JP4039620B2 - Speech synthesis apparatus and speech synthesis program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
  The present invention synthesizes speech from text data, particularly text data transmitted by data broadcasting and text broadcasting.SoundThe present invention relates to a voice synthesizer and a voice synthesis program.
[0002]
[Prior art]
Conventionally, for example, a speech synthesis method (Patent Document 1) can be used as a method of receiving text data transmitted from a transmission side by data broadcasting or text broadcasting on the reception side and generating high-quality speech synthesis data. This method (method, device according to this method) is included in the text data by referring to the speech synthesis database (included in the transmitting device) from the text data input to the transmitting device. By generating additional information such as the duration and pitch of the phoneme that is being transmitted, and transmitting this additional information to the receiving side together with the text data. The high-quality synthesized voice data is generated by synthesizing the text data with reference to (included in the side device).
[0003]
By using this method, even if a high load calculation (high load calculation) is required when synthesizing the speech on the receiving device (speech synthesizer), it was generated on the transmitting side (broadcasting station side) The load of the high load calculation is reduced by the additional information, and high-quality speech synthesis can be realized quickly (fast processing speed) by the receiving device (speech synthesizer).
[0004]
[Patent Document 1]
Japanese Patent Application Laid-Open No. 5-210395
[0005]
[Problems to be solved by the invention]
However, in the conventional method “speech synthesis method”, the same text data is often repeatedly transmitted from the transmission side to the reception side (for example, transmission by the data carousel method in data broadcasting), and the text data is received by the reception device on the reception side. Each time a message is received, it is necessary to synthesize speech, resulting in poor processing efficiency, and it becomes difficult to maintain high-quality speech synthesis. Further, in the conventional “speech synthesis method”, it is necessary to provide the same speech synthesis database between the transmission side device and the reception side device. For this reason, there is a problem that it is necessary to maintain constantly so that the same database can be maintained between the transmission side and the reception side.
[0006]
  Therefore, the object of the present invention is to solve the problems of the conventional technique described above, and to generate (maintain) high-quality synthesized speech data without having to maintain the same speech synthesis database on the transmission side and the reception side. )It is possibleSoundTo provide a voice synthesizer and a voice synthesis program.
[0007]
[Means for Solving the Problems]
  In order to achieve the above-described object, the present invention has the following configuration.
  The speech synthesizer according to claim 1, which is a speech synthesizer for synthesizing text data, a text data input means for inputting the text data, and a text data storage means for storing the text data as stored text data. And the text data input for each unit by the text data input means as input text data, compared with the stored text data, if the input text data does not match the stored text data, the input text data New text data discriminating means for discriminating as newly input new text data and storing it as stored text data in the text data storing means, and provided for voice synthesis of the new text dataInclude at least word and feature data for each speech synthesis unitUsing the voice synthesis data storage means for storing the voice synthesis data and the voice synthesis data stored in the voice synthesis data storage means, the new text data is voice-synthesized into synthesized voice data. Speech synthesis means; text data selection means for selecting input text data for each unit input by the text data input means; synthesized speech data storage means for storing synthesized speech data synthesized by the speech synthesis means; Synthetic speech data output means for outputting synthesized speech data corresponding to the input text data selected by the text data selection means, and speech synthesis data used for speech synthesis by the speech synthesizerWordsWhen,Of features corresponding to each wordReference information in association with the storage position of the speech synthesis data in the speech synthesis data storage means on the storage medium is generated and output to the synthesized speech data storage means when speech synthesis is performed by the speech synthesis means. Reference information generation and output means, and the synthesized speech data output means,in frontSelected by the text data selection meansA feature amount corresponding to a phrase included in the input text data is read from the speech synthesis data storage unit using the reference information, and is generated using the read feature amount.The synthesized voice data is output to a speaker.
[0008]
  According to this configuration, text data is input by the text data input means. The new text data discriminating means compares the stored text data storing the text data with the newly input text data, and if the input text data does not match the stored text data, the input text data is newly input. The new text data is determined and stored as stored text data in the text data storage means.The speech synthesis data storage means is used when speech synthesis of new text data, and stores speech synthesis data including at least words and feature data for each speech synthesis unit.When the new text data discriminating means discriminates it as new text data, the voice synthesizing means uses the voice synthesizing data stored in the voice synthesizing data storage means, and the new text data is synthesized by voice. Synthetic data. The synthesized speech data is recorded until the input text data is selected by the text data selecting means.SpeculatorIf the input text data is selected, the synthesized speech data corresponding to the input text data is synthesized speech data.OutOutput by force means. Note that the case where the input text data does not match the stored text data refers to a case where the input text data is considered not to match unless even part of the punctuation marks match, that is, in the new text data determination means, As long as the input text data between the punctuation marks does not completely match the stored text data, it is regarded as new text data. In addition, the data for speech synthesis used when the speech synthesis unit performs speech synthesis by the reference information generation / output unit.WordsWhen,Of features corresponding to each wordReference information associated with the storage position on the storage medium in the voice synthesis data storage means is generated and output to the synthesized voice data storage means. This reference information is then used,The synthesized speech data output means reads out the feature quantity corresponding to the phrase included in the input text data selected by the text data selection means from the speech synthesis data storage means and generates it using the read feature quantity. Synthesized voice dataOutput to the speaker. The storage position on the storage medium corresponds to, for example, time information attached on the storage medium. The reference information is obtained by associating a word or phoneme of speech synthesis data with a storage position on a storage medium.
[0009]
  The speech synthesizer according to claim 2 is the speech synthesizer according to claim 1, further comprising text data receiving means for receiving the text data by at least one of data broadcasting and text broadcasting. And
[0010]
  According to such a configuration, text data is input by the text data receiving means by at least one of data broadcasting and text broadcasting. In other words, by providing this text data receiving means, the speech synthesizer can be said to be a data receiver with a speech synthesizer function, and is broadcast by normal data broadcasting and displayed on the receiving display device. Text information such as telop and text data such as text news are converted (synthesized) into synthesized voice data and output.
[0011]
  The speech synthesizer according to claim 3 is the speech synthesizer according to claim 1 or 2, wherein the synthesized speech data recording is performed.SpeculatorWhen the synthesized speech data is stored in a stage, the synthesized speech data compressing means for compressing the synthesized speech data into compressed synthesized speech data, and the text data selecting meansSelected bySelectedinputWhen the text data corresponds to the compressed synthesized voice data, a compressed synthesized voice data decompressing unit is provided for decompressing the compressed synthesized voice data.
[0012]
  According to such a configuration, the synthesized voice data is compressed by the synthesized voice data compression means.SpeculatorWhen it is stored in the stage, it is compressed so that the amount of data is reduced. Note that the synthetic audio data compression method in the synthetic audio data compression means may be any compression method other than the MPEG-2 method. The compressed synthesized voice data compressed by the synthesized voice data compressing means corresponds to the text data selecting means.inputWhen text data is selected, it is decompressed by the compressed synthesized speech data decompressing means.
[0013]
  The speech synthesis program according to claim 4, in order to synthesize text data, the computer synthesizes text data input means for inputting the text data, text data storage means for storing the text data as stored text data, The text data input for each unit by the text data input means is compared as the input text data with the stored text data. When the input text data does not match the stored text data, the input text data is newly input. New text data discriminating means for discriminating as new text data and storing it in the text data storage means as stored text data; provided for speech synthesis of the new text dataInclude at least word and feature data for each speech synthesis unitVoice synthesis data storage means for storing voice synthesis data, and voice synthesis data synthesized using the voice synthesis data stored in the voice synthesis data storage means to produce synthesized voice data Synthesis means, text data selection means for selecting input text data for each unit input by the text data input means, synthesized speech data storage means for storing synthesized speech data synthesized by the speech synthesis means, and the text data selection Synthesized speech data output means for outputting synthesized speech data corresponding to the input text data selected by the means, speech synthesis data used for speech synthesis by the speech synthesis meansWordsWhen,Of features corresponding to each wordReference information in association with the storage position of the speech synthesis data in the speech synthesis data storage means on the storage medium is generated and output to the synthesized speech data storage means when speech synthesis is performed by the speech synthesis means. Functioning as reference information generation / output means, and the synthesized voice data output means,in frontSelected by the text data selection meansA feature amount corresponding to a phrase included in the input text data is read from the speech synthesis data storage unit using the reference information, and is generated using the read feature amount.The synthesized voice data is output to a speaker.
[0014]
  According to this configuration, text data is input by the text data input means. The new text data discriminating means compares the stored text data storing the text data with the newly input text data, and if the input text data does not match the stored text data, the input text data is newly input. The new text data is determined and stored as stored text data in the text data storage means.The speech synthesis data storage means is used when speech synthesis of new text data, and stores speech synthesis data including at least words and feature data for each speech synthesis unit.When the new text data discriminating means discriminates it as new text data, the voice synthesizing means uses the voice synthesizing data stored in the voice synthesizing data storage means, and the new text data is synthesized by voice. Synthetic data. The synthesized speech data is recorded until the input text data is selected by the text data selecting means.SpeculatorIf the input text data is selected, the synthesized speech data corresponding to the input text data is synthesized speech data.OutOutput by force means. Speech synthesis data used for speech synthesis by the speech synthesis means in the reference information generation / output meansWordsWhen,Of features corresponding to each wordReference information associated with the storage position on the storage medium in the voice synthesis data storage means is generated and output to the synthesized voice data storage means. This reference information is then used,The synthesized speech data output means reads out the feature quantity corresponding to the phrase included in the input text data selected by the text data selection means from the speech synthesis data storage means and generates it using the read feature quantity. Synthesized voice dataOutput to the speaker.
[0021]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.
(Configuration of speech synthesizer)
FIG. 1 is a block diagram of a speech synthesizer. As shown in FIG. 1, the speech synthesizer 1 includes a text data input unit 3, a text data receiving unit 5, a new text data determination unit 7, a text data storage unit 9, a speech synthesis unit 11, A synthesis database 13, a synthesized voice data storage / output unit 15, and a text data selection unit 17 are provided. The speech synthesizer 1 is connected to a speaker 19 that outputs synthesized speech data synthesized by speech synthesis, and includes a remote controller 21 for operating the speech synthesizer 1.
[0022]
This speech synthesizer 1 synthesizes text data transmitted by data broadcasting, text broadcasting, etc., or text data (text data input for each unit) input via a keyboard (not shown) or the like. Output. Further, in this speech synthesizer 1, unlike the conventional speech synthesizer (data receiver with a speech synthesizer function), even if text data and additional information are not transmitted from the transmission side, When synthesizing text data using speech synthesis data, the processing speed can be improved by generating reference information (details will be described later), and once synthesized speech data is synthesized. By storing and reusing, high-quality synthesized speech can be output (maintained) without reducing the processing capability of the apparatus 1. The text data input for each unit refers to the input unit input to the device 1 at a time, or a unit that can be grouped together, for example, each phrase unit or each sentence unit. is there.
[0023]
The text data input unit 3 is configured by a keyboard, a mouse, or the like, and inputs text data. The text data input unit 3 may be configured by an OCR that can acquire text data from existing handwritten text, or may be configured by a disk drive that handles a disk that stores text data. However, it may be configured by an input terminal that is input from the outside via a communication network or the like. The text data input unit 3 corresponds to the text data input means described in the claims.
[0024]
The text data receiving unit 5 is composed of an antenna (parabolic antenna) capable of receiving data broadcasting, text broadcasting, etc., a detection circuit, and the like, and data of current analog television broadcasting and high-definition satellite broadcasting transmitted from a broadcasting station. This is to detect text data included in various digital data transmitted through channels (data broadcasting, text broadcasting, etc.). The text data receiving unit 5 corresponds to the text data receiving means described in the claims.
[0025]
In this embodiment, the text data input unit 3 and the text data receiving unit 5 are provided with a function for setting an input unit (input unit setting function) in advance, and are input to the device 1 at a time. It is possible to set input units (sentence units separated by reception time) and units that can be grouped together, for example, phrase units and sentence units.
[0026]
The new text data discriminating unit 7 stores the text data input from the text data input unit 3 and the text data receiving unit 5 in the text data storage unit 9 as stored text data and stores the text data in the text data storage unit 9. The stored text data is compared with the subsequently input text data (input text data). When the input text data does not match the stored text data, it is determined as new text data and output to the speech synthesizer 11. is there. If it is determined as new text data, it is stored in the text data storage unit 9 as stored text data (new stored text data).
[0027]
In addition, the case where the input text data does not match the stored text data is to divide the input text data by punctuation marks, and if there are any different parts in the sentence between these punctuation marks, it is considered that they do not match. In this case, only the unmatched part is output to the speech synthesizer 11.
[0028]
For example, the stored text data is “Today is a good weather, it will be a comfortable day”, and the input text data is “Today is a good weather and the temperature is expected to be 28 degrees C.” The text data is decomposed and stored in the text data storage unit 9 as “Today is a good weather” and “It will be a comfortable day”, and the input text data “Today is a good weather”, “ The temperature is expected to be 28 degrees. ”When these are decomposed and compared, the difference between the stored text data and the input text data (new part), that is,“ the temperature is expected to be 28 degrees. ” Is determined as new text data and output to the speech synthesizer 11, and the common word “Today is good weather” is not output to the speech synthesizer 11.
[0029]
The text data storage unit 9 is constituted by a semiconductor memory, a hard disk, or the like. The text data obtained by the text data input unit 3 and the text data receiving unit 5 is determined by the new text data determination unit 7. It memorizes based on. The new text data determination unit 7 and the text data storage unit 9 correspond to the new text data determination means described in the claims.
[0030]
The speech synthesizer 11 searches the speech synthesis database 13 for speech synthesis of the new text data determined by the new text data determiner 7, and generates synthesized speech data. Note that the speech synthesis method (means) in the speech synthesizer 11 may be of any type, and for example, the means disclosed in Japanese Patent Laid-Open No. 10-49193 may be used.
[0031]
The speech synthesizer 11 stores synthesized speech data together with synthesized speech data in association with speech synthesis data used when speech synthesis of new text data is performed and a storage position of the speech synthesis database 13 on the storage medium. Reference information generating / outputting means (not shown) for outputting to the output unit 15 is provided.
[0032]
  The reference information is, for example, “Yonekura_noun_file 2” for the word “Yonekura”.3_30 ms to 70 ms_XXX ”,“ noun ”is a kind of information related to the part of speech or the like of a word or phoneme, and“ file 23"Indicates the number of a sentence containing the noun" Yonekura "in the speech synthesis database 13, and" 30ms to 70ms "is file 23(Sentence) indicates the time of utterance, and “XXX” indicates the feature amount of the word “Yonekura”.
[0033]
  In the speech synthesizer 11, when the means disclosed in Japanese Patent Laid-Open No. 10-49193 is used, it takes a long time for speech synthesis to search the speech synthesis database 13. If the reference information generated by the generation output means (not shown) is used,Synthesized speechThe data storage output unit 15 is not necessarily required to store the synthesized voice data, and the storage capacity can be reduced. The speech synthesizer 11 corresponds to the speech synthesizer described in the claims.
[0034]
The speech synthesis database 13 is composed of a large-capacity hard disk or the like, and is used when speech synthesis unit 11 synthesizes new text data. The speech synthesis unit 13 is composed of words or phonemes. ), Speech synthesis data in which speech time, feature amount, and the like are collected is stored. The speech synthesis database 13 corresponds to the speech synthesis data storage means described in the claims.
[0035]
The synthesized speech data storage output unit 15 corresponds to a so-called data buffer, stores the synthesized speech data synthesized by the speech synthesizer 11, and based on the selection data output from the text data selection unit 17. , Which outputs the stored synthesized voice data, and comprises synthesized voice data output means 15a and synthesized voice data storage means 15b.
[0036]
The synthesized voice data output means 15a is responsible for storage and output control of synthesized voice data and reference information. The synthesized voice data synthesized by the voice synthesizer 11 is stored in the synthesized voice data storage means 15b and text Based on the selection data output from the data selection unit 17, synthesized voice data corresponding to the selection data is output to the speaker 19.
[0037]
The synthesized speech data storage unit 15b stores the synthesized speech data synthesized by the speech synthesizer 11 and the reference information according to the control of the synthesized speech data output unit 15a.
The synthesized voice data storage / output unit 15 includes synthesized voice data compression means and compressed synthesized voice data decompression means (not shown).
[0038]
When the synthesized voice data compression means (not shown) stores the synthesized voice data synthesized by the voice synthesizer 11 in the synthesized voice data storage means 15b, the synthesized voice data compression means compresses the data amount of the synthesized voice data and compresses and synthesizes the synthesized voice data. It generates voice data. The compressed synthesized voice data decompressing means (not shown) outputs (reads out) the compressed synthesized voice data stored in the synthesized voice data storage means 15b based on the selection data output from the text data selection unit 17. It is to be thawed.
[0039]
By these synthesized voice data compression means (not shown) and compressed synthesized voice data decompression means (not shown), the storage capacity of the synthesized voice data storage means 15b can be suppressed to a small amount.
[0040]
The text data selection unit 17 corresponds to the voice (synthesized voice data) output from the voice synthesizer 1 based on the infrared signal (control signal) output from the remote controller 21 operated by the user of the voice synthesizer 1. Selects text data.
[0041]
  According to the speech synthesizer 1, text data is input by the text data input unit 3. When the new text data discrimination unit 7 compares the stored text data stored in the text data storage unit 9 with the newly input text data, and the input text data does not match the stored text data, The input text data is determined as newly input new text data. When the new text data discriminating unit 7 discriminates it as new text data, the voice synthesizing unit 11 uses the voice synthesizing data stored in the voice synthesizing database 13 to synthesize the new text data. , Speech synthesis data. Then, until the text data selection unit 17 selects the text data, the voice synthesis data is held in the synthesized voice data storage / output unit 15 and when the text data is selected, the voice synthesis data corresponding to the text data is obtained. It is output from the synthesized voice data storage output unit 15. For this reason,,The synthesized speech data synthesized by the speech synthesizer 11 is stored in the synthesized speech data storage / output unit 15, and only the new text data determined by the new text data determination unit 7 is synthesized by the speech synthesizer 11. Therefore, useless processing (excess speech synthesis) at the time of speech synthesis is reduced, the speech synthesis processing capability of the apparatus 1 can be maintained at a high level, and high-quality synthesized speech data can be generated. .
[0042]
The text data receiving unit 5 of the speech synthesizer 1 receives text data by at least one of data broadcasting and text broadcasting. That is, by providing the text data receiving unit 5, the speech synthesizer 1 can be said to be a data receiver with a speech synthesis function, and is broadcast by normal data broadcasting, and is displayed on the display device on the receiving side. Character information such as displayed telop and text data such as character news are converted (synthesized) into synthesized voice data and output.
[0043]
That is, text data by normal data broadcasting, text broadcasting, or the like is less likely to be frequently replaced while the user (viewer) is using the speech synthesizer (data receiver with speech synthesis function) 1. . Therefore, as soon as new text data that has not been synthesized by the speech synthesizer 11 of the speech synthesizer 1 is received by the text data receiver 5, synthesized speech data synthesized by speech synthesis is generated and synthesized by the synthesized speech data storage output unit 15 When the text data is selected (stored) in the voice data storage means 15b and the text data is selected by the text data selection section 17, the synthesized voice data corresponding to the text data is synthesized. 15 and can be read out by the speaker 19 (can be uttered).
[0044]
(Operation of speech synthesizer)
Next, the operation of the speech synthesizer 1 will be described with reference to the flowchart shown in FIG.
First, when the speech synthesizer 1 is activated, it is determined whether or not text data has been input by at least one of the text data input unit 3 and the text data receiving unit 5 (S1), and waits until it is input (S1, No). If it is input, it is stored as stored text data in the text data storage unit 9 (S2).
[0045]
Then, the stored text data stored in the text data storage unit 9 and the newly input text data are compared and discriminated by the new text data discriminating unit 7, and whether or not the text data is new text data based on the comparison discrimination result. Is determined (S3). When the new text data determination unit 7 determines that the input text data is new text data (S3, Yes), the new text data is output to the speech synthesis unit 11.
[0046]
Then, the speech synthesizer 11 searches the speech synthesis database 13 and synthesizes new text data as speech data (S4). The synthesized voice data and the reference information generated at the time of voice synthesis are output to the synthesized voice data storage output unit 15. When the synthesized voice data and the reference information are input, the synthesized voice data storage / output unit 15 stores the synthesized voice data and the reference information in the synthesized voice data storage unit 15b (S5).
[0047]
Thereafter, it is determined whether or not there is an output request (control signal) of speech (synthesized speech data) from the remote controller 21 (from the user of the speech synthesizer 1) in the text data selection unit 17 (S6). It is waited until it is determined that there is an output request (control signal) of the sound (synthesized sound data) from the remote controller 21 (S6, No), and an output request (control signal) of the sound (synthesized sound data) from the remote controller 21 is received. If it is determined that there is a voice (synthesized voice data) corresponding to the text data in accordance with the output request (control signal), the synthesized voice data output means 15 a of the synthesized voice data storage output unit 15 outputs to the speaker 19. (S7).
[0048]
(Specific examples of speech synthesis of text data by data broadcasting)
A specific example in which the text data receiving unit 5 of the speech synthesizer 1 receives text data by data broadcasting and synthesizes the speech will be described (see FIG. 1 as appropriate). The text data received by the text data receiving unit 5 includes “news”, “weather information”, “sports”, “yen and stock”, “road traffic information”, “welfare”, “recommended information”, This text data is displayed as a “menu screen” on the display screen of a display device (not shown).
[0049]
Assuming that "news", "weather information", "sports", "yen and stock", "road traffic information", "welfare", and "recommended information" have been transmitted from the broadcasting station on the sender side, “News”, “weather information”, “sports”, “yen and stock”, “road traffic information”, “welfare”, “recommended information” are stored in the text data storage unit 9 as stored text data. Subsequent transmitted “news”, “weather information”, “typhoon information”, “sports”, “yen and stock”, “road traffic information”, “welfare”, “recommended information”, “typhoon information” Are determined as new text data by the new text data determination unit 7 and synthesized by the speech synthesis unit 11. The synthesized voice data that has been synthesized by the voice synthesizer 11 is output in order to the synthesized voice data storage output unit 15 and stored in the synthesized voice data storage unit 15b.
[0050]
  If the user of the speech synthesizer 1 selects “yen and stock” with the remote controller 21, the selection data for selecting this “yen and stock” is sent from the text data selection unit 17 to the synthesized speech data storage output unit 15. It is output to the synthesized voice data output means 15a, and in this synthesized voice data output means 15a,Synthesized speechThe synthesized voice data of “yen and stock” stored in the data storage unit 15 b is read and output from the speaker 19.
[0051]
As mentioned above, although this invention was demonstrated based on one Embodiment, this invention is not limited to this.
For example, it is possible to regard the processing of each component of the speech synthesizer 1 as a speech synthesis method that considers each process as one step, or to regard the processing of each component as a speech synthesis program described in a general-purpose computer language. . In these cases, the same effect as the speech synthesizer 1 can be obtained.
[0052]
【The invention's effect】
  Claim 1,4According to the described invention, text data is input, and if this text data is newly input, it is synthesized and stored as synthesized speech data. When text data is selected, synthesized speech data corresponding to the text data is output. Since only new text data is synthesized with speech, there is no need to maintain the same database for speech synthesis between the sender that sent the text data and the receiver that received the text data, without reducing the speech synthesis processing capability. High-quality synthesized speech data can be generated.In addition, since reference information in which the voice synthesis data used for voice synthesis and the storage position on the storage medium in which the voice synthesis data is stored is associated is generated, based on this reference information, Synthetic speech data can be generated, the storage capacity of the storage medium for storing the synthetic speech data can be reduced to a small amount, and the processing for speech synthesis can be reduced.
[0053]
  Claim2According to the described invention, text data by at least one of data broadcasting and text broadcasting is input. That is, it is broadcast by normal data broadcasting, and text information such as telop displayed on the display device on the receiving side and text data such as text news can be output as synthesized voice data.
[0054]
  ClaimTo 3According to the described invention, when the synthesized voice data is stored, the compressed voice data is compressed so as to reduce the amount of data, and the compressed compressed voice data is decompressed when read out, so that the synthesized voice data is stored. The storage capacity of the storage medium to be kept can be reduced to a small amount.
[0055]
According to the fifth aspect of the present invention, the reference information in which the speech synthesis data used for speech synthesis is associated with the storage position on the storage medium in which the speech synthesis data is stored is generated. Based on this reference information, data for speech synthesis can be generated, the storage capacity of the storage medium for storing the synthesized speech data can be reduced to a small amount, and the processing at the time of speech synthesis is reduced. be able to.
[Brief description of the drawings]
FIG. 1 is a block diagram of a speech synthesizer according to an embodiment of the present invention.
FIG. 2 is a flowchart for explaining the operation of the speech synthesizer shown in FIG. 1;
[Explanation of symbols]
1 Speech synthesizer
3 Text data input section
5 Text data receiver
7 New text data discriminator
9 Text data storage
11 Speech synthesis unit
13 Database for speech synthesis
15 Synthetic voice data storage / output unit
15a Synthetic voice data output means
15b Synthetic voice data storage means
17 Text data selection part
19 Speaker
21 Remote control

Claims (4)

A speech synthesizer for speech synthesis of text data,
Text data input means for inputting the text data;
Text data storage means for storing the text data as stored text data;
The text data input for each unit by the text data input means is compared with the stored text data as input text data. When the input text data does not match the stored text data, the input text data is newly New text data determining means for determining as input new text data and storing the stored text data in the text data storing means,
The subjected the new text data in speech synthesis, and speech synthesis data storage means for data words and the feature amount storing data for speech synthesis that is part of at least every speech synthesis unit,
Using the voice synthesis data stored in the voice synthesis data storage means, voice synthesis means for synthesizing the new text data into synthesized voice data; and
Text data selection means for selecting input text data for each unit input by the text data input means;
Synthesized voice data storage means for storing synthesized voice data synthesized by the voice synthesis means;
Synthesized speech data output means for outputting synthesized speech data corresponding to the input text data selected by the text data selection means;
The words and phrases of the speech synthesis data used when speech synthesis is performed by the speech synthesis means, and the storage locations of the speech synthesis data in the speech synthesis data storage means of the feature values corresponding to the respective phrases on the storage medium A reference information generating / outputting unit that outputs the reference information to the synthesized voice data storage unit when the voice synthesizing unit synthesizes the reference information.
The synthesized speech data output means, a feature amount corresponding to words contained before Symbol input text data selected by the text data selection means, reads from the speech synthesis data storage means using said reference information, A speech synthesizer characterized by outputting synthesized speech data generated using the read feature amount to a speaker.   The speech synthesizer according to claim 1, further comprising text data receiving means for receiving the text data by at least one of data broadcasting and text broadcasting. In storing the synthesized speech data in the synthesized speech data SL 憶手 stage, the synthesized speech data compression unit to compress the synthetic speech data obtained by compressing the data amount of the synthesized speech data,
Claim the input text data is selected by said text data selection means in corresponding to the compressed synthesized speech data, characterized by comprising a compressed synthesized speech data decompression means for decompressing the compressed synthesized speech data The speech synthesizer according to claim 1 or 2. In order to synthesize text data,
Text data input means for inputting the text data;
Text data storage means for storing the text data as stored text data;
The text data input for each unit by the text data input means is compared with the stored text data as input text data. When the input text data does not match the stored text data, the input text data is newly New text data discrimination means for discriminating as input new text data and storing it as stored text data in the text data storage means,
The new text data is subjected during the speech synthesizing speech synthesis data storage means for data words and the feature amount storing data for speech synthesis that is part of at least every speech synthesis unit,
Using the voice synthesis data stored in the voice synthesis data storage unit, the new text data is voice-synthesized into synthesized voice data;
Text data selection means for selecting input text data for each unit input by the text data input means;
Synthesized voice data storage means for storing synthesized voice data synthesized by the voice synthesis means;
Synthesized voice data output means for outputting synthesized voice data corresponding to the input text data selected by the text data selection means;
The words and phrases of the speech synthesis data used when speech synthesis is performed by the speech synthesis means, and the storage locations of the speech synthesis data in the speech synthesis data storage means of the feature values corresponding to the respective phrases on the storage medium Is generated as reference information generation and output means for outputting to the synthesized voice data storage means when voice synthesis is performed by the voice synthesis means.
The synthesized speech data output means, a feature amount corresponding to words contained before Symbol input text data selected by the text data selection means, reads from the speech synthesis data storage means using said reference information, A speech synthesis program characterized by outputting synthesized speech data generated using the read feature amount to a speaker.

Images