characteristics such as the articulation location and articulation method of each phoneme according to the phoneme string and the duration of the phoneme. In this synchronization system, the estimated lip shape and the lip shape in the moving image (video) are compared with the result of arranging them on the time axis according to the position information and the duration, and the degree of approximation between them is maximum. The synthesized speech is generated by adjusting the duration for each phoneme so that the synthesized speech is synchronized with the moving image.

  Also, this type of speech synthesizer synthesizes speech for each text of a predetermined length, and moves the mouth of the character on the display screen only for the time required for the synthesized speech (time within the audiovisual tolerance). A device that synchronizes synthesized speech and moving images (display control device) is also known (see, for example, Patent Document 2).

  In the display control device described in Patent Document 2, text is divided into word strings of a predetermined length and words based on a characteristic number representing the length of text such as the number of commas, the number of periods, the number of words, and the like. The length of the sequence is counted, speech synthesis is performed for each word sequence, and a word sequence display time required to utter synthesized speech for each word sequence at a constant speech rate is calculated. In this display control device, the synthesized speech of the word string is output for the calculated word string display time, and the moving image is displayed for the word string display time (for example, the moving image display that moves the mouth of the character). ).

Therefore, in the synchronization system between the moving image and the text / speech converter described in Patent Document 1, the lips are displayed by causing the synthesized speech to be uttered in accordance with the movement of the lips displayed in the moving image. The moving image and the synthesized voice are synchronized. Moreover, the display control apparatus described in Patent Document 2 is designed to synthesize synthesized speech at a constant utterance speed by calculating the time required for utterance from the length of text.
Japanese Patent Laid-Open No. 10-171486 (paragraphs 0015 to 0030, FIGS. 2 and 3) JP-A-5-313686 (paragraphs 0023 to 0048, FIGS. 2, 6 and 7)

  However, in the synchronization system between the moving image and the text / speech converter described in Patent Document 1, the duration of each phoneme during speech synthesis is adjusted according to the lip shape, In order to synchronize with a moving image, there was a problem that it was not possible to synchronize the video of the content with the scene where the lips are not captured and the synthesized audio. For example, in the case where a synthetic voice is struck in synchronism with an unvoiced movie, the lips are not always shown, so that the voice cannot be uttered as a synthetic voice.

  Further, in the display control device described in Patent Document 2, since the entire text to be synthesized is uttered as synthesized speech having a constant utterance speed, the utterance speed is changed for each arbitrary word. I could not.

  The present invention has been made to solve the above-described problems, and a speech synthesis control device capable of easily synchronizing synthesized speech with a synchronization target signal such as a video signal having a predetermined display time, It is an object of the present invention to provide a method, a program thereof, and a synthesized voice data generation apparatus.

  The present invention was devised to achieve the above object. First, the speech synthesis control device according to claim 1, the speech synthesis speech synthesizes the text to be read out and uttered as synthesized speech. Based on a preset utterance start position indicating the start position of voice synthesis and a preset utterance start time indicating a start time at which the text is uttered as a synthesized voice, the text is synthesized and synthesized voice The speech synthesis control device controls the speech synthesizer so as to utter as the speech synthesis start timing acquisition unit and the speech synthesis start control unit.

According to such a configuration, the speech synthesis control device acquires the utterance start position and the utterance start time as the utterance start timing by the utterance start timing acquisition unit.
The utterance start position designates an arbitrary position in the text, and designates a word that starts utterance first for each predetermined utterance range. The utterance start position may be an arbitrary character in the text and an utterance start character that starts utterance first for each predetermined utterance range. In addition, the utterance start character is not limited to the character itself constituting the text, and may be a phonetic character such as a Roman character or a kana generated from the text. The utterance start time is a time measured from a predetermined reference time, and designates the start of the utterance of the utterance start character.

Then, the speech synthesis control device causes the speech synthesis start control means to synthesize the text from the utterance start position to the speech synthesizer, and to start the utterance of the speech synthesizer so that the text is uttered as synthesized speech from the utterance start time. Control.
In this case, the speech synthesis control device needs to have time measuring means for measuring time, or to obtain a time signal supplied from time measuring means provided outside. Then, the speech synthesis control device outputs the utterance start time to the speech synthesizer so that speech synthesis is started at the utterance start timing obtained based on the time measured by the time measuring means.

According to a second aspect of the present invention, there is provided a speech synthesis control apparatus according to the first aspect, wherein the speech synthesis control apparatus includes an utterance speed determining means and an utterance speed control means.
According to such a configuration, the speech synthesis control device uses the speech speed determination unit to utter speech as synthesized speech for each preset speech range indicating a speech synthesis range of text starting from the speech start position, and Determine its format. Further, the speech synthesis control device outputs the speech rate determined by the speech rate determination unit to the speech synthesis device by the speech rate control unit so that the speech rate is determined for each speech range. Controls speech synthesis.
In this case, the utterance start timing acquisition unit or the speech synthesis control unit may define the utterance range. This utterance range can be defined as a character string from an arbitrary utterance start character to a character (utterance end character) that precedes the utterance start character arranged next to the arbitrary utterance start character.

Furthermore, the speech synthesis control device according to claim 3 is the speech synthesis control device according to claim 1, wherein the speech speed determining means, the speech speed measuring means, the speech speed conversion control means, the speech speed conversion means, It is characterized by having.
According to such a configuration, the speech synthesis control device uses the speech speed determination unit to utter speech as synthesized speech for each preset speech range indicating a speech synthesis range of text starting from the speech start position, and Determine its format.

The speech synthesis control device measures the speech rate of the synthesized speech uttered by the speech synthesizer by the speech rate measuring means. The speech synthesis control device compares the speech rate determined by the speech rate determination unit by the speech rate conversion control unit with the speech rate measured by the speech rate measurement unit, and determines the synthesized speech uttered by the speech synthesis device. A speech speed control signal for converting the speech speed of the synthesized speech by the speech synthesizer is output so that the speech speed approaches the speech speed determined by the speech speed determining means.
In the speech synthesis control device, the speech rate conversion unit determines the speech rate of the synthesized speech uttered by the speech synthesizer based on the speech rate control signal output from the speech rate conversion control unit. The synthesized speech with the utterance speed converted so as to approach the utterance speed is output.

According to a fourth aspect of the present invention, there is provided the speech synthesis control apparatus according to the second or third aspect, wherein the speech rate determining means determines the speech rate format.
According to such a configuration, the time required for the speech rate determining means to speak the speech range calculated from the speech start time of the speech start position of the speech range and the next speech start time of the next speech start position where the next speech starts. And the format expressed based on the number of characters in the utterance range, the format expressed based on the number of utterance characters per unit time, the format expressed based on the utterance time required for utterance in the utterance range, or the reference speech rate It is determined as the speech rate of any one of the formats represented based on the ratio.

  The speech synthesis data generation device according to claim 5 is a text to be read out to be provided to the speech synthesis control device according to claim 1, the speech start position, and the speech corresponding to the speech start position. A speech synthesis data generation device for generating a start time, comprising speech recognition means, speech start timing acquisition means, and speech synthesis data storage means.

  According to this configuration, the speech synthesis data generating device recognizes the speech by the speech recognition unit and generates text, and the speech start position of the text generated by the speech recognition unit by the speech start timing acquisition unit and An utterance start time corresponding to the utterance start position is acquired as an utterance start timing. Then, the speech synthesis data generation device stores the text generated by the speech recognition unit and the speech start timing acquired by the speech start timing acquisition unit as speech synthesis data by the speech synthesis data storage unit.

  Note that the function of the utterance start timing acquisition unit included in the voice synthesis data generation device may be included in the voice recognition unit. In addition, the speech synthesis data generation device may be provided on either the transmission side or the reception side. Here, the “transmission side” is a broadcasting station or web provider that provides content by video and text and audio, or various storage media storing content, and is connected to the reception side by a wired or wireless data communication network. Is. The “reception side” outputs content delivered from the transmission side via a wired or wireless data communication network in synchronization with a synchronization target signal such as video and synthesized audio.

  Furthermore, the speech synthesis control method according to claim 6, the speech start control position set in advance indicating a speech synthesis start position of the text when the text to be read out is synthesized as speech and synthesized speech, A speech synthesis control method for controlling a speech synthesizer to synthesize the text and synthesize it as synthesized speech based on a preset utterance start time indicating a start time to utter the text as synthesized speech. And an utterance start timing acquisition step and a speech synthesis start control step.

  According to this procedure, the speech synthesis control method acquires the utterance start position and the utterance start time as the utterance start timing in the utterance start timing acquisition step. In the speech synthesis control method, in the speech synthesis start control step, the speech synthesizer starts speech so that the speech synthesizer synthesizes text from the utterance start position and utters the text as synthesized speech from the utterance start time. Control.

  According to a seventh aspect of the present invention, there is provided a speech synthesis control program comprising: a predetermined speech start position that indicates a speech synthesis start position of the text when the text to be read out is speech-synthesized and uttered as synthesized speech; In order to control the speech synthesizer to synthesize the text and utter it as synthesized speech based on a preset utterance start time indicating the start time to utter the text as synthesized speech, It was made to function as start timing acquisition means and speech synthesis start control means.

  According to this configuration, the speech synthesis control program acquires the utterance start position and the utterance start time as the utterance start timing by the utterance start timing acquisition unit. The speech synthesis control program controls speech start of the speech synthesizer so that the speech synthesizer causes the speech synthesizer to synthesize text from the utterance start position and utters the text as synthesized speech from the utterance start time. .

  According to the invention described in claim 1, claim 6 or claim 7, the utterance of the synthesized voice from the predetermined utterance start time and the synchronization target signal such as the video signal of the predetermined display time are adjusted to an arbitrary time. This makes it easy to synchronize.

  According to the second aspect of the present invention, the speech synthesizer can synthesize a text and synthesize a synthesized speech having a predetermined speech rate determined for each speech range. The speech synthesizer utters the synthesized speech by switching to a predetermined speech rate at the speech start timing sent from the speech synthesis start control unit by the speech rate control unit that controls the speech rate.

  According to the third aspect of the present invention, even when the speech rate cannot be switched in the speech synthesizer (means), the target speech rate can be adjusted. This makes it possible to adjust the synthesized speech, which is the output of the general-purpose speech synthesizer (means), to the target speech speed, which is economical for the user, and further, the speech synthesis control device of the present invention itself. It also leads to widespread use.

  According to the fourth aspect of the present invention, since it is possible to specify and determine the speech rate in a format that can be handled by the speech synthesizer, the speech rate format determined by the speech rate determining means and the speech rate in the speech synthesizer The speech rate control means can control the speech rate determined by the speech rate determination unit by converting it into a speech rate that can be handled by the speech synthesizer.

  According to the fifth aspect of the present invention, the voice data is converted into text by the character recognition means based on the voice spoken by the person, and the utterance start position and the utterance start time of the voice spoken by the person are uttered. Acquired as the start timing, the text and the utterance start timing can be stored as data for speech synthesis. The voice spoken by the person is not limited to the voice data that has already been recorded, but may be a case where the voice is collected and broadcast live in real time.

  Therefore, according to the present invention, there is provided a speech synthesis control device, a method thereof, a program thereof, and a synthesized speech data generation device capable of easily synchronizing synthesized speech to a synchronization target signal such as a video signal having a predetermined display time. Can be provided.

  By the way, by changing the video display speed, it is also possible to synchronize the synthesized voice of the predetermined speech speed and the video, but changing the video display speed changes the appearance, for example, In the case of a movie, it will also reduce the value of the work. Therefore, it is useful to vary the utterance speed and synchronize the video and the synthesized voice as in the present invention.

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

(First embodiment)
[Configuration of speech synthesis controller]
FIG. 1 is a block diagram showing the configuration of the speech synthesis control device according to the first embodiment of the present invention. FIG. 2 is a diagram for explaining the concept of the utterance start character of the speech synthesis control device according to the embodiment of the present invention. FIG. 3 is a diagram for explaining the concept of the speech rate of the speech synthesis control device according to the embodiment of the present invention. FIG. 3 (a) is an exemplary diagram when the speech rate is specified, and FIG. 3 (b) is a certain range. It is an illustration figure in the case of designating with the time concerning utterance of (2), (c) is an illustration figure in the case of calculating from the utterance start timing, and (d) is designated by a ratio with a predetermined standard utterance speed. It is an illustration figure in the case.

First, the configuration of the speech synthesis control device 1 will be described with reference to FIG.
The speech synthesis control device 1 synthesizes speech to be read out and utters it as synthesized speech. A speech start position that indicates a speech synthesis start position that is set in advance and a start time at which the text is uttered as synthesized speech. In order to control the speech synthesizer 7 to synthesize text and synthesize it as synthesized speech at a predetermined utterance speed based on a preset utterance start time indicating The speech synthesis data storage means 11, the analysis separation means 12, the speech start timing acquisition means 13, the speech synthesis start control means 14, the speech speed determination means 15, and the speech speed control means 16 are mainly provided. Yes.

  The voice synthesis data input means 10 is for inputting voice synthesis data into the voice synthesis control device 1. The voice synthesis data input means 10 includes, for example, a receiver for receiving terrestrial digital broadcasts, a receiver for receiving cable television broadcasts, various interfaces for connecting to a network such as the Internet, and reading from various removable media. There is a device.

  The speech synthesis data is a package (stream) of speech start timing, speech speed, and text. The utterance start timing is set in advance to indicate a preset utterance start position indicating the start position of text speech synthesis when the text is synthesized and uttered as synthesized speech, and a start time indicating the text to be uttered as synthesized speech. The utterance start time is determined. The speech synthesis data is a package of the utterance start timing and the utterance speed, and may be configured separately from the text.

  The utterance start position designates an arbitrary position in the text, and designates the first word to start utterance for each predetermined utterance range. The utterance start position may be an arbitrary character in the text and an utterance start character that starts utterance first for each predetermined utterance range. The utterance start character may be the beginning of each clause of the text, a phoneme or a character, and is not particularly limited to the character itself constituting the text, for example, a Roman character or a kana generated from the text. It may be a phonetic character.

This utterance start character may be used as a basic voice unit of the voice to be connected when outputting the synthesized voice. Examples of speech units include phonemes, syllables, VCV (Vowel-Consonant-Vowel), CVC (Consonant-Vowel-Consant), units by clustering, and composite units.
A phonetic character represents at least one ideographic reading such as a kanji character, and is composed of a character string in a time series of text sentences. Therefore, when the text is an ideographic character such as a Chinese character, the first phonogram of reading of the ideogram specifying the start of the utterance may be used as the utterance start character.

In the present specification, the text is described as a kanji-kana mixed sentence, but is not limited thereto, and for example, kanji only or kana-only Japanese may be used. The language handled as text is not limited to Japanese, and may be various languages such as English, German, Chinese, and the like. Also, for speech synthesis, it is not particularly necessary for notation that directly represents phonemes, such as kana and romaji, but it is possible to express phonetics with only letters, such as kanji kana mixed sentences. If you can't, you need to convert the text to phonetic characters. The phonetic characters are not limited to those generated from ideographic characters such as Kanji characters, but have a concept including Roman notation and kana notation that directly represent phonemes.
The speech synthesizer 7 may store a dictionary corresponding to the text for speech synthesis and output the synthesized speech in accordance with this dictionary.

  Further, as described above, the utterance start position designates an arbitrary position in the text, and designates a word as an utterance start character that starts utterance first for each predetermined utterance range. is there. Here, the concept of the utterance start character of the speech synthesis control device according to the embodiment of the present invention will be described with reference to FIG. Here, a description is given using an example of a text “A textbook that reads quickly”.

For example, in this sentence example, the utterance range starting from “yo” of the utterance start character a ₁ is “reading text”, and the utterance is terminated by “no” one character before “ha” of the utterance start character a ₂ It is a letter. Similarly, the utterance range starting from “ha” of the utterance start character a ₂ is “has been playing”, and “ru” immediately before the “ho” of the utterance start character a ₃ The utterance end character. In addition, the utterance range starting from “ho” of the utterance start character a ₃ is “Hoho no Rei”, and the character “i” before the punctuation mark “.” Is the utterance end character.

Therefore, the order for defining the utterance start character is defined according to the number of characters existing before the defined utterance start character. For example, in the sentence example, “yo” of the utterance start character a ₁ is the first order because there is no character before this character. In addition, since “ha” of the utterance start character a ₂ has a character string constituting “reading text” before this character, the number of characters of this character string (11 characters) +1 character, that is, the 12th character Defined as an order. Note that the order is similarly defined in the case of “ho” of the utterance start character a ₃ .

In addition, as described above, the order in which the utterance start characters are defined is not limited to the case where the utterance start characters are defined according to the number of characters existing before the specified utterance start characters, and the utterance start characters in the text to be read out are determined. You may make it prescribe | regulate by the method.
For example, the order can be defined by the relationship between the order of utterance start characters and the number of characters up to the next utterance start character. Referring to the example sentence in FIG. 2, when 11 characters are defined as the first, the first utterance start character is “yo” of a ₁ , and the next utterance start character is “ha” of a _2. It can be expressed. If the second character is defined as 12 characters, it can indicate that the next utterance start character is “ ₃ ” of a ₃ . In addition, when it is determined in advance that transmission is performed in order from the first, it is also possible to omit the information on what number and specify only the number of characters such as “11 characters”, “12 characters”, etc. A starting character can be defined.

  By the way, the punctuation mark “.” Is used as a signal for the end of the utterance when the utterance is started from an arbitrary utterance start character. Also, the punctuation mark “,” can also be used as a cue for the end of the utterance when the utterance is started from an arbitrary utterance start character. Punctuation marks and reading marks may or may not be counted as characters for defining the order. However, although punctuation marks and punctuation marks are not uttered, they have the role of separating sentences and sentences as “between”, and therefore it is preferable to count them when calculating the utterance speed.

Further, not only to specify the utterance start character a _1, a _2, a ₃ itself as described above, for example, 1 "only" after the character of b _1, "one" in b _2, the b ₃ " Mu specified or 1 character preceding c ₁ letter u "," the the c ₂ ", leave the designation of" Turning "of c _3, utterance start timing acquisition means 13 to determine this, utterance start The characters a ₁ , a ₂ , and a ₃ may be acquired.
Note that the number of characters before and after the designation is not limited to one character, and may be any number of characters before or after how many characters to select according to the user's preference when setting the utterance start character.

The utterance start time is a time measured from a predetermined reference time, and designates the start of the utterance of the utterance start character. This utterance start time can be represented by, for example, a relative time from the beginning of the sentence, a relative time from the previous utterance start time, a relative time from the beginning of the content, or a time based on Greenwich Mean Time. In particular, it is preferable to start timing based on the utterance start time of the first utterance start character of the text, and specify the utterance start time of subsequent utterance start characters based on the utterance start time (from the beginning of the sentence). Relative time). The time based on the Greenwich Mean Time can be used when a predetermined content is broadcast at a predetermined time.
Note that as the utterance start timing, only the utterance start character may be designated as the utterance start position. However, the utterance start time needs to be stored in the speech synthesis control device 1 as a prescribed value for each order in which the utterance start character is uttered.

The utterance speed is determined according to the relationship between the time required for utterance in a preset utterance range indicating the range of speech synthesis of text starting from the utterance start position and the number of characters in the utterance range.
Further, as described above, this utterance speed is determined by the relationship between the number of uttered characters and time (hereinafter referred to as “character number-time relationship”), but can be calculated or converted so as to have a character number-time relationship. As long as there is a decision, any method may be used. For example, in addition to the case of designating by the number of phonemes or words uttered within a certain time, the speech synthesizer 7 when calculating from the time required to output the synthesized speech in the utterance range and the interval of the speech start timing. There is a ratio to the standard speech rate.

Here, the concept of the speech rate will be described with reference to FIG. The description will be made with reference to the above-mentioned text example “A textbook that reads quickly”.
The “example of designating by speaking rate” in (a) determines the speaking rate for each speaking range based on the number of speaking characters per unit time. For example, in this sentence example, the speech rate is set to 5 characters per second for the text range of the utterance range from the utterance start character “yo” to the utterance end character “no”, and from the utterance start character “ha” The character string of the utterance range from the utterance start character “ho” to the utterance end character “i” is set to 6 characters per second in the utterance range character string “ending fast” to the end character “ru”. Hohorei is 5 characters per second.

  The “example of designating by the time required for utterance within a certain range” in (b) designates the utterance speed for each utterance range by the utterance time required for utterance within the utterance range. For example, in this sentence example, the utterance speed is set to 2.2 seconds (first) for the character string “Reading text” in the utterance range from the utterance start character “yo” to the utterance end character “no”. The character string “I am playing fast” from “ha” to the utterance end character “ru” is 2 seconds (second), and the utterance start character “ho” to the utterance end character “i” The character string “Hoho no Rei” in the utterance range is set to 1.4 seconds (third).

  The “example calculated from the utterance start timing” in (c) is an utterance speed for each utterance range based on the relationship between the time between the utterance start time and the utterance end time in the utterance range and the number of characters in the utterance range (number of characters / time). Is specified. That is, in this case, the utterance speed is specified by the utterance start character described above and the utterance start time described above. For example, in this sentence example, the utterance is started with the utterance start time of the utterance start character “yo” as a reference of 00.00 seconds, the utterance start time of the utterance start character “ha” is set to 02.20 seconds, and the utterance start character “ The utterance start time of “ho” is set to 04.20 seconds.

  The “example of designating by a ratio with a predetermined standard utterance speed” in (d) designates the utterance speed for each utterance range by the ratio to the reference utterance speed. For example, in this example sentence, the utterance speed is set to 1.0 times the character string “reading text” in the utterance range from the utterance start character “yo” to the utterance end character “no”, and from the utterance start character “ha”. The character string in the utterance range from the utterance start character "ho" to the utterance end character "i" is set to 1.2 times the character string "has been playing" until the utterance end character "ru". The column “Hou no Rei” is set to 1.0 times speed. Note that 1.0 times the speed represents the reference speech speed. For example, if the standard speaking rate is 5 characters per second, 1.2 times speed represents a speaking rate of 6 characters per second.

In addition to using the text for reading as it is, as described above, a predetermined delimiter (column) indicating the utterance start character is inserted into the text for reading, and any one of the utterance start character, the utterance start time, and the utterance speed is selected. Or you may prescribe everything.
Here, the case where it prescribes | regulates using XML (extensible Markup Language) is demonstrated.

[Description example 1 in XML]
<s time = “00:00” speed = “1.0”> Reading text </ s>… (line 1)
<s time = “2.2sec” speed = “1.2”>
<s time = “4.2sec” speed = “1.0”></s> ... (line 3)

In this case, the utterance range is specified by the tag <s>, the utterance start time of the specified utterance range is specified by the time attribute (attribute), and the utterance speed of the specified utterance range is specified by the speed attribute. .
Here, as shown in lines 1 to 3, divide the sentence example of “How to read text quickly” into the utterance range enclosed by the tag <s>, line by line. An utterance start time and an utterance speed are specified.

Note that the syntax of tag and attribute names, description methods, and the like are not limited to this example as long as the utterance start time and utterance speed can be specified.
Furthermore, even if the language to be used is not XML, any delimiter (column) may be used as long as it can be distinguished from the text for reading.

If the operation when omitted is predetermined, the specification of the utterance start time and the utterance speed may be omitted.
For example, if the utterance start time is omitted, it is defined that the utterance starts immediately after the output of the immediately preceding utterance range ends, and if the utterance speed is omitted, the utterance speed in the immediately preceding utterance range is used. (In the case of the first utterance range, the speed is set to 1 × speed). An example of description in XML in this case is shown below.

[Description example 2 in XML]
<s time = “00:00” speed = “1.0”> Reading text </ s>… (line 1)
<s speed = “1.2”> is a quick </ s>… (line 2)
<s time = “4.2sec”></s> ... (line 3)

  In this case, the utterance start time of the utterance range “Hatsusokudo wo” shown in line 2 is the time immediately after the output of the utterance range “reading text” shown in the immediately preceding line 1, and line 3 The utterance speed of the utterance range “Shihou Shiho no Rei” indicated by is 1.2 times the same as the utterance range “Hakusokudo wo” indicated by the immediately preceding line 2.

  Returning to FIG. 1 and continuing the description, the voice synthesis data storage unit 11 stores the voice synthesis data input by the voice synthesis data input unit 10. Examples of the speech synthesis data storage unit 11 include a hard disk and a semiconductor device.

The analysis / separation unit 12 reads out the speech synthesis data from the speech synthesis data storage unit 11, analyzes the speech synthesis data, separates the speech synthesis data into text, speech start timing, and speech speed, A text is sent to the speech synthesizer 7, an utterance start timing is sent to the utterance start timing acquisition means 13, and an utterance speed is sent to the utterance speed determination means 15.
For example, the analysis / separation unit 12 adds a flag before and after each data in order to distinguish the text, the utterance start timing, and the utterance speed, and the analysis / separation unit 12 recognizes the flag. However, the present invention is not limited to this, and the feature portion that specifies the utterance start timing and the utterance speed may be particularly recognized and determined. Further, when the flag is described in XML, the above-described tag can be used.

  The utterance start timing acquisition unit 13 acquires the utterance start position and the utterance start time as the utterance start timing. Here, description will be made assuming that the utterance start timing acquisition means 13 determines the utterance start character as the utterance end position with reference to the rules set in advance as described with reference to FIG. Not limited to this, it may be performed by the speech synthesis start control means 14 or by other independent means.

The speech synthesis start control means 14 outputs an utterance start signal to the speech synthesizer 7 so that the utterance is started at the utterance start time of the utterance start timing acquired by the utterance start timing acquisition means 13, and the speech synthesizer 7 7 is used to control the start of speech synthesis. The speech synthesis start control means 14 not only outputs a cue, but also counts the number of characters up to the next utterance start character, determines the utterance end character, and uses the utterance end time of the utterance end character as a cue. 7 may be output.
The utterance speed determination means 15 acquires the utterance speed to be uttered for each utterance start timing acquired by the utterance start timing acquisition means 13 via the separation analysis unit 12, and uses the acquired utterance speed as the utterance at the target utterance start timing. It is determined as a speed.

As described with reference to FIG. 3, the speech rate control unit 16 outputs the speech rate determined by the speech rate determination unit 15 to the speech synthesizer 7 and becomes the speech rate determined at each speech start timing. Thus, the speech synthesis of the speech synthesizer 7 is controlled.
As described with reference to FIG. 3B, when the speech synthesizer 7 sets the speech speed as how many characters per second, at the first time, the speech speed control means 16 , The utterance speed is calculated by 11 characters ÷ 2.2 seconds, and the calculated 5 characters per second is sent to the speech synthesizer 7 as the utterance speed. At the second time, the speech rate control means 16 calculates the speech rate by 12 characters / 2 seconds, and sends the calculated 6 characters per second to the speech synthesizer 7 as the speech rate. At the third time, the speech rate control means 16 calculates the speech rate by 7 characters / 1.4 seconds, and sends the calculated 5 characters per second to the speech synthesizer 7 as the speech rate.

If the speech synthesizer 7 sets the speech rate by time, the speech rate control means 16 sends the first to third designated times to the speech synthesizer 7 to synthesize speech. As described above, the device 7 calculates the speech rate as how many characters per second, and outputs a synthesized speech in which the speech rate is adjusted as many characters per second for each speech range.
Further, as described with reference to (c) of FIG. 3, when the speech synthesizer 7 is configured to set the speech speed at the speech start timing, the speech speed control means 16 determines each speech timing. The speech synthesizer 7 sends to the speech synthesizer 7, and the speech synthesizer 7 calculates the speech speed as how many characters per second as described above, and outputs synthesized speech with the speech speed adjusted as many characters per second for each speech range.

Since the utterance start time of the next utterance start character “ha” is 02.20 seconds, the character string “Yomiageyo text” of the utterance range from the utterance start character “yo” to the utterance end character “no” is the 2. The utterance needs to be completed in 2 seconds (first). Therefore, the speech rate control means 16 calculates the speech rate by 11 characters ÷ 2.2 seconds, and sends the calculated 5 characters per second to the speech synthesizer 7 as the speech rate.
Next, in the character string “has been playing” from the utterance start character “ha” to the utterance end character “ru”, the utterance start time of the next utterance start character “ho” is 04. Since it is 20 seconds, the utterance needs to be completed in 2 seconds calculated by (04.20-02.20). Therefore, the utterance speed control means 16 calculates the utterance speed by 12 characters / 2 seconds, and sends the calculated 6 characters per second to the speech synthesizer 7 as the utterance speed.

  By the way, when there is a sentence that follows this sentence example, the character string “Honoho” in the utterance range from the utterance start character “ho” to the utterance end character “i” is determined according to the utterance start time of the utterance start character of the next sentence. The utterance speed of “Rei” can be determined. However, if there is no sentence following this sentence example, the utterance will end with this sentence example, so simply specifying the utterance start character and the utterance start time will result in the character string “Honorei” in the utterance range. The speech rate cannot be determined.

  Therefore, in this case, for example, the utterance start character “” (space) of the next sentence and the utterance start time are specified on the assumption that the sentence further exists next to the utterance range “Hou no Rei”. Can be determined. Note that the character designated as the utterance start character as being virtually present is not limited to a space, and an arbitrary character or symbol may be assigned. In addition, a flag for distinguishing virtually existing may be added to the data string.

  Further, the speech rate when the speech rate is not designated may be determined in advance. For example, in the utterance range in which the utterance speed is not specified, the speed of the immediately preceding utterance range is continued, and in the case of the first utterance range, if the speed is set to 1.0 times in advance, the utterance start character of the next sentence is virtually You don't have to.

  Further, as described with reference to (d) of FIG. 3, when the speech synthesizer 7 is set to what speed the speech speed is set, how much the speech speed control means 16 is speech-synthesized. The speech synthesizer 7 multiplies the reference speech speed to calculate how many characters per second, and outputs the synthesized speech with the speech speed adjusted as many characters per second for each speech range. If the speech synthesizer 7 sets the speech rate by the number of characters per second, the speech rate control means 16 calculates the number of characters per second by multiplying the reference speech rate. The synthesized value is sent to the speech synthesizer 7 to output a synthesized speech in which the speech rate is adjusted to how many characters per second for each speech range.

  Here, an example of text to be read out consisting only of characters has been shown, but the speech rate is similarly applied even to auxiliary information for speech synthesis such as phoneme information or data specialized for speech synthesis that does not include characters. Can be specified. The speaking rate may be specified for any part of the text. For example, if you specify the utterance speed for each phrase of the text to be read, or specify the time to be used for utterance of the phrase for each phrase, you can create a fast-talking part or a slowly speaking part in one sentence . In addition, the speech rate determination means 15 uses a predetermined method such as applying the previous speech rate or applying the standard speech rate of the speech synthesizer 7 for the part where the speech rate is not specified. Determine the speaking rate.

[Main functional configuration of speech synthesizer, video display, and synchronizer]
The speech synthesizer 7 performs word analysis on the text, generates a phonetic word string with readings, spaces, and accents, generates a duration length and a basic frequency pattern for each phoneme of the phonetic word string, The synthesized speech generated by referring to the speech waveform database based on the fundamental frequency pattern is output as an electrical signal.
In the word analysis of the text, the speech synthesizer 7 analyzes the word by referring to the dictionary to determine “reading” of the word string, analyzes the dependency of the word string, and “between” the word strings. The phonetic word string corresponding to the word string is generated from “read” and “between” of the word strings, and “accent” is determined from the phonetic word string. The synthesized speech output as an electrical signal by the speech synthesizer 7 is output from a speaker (not shown).

  In addition, the speech synthesizer 7 particularly includes an utterance start timing setting unit and an utterance speed setting unit (all units are not shown). The speech start timing setting means sets the speech start timing sent from the speech synthesis start control means 14 for speech synthesis. The utterance start timing setting means sets the utterance start character in the text sent from the analysis / separation means 12, and sets the utterance start time of the utterance start character. The utterance speed setting means sets the utterance speed for each utterance range based on the utterance start timing sent from the utterance speed control means 16.

  The video display device 8 displays video for synchronizing with synthesized speech output from the speech synthesizer 7. Examples of the video display device 8 include a digital broadcast television receiver capable of handling text supplied as a digital signal, a personal computer, a projector, and a mobile phone with a moving image display function. Here, in particular, a television receiver is used for digital terrestrial broadcasting for which recent broadcasting has started. Although the video display device 8 is not illustrated and described, the video display device 8 separates the horizontal synchronizing means from the receiving means for receiving the broadcast wave, the synchronizing separation means for separating the video signal and the synchronizing signal, and the synchronizing signal. The apparatus mainly includes a horizontal synchronization separation unit, a vertical synchronization separation unit that separates a vertical synchronization signal from a synchronization signal, and a display unit that displays a video signal on a display screen in synchronization with the horizontal synchronization signal and the vertical synchronization signal. .

  The synchronizer 9 controls the output timings of the voice synthesizer 7 and the video display device 8 so that the synthesized voice and the video are output in synchronization. The synchronizer 9 is provided with a time measuring means (not shown) for measuring an arbitrary time, and the time measured by the time measuring means is given to the speech synthesizer 7 and the video display device 8.

  By the way, the speech synthesizer 7 outputs synthesized speech by control based on the speech start timing of the speech synthesis control device 1, while the video display device 8 outputs video based on the horizontal synchronization signal and the vertical synchronization signal. . For this reason, the speech synthesizer 7 sets the utterance start time at the utterance start timing and the output times of the horizontal synchronization signal and the vertical synchronization signal to the time of the time measuring means (not shown) of the synchronization device 9. The synthesized voice to be output and the video signal output from the video display device 8 can be synchronized.

[Operation of speech synthesis controller]
Next, the operation of the speech synthesis control device according to the first embodiment will be described with reference to FIG. 4 (refer to FIGS. 1 to 3 as appropriate).
FIG. 4 is a flowchart showing the operation of the speech synthesis control device according to the first embodiment of the present invention.
<Receiving, inputting and storing data for speech synthesis>
In the voice synthesis control device 1, when the voice synthesis data input means 10 receives voice synthesis data from the outside, the voice synthesis data storage means 11 stores the voice synthesis data (step SA1).
<Acquisition and analysis steps for speech synthesis data>
The analysis / separation means 12 reads the voice synthesis data from the voice synthesis data storage means 11 and analyzes the voice synthesis data (step SA2).

<Speech synthesis data separation step>
The analysis / separation means 12 separates the text, the utterance start timing, and the utterance speed from the speech synthesis data (step SA3).
The analysis / separation unit 12 determines whether or not text is included in the read speech synthesis data. If the text is included (Yes in step SA31), the analysis / separation unit 12 extracts the text (step SA34). Text is output to the speech synthesizer 7 (step SA8), and the process proceeds to step SA9.

If the read speech synthesis data is not text (No in step SA31), the analysis / separation means 12 determines whether the separated speech synthesis data is the speech start timing, and if it is the speech start timing (step (Yes in SA32), the utterance start timing is extracted and the process proceeds to Step SA4 (Step SA35).
If the separated speech synthesis data is not the speech start timing (No in step SA32), the analysis / separation means 12 determines whether or not the separated speech synthesis data is the speech speed, and if it is the speech speed (step (Yes in SA33), the speech rate is acquired and the process proceeds to Step SA6 (Step SA36).

  If the separated speech synthesis data is not the speech rate, the analysis / separation means 12 determines that the data is other data accompanying the speech synthesis data and moves the process to step SA9 (No in step SA33). Examples of the other data include a video signal. When the voice synthesis data and the video signal are made into one stream, the voice synthesis control device 1 first receives the stream, and the analysis / separation means 12 separates the video signal to obtain a video signal (horizontal synchronization signal and vertical synchronization signal). Can be sent to the video display device 8.

<Speech start timing acquisition step>
Then, the utterance start timing acquisition unit 13 acquires the utterance start position and the utterance start time for defining the utterance start character that is the utterance start timing extracted by the analysis / separation unit 12, and synthesizes the acquired utterance start timing with speech synthesis. It passes to the start control means 14 (step SA4).
<Speech synthesis start control step>
The voice synthesis start control means 14 uses the voice timing to start the utterance so that the voice synthesizer 7 starts utterance as synthesized voice at the utterance start time of the utterance start timing passed from the utterance start timing acquisition means 13. The synthesis start control means 14 is instructed (step SA5, step SA8), and the process proceeds to step SA9.

<Speech speed determination step>
The speech rate determination unit 15 acquires the speech rate extracted by the analysis / separation unit 12, determines whether the acquired speech rate is a format that can be set in the speech synthesizer 7, and if it is a format that can be set, The acquired speech rate is determined as the speech rate of the synthesized speech output from the speech synthesizer 7. If the format is not settable, the acquired speech rate is converted into a configurable speech rate, and the speech synthesizer 7 Is determined as the speech speed of the synthesized speech output from (step SA6).

In the case of a format that cannot be set, for example, the speech rate determination means 15 treats the speech rate as if the speech synthesizer 7 specifies “a ratio with a predetermined standard speech rate (double speed)”. In addition, the utterance speed determination means 15 "when specifying with the time required for utterance within a certain range" when the utterance speed acquired via the analysis / separation means 12 is "specified with the time required for utterance within a certain range". Is determined to be a speech rate converted into “when specified by a ratio with a predetermined standard speech rate (double speed)”. In this case, it is necessary to register a conversion rule of the speech rate format in a storage unit (not shown) that can be referred to by the speech rate control means 15.
<Speech speed control step>
The speech rate control means 16 measures the timing at which the speech rate determined by the speech rate determination means 15 is sent to the speech synthesizer 7, outputs the speech rate to the speech synthesizer 7 (step SA7, step SA8), and performs the processing in step SA9. Move to.

<Confirmation step for voice synthesis data>
If there is output target speech synthesis data stored in the speech synthesis data storage unit 11, the separation analysis unit 12 returns the process to step SA2 (Yes in step SA9), and there is no speech synthesis data. If so, the process ends (No in step SA9).

  According to the first embodiment, since the text can be uttered as synthesized speech so as to have a predetermined utterance speed at each utterance start position at a predetermined utterance start time, a video signal having a predetermined display time, etc. The synchronization target signal and the synthesized speech can be easily synchronized based on the utterance start timing. Therefore, according to the first embodiment, a movie such as a movie with a foreign language sound (so-called dubbed movie) or a silent movie, a video content such as a video and a television broadcast, etc., without being based on character information such as lips. Synchronized speech can be uttered in synchronism with arbitrary time, including quick chat and slow chat. In addition, even for audio contents such as radio broadcasts, it is possible to synthesize synthesized speech with a variable speech rate at an arbitrary time.

(Second embodiment)
Hereinafter, specific components and operations thereof will be mainly described in the second embodiment of the present invention. Note that the same components as those in the first embodiment are denoted by the same reference numerals and description thereof is omitted, and description of operations of the same components is also omitted.
[Configuration of speech synthesis controller]
FIG. 5 is a block diagram showing the configuration of the speech synthesis control device according to the second embodiment of the present invention.
The speech synthesis control device 2 speech-synthesizes the text to be read out and utters it as synthesized speech. A speech start position that indicates a speech synthesis start position that is set in advance and a start time at which the text is uttered as synthesized speech In order to control the speech synthesizer 7 to synthesize text and synthesize it as synthesized speech with a constant speech rate based on a preset speech start time indicating The speech synthesis data storage means 11, the analysis separation means 22, the speech start timing acquisition means 13, and the speech synthesis start control means 14 are mainly provided.

In the present embodiment, the speech synthesis data is composed of text and utterance start timing, and does not include the utterance speed. The speech synthesis control device 2 controls the speech synthesis device 7. Also, the speech synthesizer 7 is synchronized with the video display device 8 by the synchronization device 9 as in the case of the first embodiment.
The analysis / separation unit 22 reads out the speech synthesis data from the speech synthesis data storage unit 11, analyzes the speech synthesis data, separates the speech synthesis data into text and utterance start timing, and generates a speech synthesis device. 7 is sent, and the utterance start timing is sent to the utterance start timing acquisition means 13. Note that the text and the utterance start timing can be distinguished by the flag as described above.

[Operation of speech synthesis controller]
FIG. 6 is a flowchart showing the operation of the speech synthesis control device according to the second embodiment of the present invention. In addition, the same step number is attached | subjected to the step same as the flowchart of 1st embodiment shown in FIG. 4, and description is abbreviate | omitted.
The operation of the speech synthesis control device 2 is significantly different from the operation of the speech synthesis control device 1 in the speech synthesis data separation step.

<Speech synthesis data separation step>
The analysis separation unit 12 separates the text and the utterance start timing from the speech synthesis data (step SB3). That is, since the speech rate is not included in the speech synthesis data, the point that step SA33 and step SA36 are not provided is greatly different from the case of the first embodiment.
As in the case of the first embodiment, the analysis / separation means 22 determines whether or not the separated speech synthesis data is text, and if it is text (Yes in step SA31), extracts it as text. (Step SA34), text is output to the speech synthesizer 7 (step SA8), and the process proceeds to step SA9.

Similarly, if the separated speech synthesis data is not text (No in step SA31), the analysis / separation unit 22 determines whether the separated speech synthesis data is the speech start timing, and can determine the speech start timing. If it is yes (Yes in step SA32), it is extracted as the utterance start timing and the process proceeds to step SA4 (step SA35).
If the separated speech synthesis data is not the speech start timing (No in step SA32), the analysis / separation means 22 determines that the data is the other data associated with the speech synthesis data, and moves the process to step SA9.

  According to the second embodiment, since the speech synthesis control device 2 controls speech synthesis of the speech synthesizer 7 based on the speech start timing, a constant speech speed for each speech start position at a predetermined speech start time. The text can be uttered as synthesized speech so that

(Third embodiment)
Hereinafter, specific components and operations thereof will be mainly described in the third embodiment of the present invention. Note that the same components as those in the first embodiment are denoted by the same reference numerals and description thereof is omitted, and description of operations of the same components is also omitted. In the third embodiment, the speech synthesis control device 1 and the speech synthesis device 7 of the first embodiment described above are integrated, and the speech synthesis control function is built in the speech synthesis device 4. In the following, the means for realizing the speech synthesis control function is the same as that described in the first embodiment, and the description thereof is omitted.

[Configuration of speech synthesizer]
FIG. 7 is a block diagram showing the configuration of the speech synthesizer according to the third embodiment of the present invention.
The speech synthesizer 3 synthesizes a speech start position that indicates the speech synthesis start position of the text when the text to be read out is synthesized and uttered as a synthesized speech, and a start time at which the text is uttered as a synthesized speech. Based on the preset utterance start time shown, speech synthesis data input means 10, speech synthesis data storage means 11, analysis separation means 12, The speech start timing acquisition means 13, the speech synthesis start control means 14, the speech speed determination means 15, the speech speed measurement means 36, the speech speed conversion control means 37, the speech speed conversion means 38, and the speech synthesis means 39. And mainly.

  Therefore, the speech synthesizer 3 does not include the speech rate control unit 16 (see FIG. 1) of the speech synthesis control device 1, and instead of the speech rate control unit 16, the speech rate measurement unit 36 and the speech rate conversion A control means 37 and a speech speed conversion means 38 are provided, and a speech synthesis means 39 is further provided. Note that means other than the speech synthesizer 39 realize a speech synthesis control function for controlling speech synthesis of the speech synthesizer 39.

The utterance speed measuring means 36 measures the utterance speed of the synthesized speech output from the speech synthesis means 39. The utterance speed measuring means 36 can measure the utterance speed by, for example, recognizing the number of phonemes and measuring the time.
The speech speed conversion control means 37 compares the speech speed determined by the speech speed determination means 15 with the speech speed measured by the speech speed measurement means 36, and the speech speed of the synthesized speech output by the speech synthesis means 39 is determined. A speech speed control signal for converting the speech speed of the output of the speech synthesizer 39 is output so as to approach the speech speed determined by the speech speed determining means 15.
If the speech rate of the speech synthesizer 39 is known, the speech rate measuring unit 36 may not be provided. In that case, the speech speed conversion control means 37 may use the known speech speed instead of the speech speed measured by the speech speed measuring means 36.

The speech speed conversion means 38 is based on the speech speed control signal output from the speech speed conversion control means 37, and the speech speed of the synthesized speech output from the speech synthesis means 39 is determined by the speech speed determination means 15. The synthesized speech in which the speech speed is converted so as to be close to is output. The speech speed conversion means 38 divides the synthesized speech uttered by the speech synthesizer 39 at a fixed time period, omits a part of the divided data, and increases the reproduction speed, or separates the divided data. It is possible to slow down the reproduction speed by inserting and joining a part of them repeatedly. For example, by dividing the phoneme of “A” and omitting a part and shortening it like “Ah”, or by inserting a part and extending it like “Ah”, the speech speed can be converted. it can.
The voice synthesizer 39 includes the same means as the voice synthesizer 7 described in the first or second embodiment, and performs the same operation, so that the description thereof is omitted. Further, the speech synthesizing means 39 is synchronized with the video display device 8 by the synchronizing device 9 in the same manner as the speech synthesizing device 7 in the first embodiment.

[Operation of speech synthesis controller]
FIG. 8 is a flowchart showing the operation of the speech synthesizer according to the third embodiment of the present invention. In addition, the same step number is attached | subjected to the step same as the flowchart of 1st embodiment shown in FIG. 4, and description is abbreviate | omitted.
Compared with the operation of the speech synthesis control device 1, the operation of the speech synthesis device 3 is provided with speech synthesis means 39, so that instead of the speech synthesis step (step SC1) and speech rate control step (step SA7) The points including the speed measurement step (step SC2), the speech speed conversion control step (step SC3), and the speech speed conversion step (step SC4) are greatly different. Hereinafter, the steps SC1 to SC4 will be described.

<Speech synthesis step>
When there is a text passed from the analysis / separation unit 12, the speech synthesis unit 39 synthesizes the text according to a cue based on the utterance start position which is the utterance start timing output by the speech synthesis start control unit 14, and performs predetermined synthesis. Is uttered at each utterance start time as synthesized speech of the utterance speed (step SC1).
<Speech speed measurement step>
The speech rate measuring unit 36 measures the speech rate of the synthesized speech uttered by the speech synthesis unit 39 (step SC2).
<Speech speed conversion control step>
The speech speed conversion control means 37 compares the speech speed determined by the speech speed determination means 15 with the speech speed measured by the speech speed measurement means 36, and the speech speed of the synthesized speech spoken by the speech synthesis means 39. A speech speed control signal for converting the speech speed spoken by the speech synthesizer 39 is output so that the speech speed approaches the speech speed determined by the speech speed determiner 15 (step SC3).

<Speaking speed conversion step>
The speech speed conversion means 38 is based on the speech speed control signal output from the speech speed conversion control means 37, and the speech speed of the synthesized speech uttered by the speech synthesis means 39 is determined by the speech speed determination means 15. The synthesized speech in which the speech rate is converted so as to be close to is output (step SC4).
According to the third embodiment, each means for realizing the voice synthesis control function of the voice synthesizer 3 controls the voice synthesis of the voice synthesis means 39 based on the utterance start timing and the utterance speed. The text can be uttered as synthesized speech so as to have a predetermined utterance speed for each utterance start position at the start time.

(Fourth embodiment)
Hereinafter, specific components and operations thereof will be mainly described in the fourth embodiment of the present invention. Note that the same components as those in the first embodiment are denoted by the same reference numerals and description thereof is omitted, and description of operations of the same components is also omitted.
[Configuration of speech synthesis controller]
FIG. 9 is a block diagram showing a configuration of a speech synthesis control device according to the fourth embodiment of the present invention.
The speech synthesis control device 4 speech-synthesizes the text to be read out and utters it as a synthesized speech. The speech utterance starting position that indicates the speech synthesis start position of the text and the start time at which the text is uttered as a synthesized speech Input means 40 and text storage means for controlling the speech synthesizer 7 to synthesize text and synthesize it as synthesized speech at a predetermined utterance speed based on a preset utterance start time indicating 41, utterance start timing storage means 42, utterance speed storage means 43, utterance start timing acquisition means 13, speech synthesis start control means 14, utterance speed determination means 15, and utterance speed control means 16. In preparation. Therefore, the speech synthesis control device 4 uses the input unit 40, the text storage unit 41, and the speech start timing storage unit instead of the speech synthesis data input unit 10, the speech synthesis data storage unit 11, and the analysis separation unit 12. 42 and the speech rate storage means 43 are mainly provided.

The input unit 40 is for inputting the text, the speech start timing, and the speech speed into the speech synthesis control device 4 in a state in which the text is separated. Note that the speech start data and the speech speed may be the same data for speech synthesis and may have a data structure separated from text. In this case, the text may be input directly to the speech synthesizer 7 without going through the analysis / separation unit 12 or may be sent through the analysis / separation unit 12.
The text storage unit 41 stores the text input by the input unit 40.
The utterance start timing storage means 42 stores the utterance start timing input by the input means 40.
The utterance speed storage means 43 stores the utterance speed input by the input means 40.
The text storage means 41, the utterance start timing storage means 42, and the utterance speed storage means 43 are not limited to being configured by separate storage media, but are configured by allocating different storage areas on one storage medium. be able to.

[Operation of speech synthesis controller]
Next, referring to FIG. 10 (refer to FIG. 9 as appropriate), the operation of the speech synthesis control device according to the fourth embodiment will be described.
FIG. 10 is a flowchart showing the operation of the speech synthesis control device according to the fourth embodiment of the present invention. In addition, the same step number is attached | subjected to the step same as the flowchart of 1st embodiment shown in FIG. 4, and description is abbreviate | omitted. The operation of the speech synthesis control device 4 according to the fourth embodiment does not input speech synthesis data that is a single stream, but instead inputs text, speech start timing, and speech speed as input data. The point is different from the case of the first embodiment. Hereinafter, each step regarding handling of input data unique to the present embodiment will be described.

<Receiving step of input data>
In the speech synthesis control device 4, for example, when the input means 40 receives a broadcast wave as input data (step SD1), the input data is discriminated / stored (step SD2).
<Identification / storage step of input data>
The input means 40 determines whether or not the input data is text, and if it is text (Yes in step SD21), recognizes it as text (step SD24) and stores it in the text storage means 41 (step SD27). The text storage means 41 outputs the text to the speech synthesizer 7 in response to a command from a control means (not shown) (step SA8), and the process proceeds to step SD9.

If the input data is not text (No in step SD21), the input means 40 determines whether or not the input data is the utterance start timing. If the input data is the utterance start timing (Yes in step SD22), the input means 40 determines the utterance start timing. Recognize (step SD25) and store it in the utterance start timing storage means 42 (step SD28). The utterance start timing storage means 42 outputs the utterance start timing to the utterance start timing acquisition means 13 according to a command from a control means (not shown), and the process proceeds to step SA4.
If the input data is not the utterance start timing (No in step SD22), the input means 40 determines whether or not the input data is the utterance speed. If the input data is the utterance speed (Yes in step SD23), it is recognized as the utterance speed. (Step SD25), and it is stored in the speech rate storage means 43 (step SD28). The utterance speed storage means 43 outputs the utterance speed to the utterance speed determination means 15 according to a command from a control means (not shown), and the process proceeds to step SA6.

If the input data is not the speech rate, the input means 40 determines that the input data is the other data described above, and moves the process to step SD9 (No in step SD23).
<Input data existence confirmation step>
The input means 40 confirms whether or not input data has been received. If there is input data, the process returns to step SD2 (Yes in step SD9), and if there is no input data, the process ends (in step SD9). No).
According to the fourth embodiment, since the speech synthesis control device 4 controls speech synthesis of the speech synthesizer 7 based on the utterance start timing and the utterance speed, it is predetermined for each utterance start position at a predetermined utterance start time. The text can be uttered as synthesized speech so that the speech speed becomes.

Hereinafter, as a fifth embodiment, the configuration and operation of the speech synthesis data generation device 5 that generates the utterance start timing and the utterance speed will be described.
(Fifth embodiment)
[Configuration of data generator for speech synthesis]
FIG. 11 is a block diagram showing the configuration of the speech synthesis data generating apparatus according to the fifth embodiment of the present invention.
The speech synthesis data generation device 5 generates a text to be read, an utterance start position, and an utterance start time corresponding to the utterance start position. It mainly includes a timing acquisition unit 52, an utterance speed measurement unit 53, a voice synthesis data generation unit 54, a voice synthesis data storage unit 55, and an output unit 56.

The sound signal input means 50 converts sound waves into sound signals as electric signals and inputs them. In addition, although the signal which converts a sound wave is demonstrated here as an electrical signal, it is not restricted to this, For example, an optical signal may be sufficient.
The voice recognition means 51 recognizes the voice signal input by the voice signal input means 50 and generates text. The voice recognition means 51 performs A / D conversion on a voice signal input as an analog signal, converts it into a voice signal as a digital signal, and then refers to the spectrum of the voice as a word standard model to characterize the voice signal. The phoneme is recognized by extracting and converting the characteristic element into a character suitable for the characteristic. Then, the speech recognition means 51 refers to the word dictionary or grammar dictionary, recognizes the recognized phonemes as words and sentences, and generates text.

  The utterance start timing acquisition unit 52 acquires the utterance start time of the text generated by the speech recognition unit 51 and the utterance start time corresponding to the utterance start position as the utterance start timing. As the utterance start character, any character may be used. In particular, the utterance start character is obtained when the utterance is started from the silent state, and the utterance start time is obtained at that time.

The utterance speed measurement means 53 starts measuring the utterance speed from the time when the utterance start timing acquisition means 52 acquires it as the utterance start character. In this case, the speech rate is determined from the relationship between the time between the time of any phoneme and the time corresponding to the number of phonemes, corresponding to the time of the phonemes extracted as a feature by the speech recognition means 51. Can be calculated.
Further, the speech rate measuring unit 53 measures the speech rate of the speech input by the input unit 50 in correspondence with the text generated by the speech recognition unit 51. The utterance speed measuring means 53 can also calculate the utterance speed from the time from when the utterance start timing acquisition means 52 acquires the utterance start character to the time until the silent state and the number of characters in between.

The voice synthesizing data generating means 54 synthesizes the text generated by the voice recognizing means 51, the utterance start timing acquired by the utterance start timing acquiring means 52, and the utterance speed measured by the utterance speed measuring means 53. Generate as business data. Here, description will be given as speech synthesis data as one packaged stream, but only text is separated, and the speech start timing and speech speed are used as speech synthesis data, or each is separated separately. It may be the case.
The voice synthesis data storage unit 55 stores the voice synthesis data generated by the voice synthesis data generation unit 54.

  The output means 56 uses the voice synthesis data stored in the voice synthesis data storage means 55 as the voice synthesis control device 1, 2 or 4 of the first, second or fourth embodiment, or the third embodiment. To the voice synthesizer 3. Note that the voice synthesis data generation device 5 is not limited to being installed at the same location as the voice synthesis control device 1, 2, or 4 or the voice synthesis device 3, but is installed remotely and transmits and receives via a communication line and broadcast radio waves. You may be able to do it. When installed in the same location, the speech synthesis control device 1, 2 or 4 or the speech synthesis device 3 and the speech synthesis data generation device 5 may or may not be built in the same housing. I do not care.

[Operation of data generator for speech synthesis]
FIG. 12 is a flowchart showing the operation of the speech synthesis data generating apparatus according to the fifth embodiment of the present invention.
When receiving the sound wave, the sound signal input means 50 converts the sound wave into a sound signal as an electric signal and inputs the sound signal (step SE1). The utterance start timing acquisition means 52 acquires the utterance start timing (step SE2), and the utterance speed measurement means 53 measures the utterance speed (step SE3). Further, the voice recognition means 51 recognizes the voice signal and generates a text (step SE4). The above steps SE2 to SE4 are repeated until there is no speech signal (step SE5). If there is no speech signal (step SE6), the speech synthesis data generating means 54 determines the text, speech start timing, and speech rate as speech synthesis data. (Step SE6) and stored in the speech synthesis data storage means 55 (step SE7).

According to the fifth embodiment, the speech synthesis data generation means 5 can acquire text, speech start timing, and speech rate as packaged speech synthesis data.
In this embodiment, the voice recognition means 51, the utterance start timing acquisition means 52, and the utterance speed measurement means 53 have been described as separate functional elements. However, the functions of the utterance start timing acquisition means and the utterance speed measurement means are described as voice. The recognition means 51 may be provided.

  The means for realizing the speech synthesis control function installed in the speech synthesis control devices 1, 2 and 4 of the first, second and fourth embodiments or the speech synthesis device 3 of the third embodiment are: This can be realized by causing a general computer to execute a program and operating an arithmetic device or a storage device in the computer. This program (speech synthesis control program) can be distributed via a communication line, or can be distributed by writing on a recording medium such as a CD-ROM.

  Similarly, the speech synthesis data generation device 5 for reading in the fifth embodiment is realized by causing a general computer to execute a program and operating an arithmetic device and a storage device in the computer. Can do. The program (speech synthesis data generation program) can be distributed via a communication line, or can be distributed by writing on a recording medium such as a CD-ROM.

  Further, although the speech synthesis control devices 1, 2 and 4 of the first, second and fourth embodiments have been described as separate from the speech synthesis device 7, the speech synthesis device of the third embodiment. 3, the speech synthesis control devices 1, 2, and 4 may be mounted on the speech synthesis device 7. Further, in the speech synthesizer 3 of the third embodiment, each means for realizing the speech synthesis control function and the speech synthesizer 39 may be separated.

In each embodiment, the video signal of the moving image has been described as the synchronization target signal. However, the present invention is not limited to this, and may be an audio signal, for example. The synchronization target signal is content that is arranged in time series, such as a video signal or an audio signal. Therefore, even when a still image is slide-displayed, it can be included in the synchronization target signal in the sense that the display of the still image is switched in time series. However, in the case of slide display, a synchronization target signal including a still image and a setting value specifying how many seconds this still image is displayed may be used.

  Further, since the synchronization signal only needs to have a time series, it can also be the time itself. In this case, for example, it is useful in a case where a predetermined amount of synthesized speech is uttered by quick chat within an arbitrary time. For example, a predetermined amount of news is output as synthesized speech during a one-minute news broadcast. In this case, when the title news is broadcast for 10 seconds within 1 minute and the corresponding detailed explanation is broadcast for 20 seconds, the first 10 seconds is slower than the standard speech rate. It can be expected to be used for applications such as outputting synthesized speech by chatting, and attracting the interest of the audience, and outputting detailed explanations with a lot of content with synthesized speech a little faster than the standard speech rate.

It is a block diagram which shows the structure of the speech synthesis control apparatus which concerns on 1st embodiment of this invention. It is a figure explaining the concept of the utterance start character of the speech synthesis control apparatus which concerns on embodiment of this invention. It is a figure explaining the concept of the speech rate of the speech synthesis control apparatus which concerns on embodiment of this invention, (a) is an illustration figure in the case of designating with a speech rate, (b) is based on a certain range of speech It is an illustration figure in the case of designating by time, (c) is an illustration figure in the case of calculating from the utterance start timing, and (d) is an illustration figure in the case of designating by a ratio with a predetermined standard utterance speed. It is. It is a flowchart which shows operation | movement of the speech synthesis control apparatus which concerns on 1st embodiment of this invention. It is a block diagram which shows the structure of the speech synthesis control apparatus which concerns on 2nd embodiment of this invention. It is a flowchart which shows operation | movement of the speech synthesis control apparatus which concerns on 2nd embodiment of this invention. It is a block diagram which shows the structure of the speech synthesizer which concerns on 3rd embodiment of this invention. It is a flowchart which shows operation | movement of the speech synthesizer which concerns on 3rd embodiment of this invention. It is a block diagram which shows the structure of the speech synthesis control apparatus which concerns on 4th embodiment of this invention. It is a flowchart which shows operation | movement of the speech synthesis control apparatus which concerns on 4th embodiment of this invention. It is a block diagram which shows the structure of the data generation apparatus for speech synthesis which concerns on 5th embodiment of this invention. It is a flowchart which shows operation | movement of the data generation apparatus for speech synthesis which concerns on 5th embodiment of this invention.

Explanation of symbols

1, 2, 4 Speech synthesis control device 3, 7 Speech synthesis device 5, 6 Speech synthesis data generation device 10, 20, 30 Speech synthesis data input means 11, 21, 31 Speech synthesis data storage means 12, 22, 32 Analysis analysis means 13, 23, 33, 44 Speech start timing acquisition means 14, 24, 34, 45 Speech synthesis start control means 15, 35, 46 Speech speed determination means 36 Speech speed measurement means 37 Speech speed conversion control means 38 Speed conversion means 40 Input means 41 Text storage means 42 Speech start timing storage means 43 Speech speed timing storage means 47 Speech speed control means 50 Speech signal input means 51 Speech recognition means 52 Speech start timing acquisition means 53 Speech speed measurement means 54 Speech synthesis Data generation means 55 Speech synthesis data storage means 56 Output means

Claims (7)

A preset utterance start position indicating the start position of speech synthesis of the text when the text to be read out is synthesized and uttered as synthesized speech, and a start time indicating the start time of uttering the text as synthesized speech are preset. A speech synthesis control device for controlling the speech synthesizer to synthesize the text and utter as synthesized speech based on the utterance start time,
Utterance start timing acquisition means for acquiring the utterance start position and the utterance start time as an utterance start timing;
Speech synthesis start control means for controlling the start of speech of the speech synthesizer so that the speech synthesizer synthesizes the text from the utterance start position and utters the text as synthesized speech from the utterance start time;
A speech synthesis control device comprising: An utterance speed determining means for determining an utterance speed and a format of the utterance as a synthesized voice for each predetermined utterance range indicating a range of speech synthesis of the text starting from the utterance start position;
An utterance speed control means for outputting the utterance speed determined by the utterance speed determination means to the speech synthesizer and controlling the speech synthesis of the speech synthesizer so as to be the utterance speed determined for each utterance range; ,
The speech synthesis control device according to claim 1, comprising: An utterance speed determining means for determining an utterance speed and a format of the utterance as a synthesized voice for each predetermined utterance range indicating a range of speech synthesis of the text starting from the utterance start position;
Utterance speed measuring means for measuring the utterance speed of synthesized speech uttered by the speech synthesizer;
The utterance speed determined by the utterance speed determination means is compared with the utterance speed measured by the utterance speed measurement means, and the utterance speed of the synthesized speech uttered by the speech synthesizer is determined by the utterance speed determination means. Speech speed conversion control means for outputting a speech speed control signal for converting the speech speed of the synthesized speech by the speech synthesizer so as to approach the spoken speech speed;
Based on the speech speed control signal output from the speech speed conversion control means, the speech speed is adjusted so that the speech speed of the synthesized speech uttered by the speech synthesizer approaches the speech speed determined by the speech speed determination means. A speech rate conversion means for outputting the converted synthesized speech;
The speech synthesis control device according to claim 1, comprising:   The time required for the utterance of the utterance range calculated by the utterance speed determining means calculated from the utterance start time of the utterance start position of the utterance range and the next utterance start time of the next utterance start position of starting the next utterance, and The format expressed based on the number of characters in the utterance range, the format expressed based on the number of utterance characters per unit time, the format expressed based on the utterance time required for utterance in the utterance range, or the standard utterance speed The speech synthesis control device according to claim 2 or 3, wherein the speech synthesis control device is determined as an utterance speed of any one of the formats represented based on a ratio to. A speech synthesis data generation device for generating the text to be read to provide to the speech synthesis control device according to claim 1, the speech start position, and the speech start time corresponding to the speech start position,
Speech recognition means for recognizing speech and generating the text;
Utterance start timing acquisition means for acquiring, as the utterance start timing, the utterance start position of the text generated by the speech recognition means and the utterance start time corresponding to the utterance start position;
Speech synthesis data storage means for storing the text generated by the speech recognition means and the speech start timing acquired by the speech start timing acquisition means as speech synthesis data;
A data generating apparatus for speech synthesis, comprising: A preset utterance start position indicating the start position of speech synthesis of the text when the text to be read out is synthesized and uttered as synthesized speech, and a start time indicating the start time of uttering the text as synthesized speech are preset. A speech synthesis control method for controlling a speech synthesizer to synthesize the text and utter as synthesized speech based on the utterance start time,
An utterance start timing acquisition step of acquiring the utterance start position and the utterance start time as an utterance start timing;
A speech synthesis start control step of controlling the start of speech of the speech synthesizer so that the speech synthesizer synthesizes the text from the utterance start position and utters the text as synthesized speech from the utterance start time;
A speech synthesis control method comprising: A preset utterance start position indicating the start position of speech synthesis of the text when the text to be read out is synthesized and uttered as synthesized speech, and a start time indicating the start time of uttering the text as synthesized speech are preset. In order to control the speech synthesizer to synthesize the text and utter as synthesized speech based on the utterance start time,
An utterance start timing acquisition means for acquiring the utterance start position and the utterance start time as an utterance start timing;
Speech synthesis start control means for controlling the start of speech of the speech synthesizer so as to cause the speech synthesizer to synthesize the text from the utterance start position and to utter the text as synthesized speech from the utterance start time;
A speech synthesis control program characterized by being made to function as.

Images