JP2005208421A - Document read-aloud system and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enable a document read-aloud apparatus to properly analyze distributed character data and correctly read the character data aloud although the document read-aloud apparatus refers to a dictionary having limited capacity by allowing a language analyzing server to analyze the character data distributed from an information providing server, extract a morpheme that the document read-aloud apparatus misreads, and transmit data showing a reading of the extracted morpheme to the document read-aloud apparatus. <P>SOLUTION: A document read-aloud system has: the information providing server 21 which distributes character data; the document read-aloud apparatus which analyzes and reads the character data aloud; and the language analyzing server 31 which extracts a morpheme that the document read-aloud apparatus misreads and transmits the data showing the reading of the extracted morpheme to the document read-aloud apparatus, wherein the document read-aloud apparatus analyzes and reads the character data aloud by using the data showing the reading of the extracted morpheme. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a text reading system and method.

Conventionally, in a navigation apparatus mounted on a vehicle such as an automobile, when an operator such as a driver operates a predetermined input unit to set a destination, the destination and the current position detection processing unit detect the destination. Based on the current position of the vehicle, a route from the current position to the destination is searched, and the searched route is guided. In recent years, a navigation device has been proposed in which character data such as news distributed from an information center is acquired and displayed, and the news or the like is output by voice and read as necessary (for example, Patent Document 1). reference.). Thereby, the operator can grasp news and the like by listening to the read-out voice without gazing at the display screen of the navigation device.
JP 2003-121187 A

  However, in the conventional navigation device, since the character data such as the news is a kana-kana mixed sentence, the phonetic characters such as hiragana and katakana are read-only for reading out the contents of the news etc. Data is required. Therefore, it is necessary for the information center to distribute the phonetic character data together with the character data of the kana-kana mixed sentence, which increases the burden. In addition, the amount of data to be distributed is increased, the communication time is increased, and the communication cost is increased. Furthermore, although it is conceivable to distribute voice data, the voice data has a larger data amount than the character data, so that the communication time is further increased and the communication cost is further increased.

  Recently, however, a system has been developed that analyzes kanji-kana mixed sentences and assigns how to read kanji. However, in this system, a dictionary that requires a large storage capacity such as a dictionary for sentence analysis is used, and it has been difficult to apply to a device having a limited storage capacity such as a navigation device. In addition, in order to apply the system to a device having a limited storage capacity such as a navigation device, it is necessary to limit the capacity of a dictionary such as a sentence analysis dictionary. Because they are often mistakenly read out, they cannot be heard.

  The present invention solves the above-mentioned conventional problems, the linguistic analysis server analyzes the character data distributed from the information providing server, extracts the morpheme in which the reading error occurs in the document reading device, and reads the morpheme Is transmitted to the document reading device, the document reading device properly analyzes the distributed character data while referring to the dictionary having a limited capacity, and is surely analyzing the character data. An object is to provide a text-to-speech reading system and method that can be read aloud.

  Therefore, in the text-to-speech reading system of the present invention, an information providing server that distributes character data, a document reading device that analyzes and reads the character data, and a morpheme that causes a reading error in the document reading device, A language analysis server that transmits data indicating how to read the morpheme to the document reading device, and the document reading device analyzes and reads the character data using the data indicating how to read the morpheme.

  In another sentence reading system of the present invention, the character data is a kanji / kana mixed sentence, and the document reading apparatus analyzes the kanji / kana mixed sentence with reference to a language analysis dictionary and the language analysis dictionary. A speech synthesis processing unit.

  In yet another text-to-speech system according to the present invention, the language analysis server further includes a comparison speech synthesis processing unit equivalent to the speech synthesis processing unit of the document reading device, and the comparison speech synthesis processing unit. A server speech synthesis processing unit with high processing capability is provided, comparing the analysis result of the comparison speech synthesis processing unit and the analysis result of the server speech synthesis processing unit, and extracting the morpheme in which the reading error occurs, Data indicating how to read the morpheme is created.

  In still another text-to-speech system of the present invention, the language analysis server further includes a comparison language analysis dictionary equivalent to the language analysis dictionary of the document reading device, and the comparison language analysis dictionary. A server language analysis dictionary having a large storage capacity and abundant dictionary contents is provided, the comparison speech synthesis processing unit refers to a comparison language analysis dictionary, and the server speech synthesis processing unit is used for server language analysis Browse the dictionary.

  In still another text reading system according to the present invention, the data indicating how to read the morpheme is a list of morphemes in which the reading error occurs and a list of how to read the morpheme.

  In still another text-to-speech system according to the present invention, the data indicating how to read the morpheme is replaced with a phonetic character representing the morpheme in which the reading error occurs in the character data. Character data.

  In the text reading method of the present invention, the information providing server distributes character data, the language analysis server extracts a morpheme in which a reading error occurs in the document reading device, and data indicating how to read the morpheme is read out from the document reading device. The document reading device analyzes and reads out the character data using data indicating how to read the morpheme.

  The language analysis server of the present invention is a language analysis server that transmits data indicating how to read to a vehicle-mounted device that reads out character data with reference to the language analysis dictionary, the server language analysis dictionary, and the vehicle-mounted device A language analysis dictionary for a server, a server speech synthesis processing unit that analyzes the character data with reference to the server language analysis dictionary, and a language analysis dictionary for the vehicle-mounted device that analyzes the character data A speech synthesis processing unit for in-vehicle devices, comparing the analysis result of the speech synthesis processing unit for server and the analysis result of the speech synthesis processing unit for in-vehicle device, and extracting morphemes that cause reading errors; Data indicating how to read the morpheme is created.

  In the text-to-speech device of the present invention, the character data distributed by the information providing server, and the receiving unit that receives the data transmitted from the language analysis server and indicating the reading of the morpheme that causes the reading error in the character data, A language analysis unit that analyzes the character data using data indicating how to read morphemes, and a speech waveform generation unit that creates speech waveform data for reading the character data.

  According to the present invention, the document reading device can properly analyze the distributed character data and read the character data without fail while referring to a dictionary with a limited capacity.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a conceptual diagram showing the configuration of a text-to-speech system according to the first embodiment of the present invention.

  In the figure, 10 is a text reading system. Reference numeral 21 denotes an information providing server, which is a computer provided with arithmetic means such as a CPU and MPU, storage means such as a semiconductor memory, a magnetic disk, and an optical disk, a communication interface, and the like. The information providing server 21 may not be a single computer, but may be a so-called distributed server in which a plurality of computers are organically coupled. Another system may be constructed in the information providing server 21. Furthermore, the information providing server 21 may be one of systems constructed in another computer.

  Reference numeral 11 denotes an in-vehicle device as a document reading device operated by an operator, and is mounted on a vehicle 12 such as a passenger car, a truck, a bus, and a motorcycle. There are actually a large number of in-vehicle devices, but in the present embodiment, for convenience of explanation, the in-vehicle devices 11 are represented. The operator is a driver of the vehicle 12, a passenger, or the like.

  The in-vehicle device 11 includes a calculation device such as a CPU and an MPU, a storage device such as a semiconductor memory, a magnetic disk, and an optical disc, a liquid crystal display, an LED (Light Emitting Diode) display, a display device such as a CRT, a keyboard, a joystick, and a cross. An input device such as a key, a push button, a remote controller, and a touch panel, an audio output device such as a speaker 11a described later, and a transmission / reception device such as a communication interface and a wireless LAN (Local Area Network) device. The in-vehicle device 11 is, for example, a navigation device mounted on the vehicle 12, but a stationary phone, a mobile phone, a PHS (Personal Handy-Phon System) phone, a personal digital assistant, a PDA (Personal Digital Assistant), a personal computer, Any device such as a game machine or a digital TV may be used. And although the said vehicle equipment 11 does not necessarily need to be mounted in the vehicle 12, it demonstrates as what is mounted in the vehicle 12 here.

  Further, the in-vehicle device 11 may have a current position detection device (not shown). When the in-vehicle device 11 is a navigation device, for example, the current position detection device is generally based on GPS (Global Positioning System), geomagnetic sensor, distance sensor, steering sensor, beacon sensor, gyro sensor, etc. Detect position. Further, for example, in the case of a mobile phone, a portable information terminal, etc., generally, based on communication with the base station where the mobile phone, the portable information terminal, etc. are located, the position of the base station is set as the current position. To detect.

  Here, the information providing server 21 and the in-vehicle device 11 are connected to each other via a network 22 so that they can communicate with each other. The network 22 is any communication network such as a wired or wireless public communication network, a dedicated communication network, a mobile phone network, the Internet, an intranet, a LAN, a WAN (Wide Area Network), and a satellite communication network. They may be combined appropriately. Moreover, it may communicate using CS broadcasting or BS broadcasting by a broadcasting satellite, may communicate using terrestrial digital television broadcasting, may communicate using FM multiplex broadcasting, Road and vehicle communication may be performed using an optical beacon, a radio beacon, an antenna, or the like installed on the side of the road or on the road.

  In the present embodiment, description will be made assuming that the in-vehicle device 11 is connected to the network 22 by communicating with the access point 23 connected to the network 22 by a wireless LAN system. The access point 23 is, for example, a wireless LAN adapter, and uses 2.4 [GHz] ISM band, 5.2 [GHz] band, 19 [GHz] band quasi-millimeter wave or the like, or uses infrared rays. Thus, communication with the in-vehicle device 11 is performed by a wireless LAN system. The wireless LAN system in this case is, for example, IEEE802.11a, IEEE802.11b, IEEE802.11e, IEEE802.11g, or Bluetooth used as a home wireless system. Good. The access point 23 may be located at any place, may be an operator's residence, or may be used by a large number of people such as a gas station, a restaurant, a parking lot. It may be arranged in a parking lot or the like.

  The information providing server 21 distributes various types of information such as news, weather forecasts, company information, stock market conditions, and exchange market conditions as text data, that is, character data. Since the character data is originally data for displaying various information as text on the display device of the in-vehicle device 11, it is usually configured as a kanji-kana mixed text. Since the amount of character data is smaller than that of voice data, it can also be transmitted using TV radio waves.

  Moreover, the vehicle-mounted apparatus 11 in this Embodiment is the receiving part which receives the data which the said information provision server 21 and the language analysis server 31 transmit, the speech synthesis process part 14, the language analysis dictionary 15, and a user from a functional viewpoint. It has a document reading processing unit 13 including a definition language analysis dictionary 16 and a voice data generation dictionary 17. The text-to-speech processing unit 13 reads out the text data by analyzing the text of the kanji / kana mixed text in the text data received from the information providing server 21 and assigning the reading style of the text. .

  A language analysis server 31 is communicably connected to the network 22. The language analysis server 31 is a computer provided with arithmetic means such as a CPU and MPU, storage means such as a semiconductor memory, a magnetic disk, and an optical disk, a communication interface, and the like. The language analysis server 31 may not be a single computer but may be a so-called distributed server in which a plurality of computers are organically coupled. Another system may be constructed in the language analysis server 31. Furthermore, the language analysis server 31 may be configured integrally with another computer, or may be one of systems constructed in another computer. For example, the language analysis server 31 may be configured integrally with the information providing server 21 or may be one of the systems constructed in the information providing server 21.

  Here, the language analysis server 31 includes a server speech synthesis processing unit 32, a server language analysis dictionary 33, an in-vehicle device speech synthesis processing unit 34 as a comparison speech synthesis processing unit, and a functional viewpoint. It has an in-vehicle device language analysis dictionary 35 as a comparative language analysis dictionary. The in-vehicle device speech synthesis processing unit 34 has the same processing capability as the speech synthesis processing unit 14 in the in-vehicle device 11, and the in-vehicle device language analysis dictionary 35 is in the language analysis of the in-vehicle device 11. It has a storage capacity equivalent to that of the dictionary 15 and a dictionary content equivalent. The server speech synthesis processing unit 32 has a higher processing capability than the speech synthesis processing unit 14 of the in-vehicle device 11, and the server language analysis dictionary 33 is the language analysis dictionary 15 of the in-vehicle device 11. It has a larger storage capacity and rich dictionary content. For example, the storage capacity of the language analysis dictionary 15 of the in-vehicle device 11 is about 1 [MB] or less, whereas the storage capacity of the server language analysis dictionary 33 is about 100 [MB].

  Then, the language analysis server 31 analyzes the kanji-kana mixed sentence in the character data received from the information providing server 21 and assigns how to read kanji and symbols. In this case, the server speech synthesis processing unit 32 and the on-vehicle device speech synthesis processing unit 34 analyze the sentences of kanji / kana mixed sentences in the same character data. The server speech synthesis processing unit 32 analyzes the sentence with reference to the server language analysis dictionary 33 having a high processing capacity and a large storage capacity and abundant dictionary contents. It can be analyzed and how to read kanji can be assigned accurately. On the other hand, the in-vehicle device speech synthesis processing unit 34 has a processing capability equivalent to that of the speech synthesis processing unit 14 of the in-vehicle device 11, and also has a storage equivalent to the language analysis dictionary 15 of the in-vehicle device 11. Since the sentence is analyzed with reference to the language analysis dictionary 35 for the vehicle-mounted device having the capacity and the equivalent dictionary content, it can be analyzed only with the same accuracy as the language analysis dictionary 15 of the vehicle-mounted device 11, Misreading kanji occurs. Therefore, the language analysis server 31 compares the analysis result of the server speech synthesis processing unit 32 with the analysis result of the on-vehicle device speech synthesis processing unit 34, and extracts morphemes having different readings. A morpheme is the smallest meaningful language unit.

  Here, since the extracted morpheme can be considered as a morpheme in which a reading error occurs in the in-vehicle device 11, the morpheme and the morpheme that are analyzed by the server speech synthesis processing unit 32 are used as data indicating how to read the morpheme. The reading list data is transmitted to the in-vehicle device 11. Then, the in-vehicle device 11 stores the received list data in the user-defined language analysis dictionary 16. It should be noted that the data amount of the list data is normally considered to be about several [KB]. Then, the speech synthesis processing unit 14 of the in-vehicle device 11 refers to the language analysis dictionary 15 and the user-defined language analysis dictionary 16 to analyze the sentence of the kanji / kana mixed sentence in the character data received from the information providing server 21. And assign how to read kanji and symbols. Therefore, the sentence can be analyzed accurately, and how to read kanji and symbols can be definitely assigned. Thus, since the in-vehicle device 11 analyzes and reads out the character data received from the information providing server 21 using the data indicating how to read the morpheme, the character data can be read out without fail.

  Next, the configuration of the document reading processing unit 13 of the in-vehicle device 11 will be described in detail.

  FIG. 2 is a diagram showing the configuration of the in-vehicle device according to the first embodiment of the present invention.

  Here, the case where the said vehicle-mounted apparatus 11 receives the character news delivered as character data from the information provision server 21 via the network 22 and the access point 23 is demonstrated. The text news is a sentence in a mixed kanji kana that “the yen exchange rate in the Tokyo foreign exchange market as of 9:00 am on the 5th is 1 dollar = 109 yen 57-60 yuan”. As shown in FIG. 2, the voice synthesis processing unit 14 of the in-vehicle apparatus 11 includes a language analysis unit 14a, a voice waveform generation unit 14b, and a voice waveform reproduction unit 14c. In addition, a speaker 11a is connected to the audio waveform reproducing unit 14c as an audio output device.

  Then, the language analysis unit 14a refers to the two language analysis dictionaries, the language analysis dictionary 15 and the user-defined language analysis dictionary 16, to decompose the character news into morphemes. It is assumed that the list data received from the language analysis server 31 is temporarily registered in the user-defined language analysis dictionary 16. In this case, the text news is “/ 5 / day / am / 9 / hour / present / of / Tokyo / foreign / exchange / market / of / yen / market / ha /, / 1 / dollar / = / 109 /. Yen / 57 / ˜ / 60 / sen /./ ”.

  Further, the language analysis unit 14a refers to the two language analysis dictionaries and creates a character string for reading each morpheme. When there are a plurality of readings for one morpheme, reference is made to preset priority data. In this case, the text string of how to read the text news is “/ someday / gozen / kuji / genzai / no / tokyo / gaikoku / kaze / shijo / no / en / soba / wa”. /, /Ichi/Doru//Hyakukyu/En/Gonana/Kara/Rokuji/Sen/./ ”.

  Then, the speech waveform generation unit 14b refers to the speech data generation dictionary 17 and creates speech waveform data from the reading character string created by the language analysis unit 14a. The speech waveform data is a binary code composed of 0 and 1. In addition, the audio waveform reproduction unit 14c reproduces the audio waveform data created by the audio waveform generation unit 14b and outputs the audio waveform data from the speaker 11a. As a result, the text news will be read out as "Someday, there will be some sort of text that will be read someday". .

  Next, the operation of the text-to-speech reading system 10 having the above configuration will be described.

  FIG. 3 is a first flowchart showing the operation of the text-to-speech system according to the first embodiment of the present invention. FIG. 4 is a second flowchart showing the operation of the text-to-speech system according to the first embodiment of the present invention. is there.

  Here, the operation when the operator receives the text news requested by operating the in-vehicle device 11 from the information providing server 21 will be described. In this case, the operator operates the in-vehicle device 11 to cause the in-vehicle device 11 to perform an operation of receiving text news of a desired genre from the information providing server 21 and reading it out. Here, it is assumed that the operator has selected text news regarding the foreign exchange market conditions. Then, the in-vehicle device 11 transmits a text news request for requesting transmission of text news to the information providing server 21 via the access point 23 and the network 22.

  On the other hand, the information providing server 21 is in a state waiting to receive a text news request waiting for a text news request from the in-vehicle device 11, and continuously determines whether or not a text news request has been received. Then, when a text news request is received from the in-vehicle device 11, the information providing server 21 selects and determines text news of a genre corresponding to the content of the text news request. Here, it is determined as character news to be transmitted character news regarding the foreign exchange market conditions. Subsequently, the information providing server 21 transmits a character news analysis request for analyzing character news to the language analysis server 31 via the network 22.

  Here, the language analysis server 31 is waiting for a character news analysis request to be waiting for a character news analysis request, and continuously determines whether or not a character news analysis request has been received. When receiving the character news analysis request from the information providing server 21, the language analysis server 31 transmits a character news request for requesting transmission of the character news to be analyzed to the information providing server 21.

  In this case, whether or not the information providing server 21 is waiting for the character news request waiting for the character news request from the language analysis server 31 after transmitting the character news analysis request, and has received the character news request? Judgment is continuously made. When receiving a text news request from the language analysis server 31, the information providing server 21 transmits the determined text news, that is, text news regarding foreign exchange market conditions, to the language analysis server 31. Further, the information providing server 21 notifies the language analysis server 31 of the transmission destination of the analysis result. In this case, the destination of the analysis result is the in-vehicle device 11, and the ID number, IP (Internet Protocol) address, MAC (Media Access Control) address, etc. of the in-vehicle device 11 are transmitted to the language analysis server 31. The

  On the other hand, after transmitting the text news request, the language analysis server 31 is waiting for the text news reception waiting for the text news, and continuously determines whether or not the text news has been received. When character news is received from the information providing server 21, character news analysis processing is executed. In this case, the character news is analyzed by the server speech synthesis processing unit 32 and the on-vehicle device speech synthesis processing unit 34, respectively. Then, the language analysis server 31 compares the analysis result of the server speech synthesis processing unit 32 with the analysis result of the in-vehicle device speech synthesis processing unit 34, and extracts morphemes having different readings. Subsequently, the language analysis server 31 creates list data of the extracted morphemes and how to read the extracted morphemes analyzed by the server speech synthesis processing unit 32, and uses the list data as an analysis result. It transmits to the vehicle equipment 11. Then, the language analysis server 31 transmits an analysis end notification to the effect that the character news analysis process has ended, to the information providing server 21 and ends the process.

  In addition, whether the information providing server 21 is in a state waiting for the analysis of the character news waiting for the analysis end notification from the language analysis server 31 after notifying the transmission destination of the analysis result, and whether the analysis end notification has been received. Judgment is continuously made. When receiving the analysis end notification from the language analysis server 31, the information providing server 21 transmits the determined character news, that is, the character news regarding the foreign exchange market conditions, to the in-vehicle device 11 and ends the processing. To do.

  On the other hand, after transmitting the text news request, the in-vehicle device 11 is in a state waiting for the analysis result and the text news reception waiting for the text news, whether the analysis result has been received, and the text news. It is continuously determined whether or not it has been received. When the analysis result is received from the language analysis server 31, the in-vehicle device 11 temporarily registers the extracted morphemes and list data of how to read them in the user-defined language analysis dictionary 16. Subsequently, when the character news about the foreign exchange market situation is received from the information providing server 21, the in-vehicle device 11 executes a character news analysis process. In this case, the language analysis unit 14a of the in-vehicle device 11 analyzes the character news with reference to the language analysis dictionary 15 and the user-defined language analysis dictionary 16 in which the list data is temporarily registered. To do. Subsequently, the speech waveform generation unit 14b of the in-vehicle device 11 refers to the speech data generation dictionary 17 and performs speech waveform generation processing for generating speech waveform data from the reading character string created by the language analysis unit 14a. Execute. Subsequently, the audio waveform reproduction unit 14c of the in-vehicle device 11 performs audio waveform reproduction processing for reproducing the audio waveform data generated by the audio waveform generation unit 14b and outputting the data from the speaker 11a. As a result, the text news is definitely read out. And the said vehicle-mounted apparatus 11 deletes the said list data temporarily registered from the dictionary 16 for user-defined language analysis, and complete | finishes a process.

Next, a flowchart will be described.
Step S <b> 1 The in-vehicle device 11 transmits a text news request to the information providing server 21.
Step S2 The information providing server 21 waits for a text news request from the in-vehicle device 11.
Step S3 The information providing server 21 determines whether or not a text news request has been received. If a text news request has been received, the process proceeds to step S4. If a text news request has not been received, the process returns to step S2.
Step S4 The information providing server 21 determines text news to be transmitted to the in-vehicle device 11.
Step S5 The information providing server 21 transmits a text news analysis request.
Step S6 The language analysis server 31 waits for a character news analysis request.
Step S7: The language analysis server 31 determines whether or not a character news analysis request has been received. If a character news analysis request has been received, the process proceeds to step S8. If a character news analysis request has not been received, the process returns to step S6.
Step S8: The language analysis server 31 transmits a character news request to be analyzed.
Step S9 The information providing server 21 waits for a text news request from the language analysis server 31.
Step S10: The information providing server 21 determines whether a text news request has been received. If a text news request has been received, the process proceeds to step S11. If a text news request has not been received, the process returns to step S9.
Step S11 The information providing server 21 transmits text news.
Step S12 The information providing server 21 notifies the transmission destination of the analysis result.
Step S13 The language analysis server 31 stands by for character news.
Step S14: The language analysis server 31 determines whether text news has been received. If character news has been received, the process proceeds to step S15. If character news has not been received, the process returns to step S13.
Step S15 The language analysis server 31 executes a character news analysis process.
Step S16: The language analysis server 31 transmits the analysis result.
Step S17 The language analysis server 31 transmits an analysis end notification and ends the process.
Step S18 The information providing server 21 stands by for the end of character news analysis.
Step S19: The information providing server 21 determines whether or not an analysis end notification has been received. If an analysis end notification has been received, the process proceeds to step S20. If an analysis end notification has not been received, the process returns to step S18.
Step S20 The information providing server 21 transmits the text news and ends the process.
Step S21 The in-vehicle device 11 waits for the end of the text news reception.
Step S22 The in-vehicle device 11 determines whether or not an analysis result has been received. When the analysis result is received, the process proceeds to step S23, and when the analysis result is not received, the process proceeds to step S24.
Step S <b> 23 The in-vehicle device 11 temporarily registers the list data in the user-defined language analysis dictionary 16.
Step S24: The in-vehicle device 11 determines whether or not text news has been received. If character news has been received, the process proceeds to step S25, and if character news has not been received, the process returns to step S21.
Step S25 The in-vehicle device 11 executes a character news analysis process.
Step S26 The in-vehicle device 11 executes a voice waveform generation process.
Step S27 The in-vehicle device 11 reads out the text news (voice waveform reproduction process).
Step S28 The in-vehicle device 11 deletes the list data from the user-defined language analysis dictionary 16 and ends the process.

  Next, the character news analysis process executed by the language analysis server 31 will be described in detail. Here, a case will be described in which the text news is text news related to the foreign exchange market situation and is a sentence in a kanji-kana mixed sentence.

  FIG. 5 is a flowchart showing the operation of the character news analysis process of the language analysis server in the first embodiment of the present invention.

  First, the language analysis server 31 receives character news to be analyzed from the information providing server 21. Here, the text news is text news regarding the foreign exchange market conditions, and as shown in FIG. 5 (a), "the yen exchange rate of the Tokyo foreign exchange market as of 9:00 am on the 5th is 1 dollar = 109 yen This is a kanji-kana mixed sentence "57-60 sen." Then, the language analysis server 31 causes the server speech synthesis processing unit 32 and the in-vehicle device speech synthesis processing unit 34 to analyze the character news.

  Here, the server speech synthesis processing unit 32 refers to the server language analysis dictionary 33 and the priority data having a high processing capacity and having a large storage capacity and abundant high-precision dictionary contents. Since the text of the text news is analyzed, it can be analyzed with high accuracy, and how to read kanji can be assigned accurately. Then, the server speech synthesis processing unit 32 divides the text of the text news into morphemes. In this case, as shown in FIG. 5B, the text news is “/ 5 / day / am / 9 / hour / present / of / Tokyo / foreign / exchange / market / of / yen / market / /, /1/dollar/=/109/yen/57/˜/60/sen/./ ”. Subsequently, the server speech synthesis processing unit 32 assigns a reading to each morpheme and creates a phonetic character string as a character string for the reading. As shown in FIG. 5 (c), the phonetic character string is “/ someday / gozen / kuji / genzai / no / tokyo / gaikoku / kaseki / shijo / no / en / Soba / Wa /, / Ichi / Doru // Hyakukyu / En / Goryuna / Kara / Rokuji / Sen /./ ”.

  Further, the on-board device speech synthesis processing unit 34 has a lower processing capability than the server speech synthesis processing unit 32, and has a small storage capacity and not abundant but low-precision dictionary contents. Since the text of the text news is analyzed with reference to the priority data, it cannot be analyzed with high accuracy, and a reading error of kanji occurs. The in-vehicle device speech synthesis processing unit 34 has the same processing capacity as the in-vehicle device 11 speech synthesis processing unit 14, and the in-vehicle device language analysis dictionary 35 and the storage capacity and dictionary contents of the priority data are also described above. This is equivalent to the language analysis dictionary 15 and priority data of the in-vehicle device 11.

  Then, the on-vehicle device speech synthesis processing unit 34 divides the text of the text news into morphemes. In this case, as shown in FIG. 5 (d), the text news is “/ 5 / day / am / 9 / hour / present / of / Tokyo / foreign / exchange / market / of / yen / market / /, /1/dollar/=/109/yen/57/˜/60/sen/./ ”. Subsequently, the on-vehicle device speech synthesis processing unit 34 assigns a reading to each morpheme, and creates a phonetic character string as a character string for the reading. As shown in FIG. 5 (e), the phonetic character string is “/ go / nichi / gozen / kyu / ji / genzai / no / tokyo / gaikoku / kaseki / ichiba”. /No/en/soba/wa/,/ichi/duru/ikol/hyakukyu/en/junyana//rokuju/sen/./ ".

  Subsequently, the language analysis server 31 compares the phonetic character string created by the server voice synthesis processing unit 32 with the phonetic character string created by the on-vehicle device voice synthesis processing unit 34, and the reading is different. Extract morphemes. The morphemes with different readings in the two phonetic character strings are the morphemes surrounded by circles in FIGS. 5 (f) and 5 (g). The morpheme extracted in this way can be considered as a morpheme that causes a reading error in the in-vehicle device 11. Therefore, the language analysis server 31 creates morphemes with different readings and list data of the readings. Note that the reading included in the list data is the reading assigned by the server speech synthesis processing unit 32 that is considered to be accurate. That is, the list data is list data of morphemes in which reading errors occur in the in-vehicle device 11 and correct readings thereof. As shown in FIG. 5 (h), the list data includes ““ 5 days ”: / someday,“ 9 o'clock ”: / lottery,“ market ”: / jyo,“ = ”: / (silence), From “~”: /, “60”: Rokuji ”.

  Then, the language analysis server 31 transmits the list data to the in-vehicle device 11 and ends the process.

Next, a flowchart will be described.
Step S15-1: Character news to be analyzed is received from the information providing server 21.
Step S15-2: The server speech synthesis processing unit 32 divides the document into morphemes.
Step S15-3: The server speech synthesis processing unit 32 assigns a reading to each morpheme and creates a phonetic character string.
Step S15-4 The on-vehicle device speech synthesis processing unit 34 divides the sentence into morphemes.
Step S15-5: The on-vehicle device speech synthesis processing unit 34 assigns a reading to each morpheme and creates a phonetic character string.
Step S15-6: The two phonetic character strings are compared and different morphemes are extracted.
Step S15-7: List data of morphemes in which reading errors occur.
Step S15-8 The list data is transmitted to the in-vehicle device 11 and the process is terminated.

  Thus, in this embodiment, the character data transmitted from the information providing server 21 to the in-vehicle device 11 is analyzed by the language analysis server 31. Here, the language analysis server 31 has a high processing capacity, a server speech synthesis processing unit 32 that analyzes sentences by referring to the server language analysis dictionary 33 having a large storage capacity and abundant dictionary contents; The in-vehicle device language analysis dictionary 35 has the same processing capability as the speech synthesis processing unit 14 of the in-vehicle device 11 and has the same storage capacity and equivalent dictionary content as the language analysis dictionary 15 of the in-vehicle device 11. And a speech synthesis processing unit 34 for in-vehicle devices that analyzes the sentence. Then, the language analysis server 31 compares the results of analyzing the text data of the character data by the server speech synthesis processing unit 32 and the on-vehicle device speech synthesis processing unit 34, and a reading error occurs in the on-vehicle device 11. A list data of the morpheme and its correct reading is created, and the list data is transmitted to the in-vehicle device 11. Therefore, the voice synthesis processing unit 14 of the in-vehicle device 11 analyzes the text of the character data with reference to the list data, so that the character data can be read out without fail. Therefore, the in-vehicle device 11 appropriately analyzes the distributed character data and uses the character synthesis data while using the speech synthesis processing unit 14 with limited processing capability and the language analysis dictionary 15 with limited capacity. You can definitely read the data.

  Further, since the character data transmitted from the information providing server 21 to the in-vehicle device 11 is character data of a kanji / kana mixed sentence, the data amount is smaller than character data consisting only of phonetic characters such as hiragana and katakana. Therefore, the communication load for transmitting from the information providing server 21 to the in-vehicle device 11 is small, the communication time can be shortened, and the communication cost can be reduced. Furthermore, since the list data has a small data amount, the communication load for transmitting from the language analysis server 31 to the in-vehicle device 11 is small, the communication time can be shortened, and the communication cost can be reduced. it can. Furthermore, it is not necessary to increase the capacity of the user-defined language analysis dictionary 16 of the in-vehicle device 11 that temporarily registers the list data.

  Next, a second embodiment of the present invention will be described. In addition, about the thing which has the same structure as 1st Embodiment, the description is abbreviate | omitted by providing the same code | symbol. The description of the same operation and the same effect as those of the first embodiment is also omitted.

  In the present embodiment, the language analysis server 31 does not transmit a list of morphemes that cause reading errors in the in-vehicle device 11 and correct readings thereof to the in-vehicle device 11, and represents the correct reading of the morphemes included in the list data. The character data replaced with the phonetic character is transmitted to the in-vehicle device 11. In this case, the user-defined language analysis dictionary 16 in the in-vehicle device 11 can be omitted. In addition, since the structure about another point is the same as that of the said 1st Embodiment, the description is abbreviate | omitted.

  Next, the operation of the text-to-speech system 10 in the present embodiment will be described.

  FIG. 6 is a first flowchart showing the operation of the text-to-speech system in the second embodiment of the present invention, and FIG. 7 is a second flowchart showing the operation of the text-to-speech system in the second embodiment of the present invention. is there.

  Similar to the first embodiment, an operation when the operator receives from the information providing server 21 the text news requested by operating the in-vehicle device 11 will be described. The operation from the time when the information providing server 21 transmits the text news to the language analysis server 31 and notifies the destination of the analysis result is the same as that in the first embodiment, and the description thereof is omitted. To do.

  In the present embodiment, the information providing server 21 terminates the process when the language analysis server 31 is notified of the transmission destination of the analysis result. On the other hand, after transmitting the text news request, the language analysis server 31 is waiting for the text news reception waiting for the text news, and continuously determines whether or not the text news has been received. When character news is received from the information providing server 21, character news analysis processing is executed. In this case, the character news is analyzed by the server speech synthesis processing unit 32 and the on-vehicle device speech synthesis processing unit 34, respectively. Then, the language analysis server 31 compares the analysis result of the server speech synthesis processing unit 32 with the analysis result of the in-vehicle device speech synthesis processing unit 34, and extracts morphemes having different readings. Subsequently, the language analysis server 31 creates list data of the extracted morphemes and how to read the extracted morphemes analyzed by the server speech synthesis processing unit 32. Subsequently, the language analysis server 31 replaces the character corresponding to the extracted morpheme in the character news with a phonetic character representing how to read the morpheme, and executes a phonetic character conversion process. Then, the character news reorganized by the phonogram conversion processing, that is, the phonogram conversion processed character news is transmitted to the in-vehicle device 11 as an analysis result, and the processing ends.

  On the other hand, after transmitting the text news request, the in-vehicle device 11 is in a state waiting for the text news reception waiting for the text news, and continuously determines whether or not the text news has been received. When the phonetic characterized processed character news as an analysis result is received from the language analysis server 31, the in-vehicle device 11 executes a character news analysis process. In this case, the language analysis unit 14a of the in-vehicle device 11 refers to the language analysis dictionary 15 and analyzes the phonogram conversion processed character news. Subsequently, the speech waveform generation unit 14b of the in-vehicle device 11 refers to the speech data generation dictionary 17 and performs speech waveform generation processing for generating speech waveform data from the reading character string created by the language analysis unit 14a. Execute. Subsequently, the audio waveform reproduction unit 14c of the in-vehicle device 11 performs audio waveform reproduction processing for reproducing the audio waveform data generated by the audio waveform generation unit 14b and outputting the data from the speaker 11a. As a result, the phonetic-letterized character news is definitely read out. And the said vehicle-mounted apparatus 11 complete | finishes a process.

Next, a flowchart will be described.
Step S31 The in-vehicle device 11 transmits a text news request to the information providing server 21.
Step S32 The information providing server 21 waits for a text news request from the in-vehicle device 11.
Step S33 The information providing server 21 determines whether or not a text news request has been received. If a text news request has been received, the process proceeds to step S34. If a text news request has not been received, the process returns to step S32.
Step S34 The information providing server 21 determines the text news to be transmitted to the in-vehicle device 11.
Step S35 The information providing server 21 transmits a character news analysis request.
Step S36 The language analysis server 31 waits for a text news request.
Step S37: The language analysis server 31 determines whether or not a character news analysis request has been received. If a character news analysis request has been received, the process proceeds to step S38. If a character news analysis request has not been received, the process returns to step S36.
Step S38: The language analysis server 31 transmits a character news request to be analyzed.
Step S39 The information providing server 21 waits for a text news request from the language analysis server 31.
Step S40 The information providing server 21 determines whether or not a text news request has been received. If a text news request has been received, the process proceeds to step S41. If a text news request has not been received, the process returns to step S39.
Step S41 The information providing server 21 transmits text news.
Step S42: The information providing server 21 notifies the transmission destination of the analysis result and ends the process.
Step S43 The language analysis server 31 stands by for character news.
Step S44: The language analysis server 31 determines whether text news has been received. If character news has been received, the process proceeds to step S45, and if character news has not been received, the process returns to step S43.
Step S45 The language analysis server 31 executes a character news analysis process.
Step S46: The language analysis server 31 transmits the analysis result and ends the process.
Step S47 The in-vehicle device 11 waits for the end of the text news reception.
Step S48 The in-vehicle device 11 determines whether or not text news has been received. If character news has been received, the process proceeds to step S49. If character news has not been received, the process returns to step S47.
Step S49 The in-vehicle device 11 executes a character news analysis process.
Step S50 The in-vehicle device 11 executes a voice waveform generation process.
Step S51 The in-vehicle device 11 reads out the text news (voice waveform reproduction process) and ends the process.

  Next, the character news analysis process executed by the language analysis server 31 will be described in detail. Here, as in the first embodiment, a case will be described in which the text news is text news related to foreign exchange market conditions and is a sentence in a kanji / kana mixed sentence.

  FIG. 8 is a flowchart showing the operation of the character news analysis process of the language analysis server in the second embodiment of the present invention.

  First, the language analysis server 31 receives character news to be analyzed from the information providing server 21. Here, the text news is text news regarding the foreign exchange market conditions, and as shown in FIG. 8A, “the yen exchange rate of the Tokyo foreign exchange market as of 9:00 am on the 5th is 1 dollar = 109 yen. This is a kanji-kana mixed sentence "57-60 sen." Note that the morphemes that cause erroneous reading in the in-vehicle device 11 and the operations up to the creation of the correct reading list data are the same as those in the first embodiment, and thus the description thereof is omitted.

  Subsequently, the language analysis server 31 replaces the character corresponding to the extracted morpheme in the character news, that is, the character corresponding to the morpheme in which the reading error occurs, with the phonetic character representing the correct reading of the morpheme included in the list data. The phonogram conversion process is executed. In this case, as shown in FIG. 8 (i), the text news is as follows: “<someday> AM <Lottery> The current yen exchange rate of Tokyo Foreign Exchange <Shoji> is 1 dollar <-> 109 yen57 < "> <Rokuji> Sen." Then, the language analysis server 31 transmits the phonogram conversion processed character news as the character news reorganized by the phonogram conversion processing to the in-vehicle device 11 and ends the processing.

Next, a flowchart will be described.
Step S45-1 Character news to be analyzed is received from the information providing server 21.
Step S45-2: The server speech synthesis processing unit 32 divides the document into morphemes.
Step S45-3: The server speech synthesis processing unit 32 assigns a reading to each morpheme and creates a phonetic character string.
Step S45-4: The on-vehicle device speech synthesis processing unit 34 divides the sentence into morphemes.
Step S45-5: The on-vehicle device speech synthesis processing unit 34 assigns a reading to each morpheme and creates a phonetic character string.
Step S45-6: The two phonetic character strings are compared, and different morphemes are extracted.
Step S45-7: List data of morphemes in which reading errors occur.
Step S45-8: The phonogram conversion-processed character news is transmitted to the in-vehicle device 11 and the process is terminated.

  As described above, in the present embodiment, the language analysis server 31 creates a list of morphemes that cause reading errors in the in-vehicle device 11 and correct readings thereof, and the characters corresponding to the morphemes that cause reading errors in character news. Is replaced with a phonetic character conversion process that replaces the phonetic character with a phonetic character that represents the correct reading of the morphemes contained in the list data. Then, the phonetic characterized processed character news is transmitted to the in-vehicle device 11. Therefore, the speech synthesis processing unit 14 of the in-vehicle device 11 can definitely read the character data. Moreover, since the user-defined language analysis dictionary 16 can be omitted, the configuration of the in-vehicle device 11 can be simplified.

  The character data transmitted from the language analysis server 31 to the in-vehicle device 11 is basically character data of a kana-kana mixed sentence, and only the characters corresponding to the morpheme in which reading errors occur are replaced with phonograms. As a result, the amount of data can be minimized. Therefore, the communication load for transmitting from the language analysis server 31 to the in-vehicle device 11 is small, the communication time can be shortened, and the communication cost can be reduced.

  In the first and second embodiments, the case where the character data distributed from the information providing server 21 is so-called pull-type content distributed in response to a request from the in-vehicle device 11 has been described. However, the character data may be so-called push type content or broadcast type content that is distributed from the information providing server 21 to the in-vehicle device 11 without a request from the in-vehicle device 11.

  The present invention is not limited to the above-described embodiment, and various modifications can be made based on the spirit of the present invention, and they are not excluded from the scope of the present invention.

It is a conceptual diagram which shows the structure of the text reading system in the 1st Embodiment of this invention. It is a figure which shows the structure of the vehicle-mounted apparatus in the 1st Embodiment of this invention. It is a 1st flowchart which shows operation | movement of the text-to-speech reading system in the 1st Embodiment of this invention. It is a 2nd flowchart which shows operation | movement of the text-to-speech system in the 1st Embodiment of this invention. It is a flowchart which shows the operation | movement of the character news analysis process of the language analysis server in the 1st Embodiment of this invention. It is a 1st flowchart which shows operation | movement of the text reading system in the 2nd Embodiment of this invention. It is a 2nd flowchart which shows operation | movement of the text-to-speech system in the 2nd Embodiment of this invention. It is a flowchart which shows the operation | movement of the character news analysis process of the language analysis server in the 2nd Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Text-to-speech reading system 11 In-vehicle device 14 Speech synthesis processing unit 15 Language analysis dictionary 21 Information providing server 31 Language analysis server 32 Server speech synthesis processing unit 33 Server language analysis dictionary 34 In-vehicle device speech synthesis processing unit 35 In-vehicle device Language analysis dictionary

Claims (9)

(A) an information providing server for distributing character data;
(B) a document reading device that analyzes and reads out the character data;
(C) having a language analysis server that extracts morphemes in which reading errors occur in the document reading device and transmits data indicating how to read the morphemes to the document reading device;
(D) The text-to-speech device reads out the text data by analyzing the text data using data indicating how to read the morpheme. (A) The character data is a kanji / kana mixed sentence,
(B) The text-to-speech system according to claim 1, wherein the text-to-speech device includes a language analysis dictionary and a speech synthesis processing unit that analyzes the kanji-kana mixed text with reference to the language analysis dictionary. (A) The language analysis server includes a comparative speech synthesis processing unit equivalent to the speech synthesis processing unit of the document reading device, and a server speech synthesis processing unit having a higher processing capability than the comparative speech synthesis processing unit. Prepared,
(B) Compare the analysis result of the comparison speech synthesis processing unit with the analysis result of the server speech synthesis processing unit, extract the morpheme in which the reading error occurs, and create data indicating how to read the morpheme The text-to-speech system according to claim 2. (A) The language analysis server includes a comparative language analysis dictionary equivalent to the language analysis dictionary of the document reading device, and a server having a larger storage capacity and abundant dictionary contents than the comparison language analysis dictionary. Language analysis dictionary
(B) The text-to-speech system according to claim 3, wherein the comparative speech synthesis processing unit refers to a comparative language analysis dictionary, and the server speech synthesis processing unit refers to a server language analysis dictionary. The text-to-speech system according to any one of claims 1 to 4, wherein the data indicating how to read the morpheme is morpheme in which the reading error occurs and list data of how to read the morpheme. The data indicating how to read the morpheme is character data obtained by replacing a character corresponding to the morpheme in which the reading error occurs in the character data with a phonetic character indicating how to read the morpheme. The text-to-speech system described in the section. (A) The information providing server distributes character data,
(B) the language analysis server extracts a morpheme in which a reading error occurs in the document reading device, and transmits data indicating how to read the morpheme to the document reading device;
(C) The text-to-speech device analyzes the text data using data indicating how to read the morpheme and reads the text. (A) a language analysis server that transmits data indicating how to read to a vehicle-mounted device that reads out character data with reference to a language analysis dictionary;
(B) a language analysis dictionary for servers;
(C) an on-vehicle device language analysis dictionary;
(D) a server speech synthesis processing unit that analyzes the character data with reference to the server language analysis dictionary;
(E) having a speech synthesis processing unit for in-vehicle device that analyzes the character data with reference to the dictionary for language analysis for in-vehicle device,
(F) Comparing the analysis result of the speech synthesis processing unit for the server and the analysis result of the speech synthesis processing unit for the in-vehicle device, extracting a morpheme in which a reading error occurs, and creating data indicating how to read the morpheme A language analysis server characterized by (A) a receiving unit that receives character data distributed by the information providing server and data indicating how to read a morpheme that is read by the language analysis server and that causes a reading error in the character data;
(B) a language analysis unit that analyzes the character data using data indicating how to read the morpheme;
(C) A text reading device having a voice waveform generation unit for generating voice waveform data for reading the character data.

Images