JP4237860B2 - Data reading apparatus and recording medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention provides, for example, a data reading apparatus and fixed data for speech synthesis of a plurality of fixed data when there are a plurality of fixed data having a specific format composed of a plurality of data such as weekly weather forecast data. The present invention relates to a recording medium on which a computer program is read.
[0002]
[Prior art]
The conventional data reader reads out standard data, creates a sentence by embedding the standard data in a specific sentence example, converts it into an intermediate notation that expresses reading, accent, and intonation, synthesizes speech, and outputs it through a speaker, etc. It was something to do. In this case, since the sentence is created by embedding the fixed form data in the same sentence example, if there are several fixed form data, only the data part is different and the monotonous sentence is repeated. For this reason, if there are a lot of standard data, the listener is often frustrated or tired while listening, and it often takes a long time to obtain all necessary information.
[0003]
[Problems to be solved by the invention]
In recent years, the performance of CPUs of computers, the higher integration of memories, and the lowering of their costs have progressed rapidly, and computers with small size, high performance, and large capacity have become increasingly smaller. Therefore, it is possible to incorporate a voice synthesizer that requires a high-performance CPU and a large-capacity memory, and various devices that provide information by voice are rapidly spreading.
[0004]
With the rapid spread of devices incorporating a speech synthesizer, more people have more opportunities to receive information provided by speech synthesized speech. However, since it is necessary to convey information by voice in order one by one, the amount of information that can be received per unit time is very small compared to visual information. Therefore, even if the required information can be read with a glance, if it is received as audio information, it will take a long time to obtain the required information because the information will be read out in order by voice. .
[0005]
An example in which it appears most prominently is information represented by a table having a lot of fixed data. The information shown in the table can be seen in a short time when viewed with the eye, and it is easy to obtain only the necessary part of the information. If you are at the end of, you must endure listening to unnecessary information, and even if you need all the information, the same pattern sentences will be repeated many times, so the more information you have The listening time is long and the text is monotonous, so I'm tired of listening to it on the way.
[0006]
Services that provide information by voice were initially limited to devices such as telephones that have only a voice output device, but recently, even if they have a display screen such as a display, such as car navigation, Since it is dangerous to watch the screen while driving, there are an increasing number of devices that provide information not only on the screen but also by voice.
[0007]
In order to further spread the use of devices that provide speech information by speech synthesis, there is a demand for technical means for transmitting a number of standard data in a short time and without getting tired.
[0008]
[Means for Solving the Problems]
In the present invention, a plurality of standard data having a specific format composed of one or more data that can be represented in a table is not read repeatedly by repeating the same sentence, but the sentence is changed, and repeated parts can be omitted. This makes it possible to read out information by voice without getting bored in a shorter time by omitting such parts or changing the data order.
[0009]
A data reading device according to the present invention is a position for embedding each data in the fixed data in the data reading device that synthesizes a sentence based on fixed data having a specific format composed of one or more data and outputs the sound. Including The content of the fixed character part of the text is different A sentence example storage unit for storing a plurality of sentence examples, and reading the fixed form data, Read Example sentences to be embedded for each fixed form data In the prescribed way Each of the data in the standard data is embedded in a predetermined position, and a text synthesizing unit is provided for synthesizing the text.
[0010]
With such a configuration, it is possible to listen to voice information without getting tired because information can be heard by voice output of different sentences for each fixed data, instead of repeating the same sentence only by changing the data portion as in the conventional case.
[0011]
Further, the data reading apparatus according to the present invention is a data reading apparatus for synthesizing sentences based on fixed data having a specific format composed of one or more data and outputting the sound, and a fixed data storage for storing a plurality of the fixed data Part and a position for embedding each data in the standard data The content of the fixed character part of the text is different From the sentence example storage unit for storing a plurality of sentence examples, and the sentence example storage unit, Read Example sentences to be embedded for each fixed form data In the prescribed way A sentence example selection unit that selects and, based on the fixed form data and the selected sentence example, embeds each data in the fixed form data at a predetermined position of the selected sentence example, and combines a document, A sentence analysis unit that analyzes the synthesized document and converts it into an intermediate notation for expressing speech, and a speech synthesis unit that synthesizes and outputs speech based on the intermediate notation, respectively. .
[0012]
With such a configuration, it is possible to listen to voice information without getting tired because information can be heard by voice output with different sentences for each fixed data, instead of repeating the same sentence only by changing the data part as in the prior art.
[0013]
Further, the data reading apparatus according to the present invention is a data reading apparatus for synthesizing sentences based on fixed data having a specific format composed of one or more data and outputting the sound, and a fixed data storage for storing a plurality of the fixed data Part and a position for embedding each data in the standard data The content of the fixed character part of the text is different A sentence example storage unit that stores a plurality of sentence examples in an intermediate notation format for expressing speech, and the sentence example storage unit, Read Example sentences to be embedded for each fixed form data In the prescribed way Based on the selected sentence example selection part, the fixed form data conversion part for converting each data in the fixed form data into an intermediate notation, each data in the converted fixed form data and the selected sentence example, the fixed form data Each of the data is embedded in a predetermined position of the selected sentence example, and a sentence synthesizing unit for synthesizing a document and a voice synthesizing unit for synthesizing and outputting speech based on the synthesized sentence are provided. .
[0014]
With this configuration, since the sentence example is in the intermediate notation from the beginning, when reading the fixed form data, it is only necessary to convert the fixed form data into the intermediate notation, which eliminates the need for parsing and makes it easy to change to the intermediate notation. Therefore, the processing speed can be improved.
[0015]
Also, the data reading apparatus according to the present invention includes a duplication check flag indicating whether or not each piece of the standard data performs a duplication check, and the sentence example storage unit collectively represents the duplication portion of the standard data. A sentence example is stored, and the sentence example selection unit has a function of checking a duplicate part of the fixed form data based on the duplication check flag and selecting a sentence example for collecting the corresponding duplicate parts.
[0016]
With this configuration, if there is the same data of the same type in the standard data, you can create a sentence by grouping them together, so you can shorten the sentence and at the same time change the sentence without getting bored. , Listen to audio information in a shorter time.
[0017]
Further, the data reading apparatus according to the present invention includes a data set flag in which each data of the fixed form data indicates a set of data to be read out together in the fixed form data, and the example storage unit stores the data set in a data set unit. A plurality of sentence examples including only the above data are stored, and the preceding sentence example selection unit has a function of reading all fixed form data at once and selecting a corresponding sentence example for a specified data set unit.
[0018]
With this configuration, it is possible to first output high-priority data in a plurality of fixed data in a voice and then output the subsequent data as detailed data. If you don't need it later, you can listen to it later.
[0019]
Further, a computer-readable recording medium recording a program to be executed by a computer according to the present invention includes a step of synthesizing a sentence based on fixed data having a specific format composed of one or more data, and outputting the sentence as a voice A computer-readable recording medium storing a program for causing a computer to execute the step of including a position for embedding each data in the standard data The content of the fixed character part of the text is different Storing a plurality of sentence examples; reading the fixed form data; Read Example sentences to be embedded for each fixed form data In the prescribed way It is a computer-readable recording medium on which a program for causing a computer to execute a step of selecting and changing, embedding each data in the fixed form data at a predetermined position, and synthesizing a sentence.
[0020]
This configuration realizes a data reading device that allows users to listen to voice information without getting bored because they can listen to information by voice output with different sentences for each standard data, instead of repeating the same sentence just by changing the data part as before. I can do it.
[0021]
Further, a computer-readable recording medium recording a program to be executed by a computer according to the present invention includes a step of synthesizing a sentence based on fixed data having a specific format composed of one or more data, and outputting the sentence as a voice A computer-readable recording medium storing a program for causing a computer to execute the step of storing, a step of storing a plurality of the fixed data, and a position for embedding each data in the fixed data The content of the fixed character part of the text is different From the step of storing a plurality of sentence examples and the stored sentence examples, Read Example sentences to be embedded for each fixed form data In the prescribed way A step of selecting, a step of embedding each data in the fixed form data in a predetermined position of the selected sentence example based on the fixed form data and the selected sentence example, and a step of combining the document A computer-readable recording medium having recorded thereon a program for causing a computer to analyze a document and convert it into an intermediate notation for expressing speech and to synthesize and output speech based on the intermediate notation It is characterized by being.
[0022]
This configuration realizes a data reading device that allows users to listen to voice information without getting bored because they can listen to information by voice output with different sentences for each standard data, instead of repeating the same sentence just by changing the data part as before. I can do it.
[0023]
DETAILED DESCRIPTION OF THE INVENTION
The principle and basic operation flow of the present invention will be described with reference to the basic configuration diagram of the present invention in FIG. 1 and the flowchart showing the basic processing flow in FIG.
[0024]
First, the sentence example selection unit 1 inputs fixed form data from the fixed form data storage unit 1-1 (step S1), and when the fixed form data is not finished (step S2: NO), the predetermined selection method determined in advance is used. A sentence example is selected (step S3). If the standard data is finished (step S2: YES), the process is finished.
[0025]
Standard data and sentence examples are usually created in advance by the data provider, so in the basic configuration diagram of FIG. 1, it is assumed that the input unit is not provided and is already registered. You may make it carry out by providing a part. Standard data and sentence examples can be entered using a normal editor. The steps up to this step are the feature of the present invention, and it is possible to read out different sentences using different sentence examples for each fixed data, without repeating the same data with the fixed data changing only the data part. As a result, the listener can listen to the reading information without getting bored.
[0026]
The subsequent steps are the same as those of the conventional text data speech-to-speech device, and the general example is taken. The present invention is not limited to this pattern. The mechanism may be used. The text synthesizing unit 2 receives the standard data and the text example, and synthesizes the text by incorporating the standard data into the text example (step S4). When the sentence analysis unit 3 receives the synthesized sentence, the sentence analysis unit 3 performs morpheme analysis / syntax analysis of the sentence using the morpheme dictionary 3-1, and converts the sentence into an intermediate notation representing speech (step S5). The speech synthesizer 4 receives the intermediate notation, synthesizes speech using the speech unit dictionary 4-1, and reads out the text as speech (step S6).
[0027]
Here is an example of what the intermediate notation is.
[0028]
For example, if the synthesized text is "The weather on July 13 in eastern Kanagawa will be sunny and cloudy. The probability of rain from 0:00 to 6:00 will be 20%.""Kanagawa Ken Ken Tobuno, Shigiga Juice Nichino Tenkiwa, Hare Nochi Kumori Desho. And so on. Katakana expresses readings, "'" and "_" indicate accent information, and "," indicates that a pose is taken. The intermediate notation given here is merely an example, and a notation different from the notation used here may be used.
[0029]
Embodiment 1
An example will be described in which a plurality of sentence examples having different sentence contents are prepared, a sentence example is selected for each fixed form data, and is read out by voice. Since the present embodiment is the same as the basic configuration diagram of FIG. 1, the detailed description of each function part will be omitted, and the detailed function of the sentence example selection unit 1 will be described.
[0030]
In the sentence example storage unit 1-2, a plurality of sentence examples are stored in a form in which the contents of the fixed character portion of the sentence are different in a state where all the data of the fixed form data can be embedded. There can be a plurality of methods for specifying where each data of the fixed form data is embedded in the sentence example. Some of the examples are given as weather data in which the fixed form data includes the district name, the date and the weather.
[0031]
First, in the first pattern of the data specifying method in the sentence example, the contents of the sentence example are set to “<< 1 >><< 2 >> weather will be << 3 >>." The number between the symbols << and >> indicates what number of standard data is to be inserted. That is, << 1 >> represents the district name as the first data, << 2 >> represents the date as the second data, and << 3 >> represents the weather as the third data. As a result, it is possible to know what position of the fixed form data should be put in which position in the sentence example, and not only a sentence example in which the fixed character part of the sentence example is changed while keeping the data order, but also a sentence example in which the data order is changed can be created. Because it becomes possible, it becomes possible to add many different variations to the document to be read, and it is possible to make the listener less tired.
[0032]
Next, as the second pattern of the data designation method in the sentence example, in the first pattern, it is difficult to know which data the numerical data indicates, and when it is corrected later, the data Add a comment that indicates what data the number represents rather than just a number, and say “<< 1: District Name >>'s<< 2: Monday & Sunday >> weather will be << 3: Weather >>." You can make it cheaper to understand what the variable is. Thereby, in addition to the advantage of a 1st pattern, the maintainability at the time of adding and correcting a sentence example can be made easy.
[0033]
Next, as a third pattern of the data specifying method in the sentence example, data in which data names to be applied in the order of the data of the fixed form data are prepared, and “<< district name >>, << month day >>, << weather >> In other words, it is possible to indicate what kind of variable each data of the standard data is assigned to. The symbols «and» indicating the variable may be omitted if the variable delimiter is known, and may be “district name, date, weather”. Data specifying the variable name to be assigned to each data of the fixed form data may be stored as the first fixed form data in the fixed form data storage unit 1-1 or the first case data in the sentence example storage part 1-2. It may be entered as sentence example data, or may be input as a parameter of the program of the sentence example selection unit 1. As a result, it is possible to directly specify in the sentence example by data name which data should be embedded. That is, when data specifying a variable name is entered as the first sentence example data in the sentence example storage unit 1-2, the sentence example selecting unit 1 reads the data specifying the variable name in the first case of the sentence example, and forms data The first data of can be recognized as the district name, the second data is the date of the month, and the third data is the weather, so in the example, without the number indicating the number of the data, Use the variable name specified in the first case of the sentence, such as “Weather will be the weather of“ Monday ”in“ Region Name ”. As a result, in the first case of the sentence example, if you specify only the variable name to be given to each data of the fixed form data, when making the sentence example, it is not necessary to be aware of what number of the fixed form data, the sentence example is more, You can add and modify in a natural way.
[0034]
In this example, the variable name is input as the first data of the sentence example, but this is to enable use as a general-purpose data reading device that changes the reading data. For example, the variable name used in the fixed data may be defined and held in the program.
[0035]
Using any of these sentence pattern patterns, not only the various sentence examples in which the parts other than the variables are changed, but also the order of the variables itself is arbitrary, so that it is possible to add various variations to the document to be read. Can make it harder to get bored.
[0036]
In addition, as the fourth pattern of the data specification method in the sentence example, the data of the standard data will be embedded in the sentence example in order from the front, and the weather of <<<<<< will be <<>>. The number or variable name indicating the data may be omitted and only the symbol indicating the position where the data is inserted may be used. However, this pattern cannot handle sentence examples in which the order of data is changed.
[0037]
In the case of the first and second patterns described above, the sentence example selection unit 1 prepares a data table (not shown) numbered in order from 1, reads the standard data, and stores each data in order. . In the case of the third pattern, the sentence example selection unit 1 adds the data name column of the two-dimensional data table (not shown) composed of the data name and data to the fixed data storage unit 1-1 or the example sentence storage unit 1-2. When the data that lists the first data name is read, or the parameters that represent the data are read, the data names are stored in order, and the standard data is read, each data is sequentially assigned to the corresponding data name in the data table. Store in part. In the case of the fourth pattern, it is not necessary to create a data table. The fixed form data passed from the sentence selection unit 1 to the sentence composition unit 2 is passed in the form of a data table in the case of the first to third patterns, and is passed in the form of the fixed form data in the case of the fourth pattern. It is.
[0038]
Next, the sentence example storage unit 1 selects a sentence example from among a plurality of sentence examples for each fixed form data, and there are several possible methods, and examples of the method are given below. In the present embodiment, it is assumed that the sentence example storage unit 1-2 stores sentence examples that are gradually simplified in order from the most polite sentence example.
[0039]
The first method is a method of using the first sentence example to the last sentence example repeatedly and repeatedly. This method reads sentence examples and standard data in order from the beginning, and uses both data as they are. If the number of fixed form data is larger than that of the sentence example, after reading the last sentence example, it returns to the beginning of the sentence example and reads it. In this way, the sample sentences are read while being cycled until there is no fixed data. This makes it easy to avoid using the same sentence example continuously.
[0040]
The second method is a method of obtaining the number of sentence examples to be used by generating a positive integer pseudo random number equal to or less than the number of sentence examples. At this time, the random number of past times is stored, and when the same random number as the stored random number is generated, the same sentence example is continued by recalculating the random number until an unstored random number appears. It may not be used or repeated at short intervals. When the third pattern of the data designation method in the above sentence example is used and data in which the data names are listed is stored in the sentence example storage unit 1-2, the first data in the sentence example is a list of data names. Since it is not a sentence example, the sentence example to be used by generating a positive integer pseudo-random number of the number -1 in the sentence example uses the pseudo-random number plus the first sentence example. In the first method, the same sentence example is repeated at regular intervals, so if you listen to it many times, you can predict the next sentence example, or the most polite sentence example comes after the most simplified sentence example, This method makes it impossible to predict which sentence will come next.
[0041]
The third method is a method in which a predetermined first few sentence examples are used only once, and subsequent repeated sentence examples are read out using the first method or the second method. The first few sentence examples use sentence examples using polite expressions, and thereafter, sentence examples with simplified expressions are stored. In order to indicate the number of polite sentence examples at the beginning and the simplified example sentences to be repeated, information on how many polite sentence examples are set in the program of the sentence example selection unit 1 in advance. However, in order to make it universal, it is preferable to provide a flag indicating whether or not the sentence example is repeated at the head of each sentence example. An example of a sentence example is shown in FIG. In this example, the first and second sentence examples are used only once, and the third and fourth sentence examples are repeated. For the listener, it is preferable that the sentence to be heard is initially polite and simplified gradually. However, the first method may return from a simplified expression to a polite expression when repeated, and the second method suddenly starts with a simplified expression or a polite expression from a simplified expression. However, such a problem can be avoided by this method.
[0042]
The sentence selection unit 1 in the case where there is a sentence example as shown in FIG. 4A, using the first pattern as the data designating method in the sentence example, using the third pattern using the pseudo-random number as the sentence selecting method. The process flow (corresponding to steps S1 to S3 in FIG. 2) will be described with reference to the flowchart in FIG.
[0043]
The sentence example selection unit 1 reads a sentence example (step S11), determines whether a flag indicating whether the sentence example is to be repeatedly used is set for the sentence example (step S12), and if the repetition flag is not set (step S12). S12: NO) reads the fixed form data (step S13), and if the fixed form data is not finished (step S14: NO), the sentence example and the fixed form data are transferred to the sentence composition unit (step S15). Since the sentence example in which the repetition flag is not set is used only once, the process returns to step S11 to read the next sentence example. If the standard data is finished (step S14: YES), the program is finished.
[0044]
When the repetition flag of the sentence example is set (step S12: YES), the sentence example is stored in the table, and the number of sentence examples is counted (step S16). Then, the next sentence example is read (step S17). In order to read all the sentence examples in which the repetition flag is set, when the sentence examples are not completed (step S18: NO), the process returns to step S16 and repeats until the sentence examples are completed.
[0045]
When the example sentence to be used repeatedly is stored in the table, the fixed form data is read (step S19). When the fixed form data is not finished (step S20: NO), a positive integer pseudo-random number equal to or less than the count of the example sentence stored in the table is stored. It is generated and a sentence example to be used is selected (step S21). The read fixed form data and the selected sentence example are passed to the text synthesis unit 2 (step S22), and the process returns to step S19 to read the next fixed form data. When the standard data is finished (step S20: YES), the program is finished.
[0046]
Embodiment 2
Next, an embodiment in which a sentence example is stored in the form of intermediate notation and only the fixed form data is converted into the intermediate notation to eliminate the need for sentence analysis will be described with reference to FIG. The description of the same parts as in FIG. 1 will be omitted as appropriate.
[0047]
In this embodiment, the sentence example selection unit 1 reads the fixed form data, performs duplication check, etc., and selects the sentence example, and then the sentence example is in an intermediate notation. The data is transferred to the fixed data conversion unit 5. The fixed data conversion unit 5 converts each data of the received fixed data into an intermediate notation by using the fixed data morpheme dictionary 5-1, and passes it to the text synthesis unit 2. The sentence synthesizing unit 2 synthesizes the intermediate notation sentence from the received intermediate notation sentence example and the fixed form data, passes it to the voice synthesizing unit 4 and reads it out as voice information.
[0048]
The morpheme dictionary for fixed data is a subset of the morpheme dictionary 3-1, and it is only necessary to store intermediate notation of only the morphemes included in the fixed data. However, if there is no problem in the capacity of the hard disk, there is no problem even if the morpheme dictionary 3-1 is used.
[0049]
Regarding the data names in the sentence examples, the symbols «and» indicating that they are variable names may be left in the intermediate notation, but in order to make the variable names easier to understand, it is preferable not to use the intermediate notation. .
[0050]
According to this embodiment, since it is necessary to store the sentence example in the intermediate notation, it is more troublesome to input the sentence example directly in the intermediate notation than to input a normal sentence. In that case, a normal sentence may be input first, the process corresponding to the sentence analysis unit 3 described in Embodiment 1 may be executed, and a sentence example of intermediate notation may be created in advance. In addition, it is conceivable that the fixed form data is set to an intermediate notation from the beginning. In that case, the fixed data converter 5 shown in FIG. 5 is not necessary.
[0051]
The flow of processing in the present embodiment will be described using the flowchart of FIG.
[0052]
When the fixed form data is input from the fixed form data storage unit 1-1 (step S31) and the fixed form data is not completed (step S32: NO), the sentence example is selected from the sentence example storage unit 1-2 by a predetermined method (step S33). To do. If the standard data is complete (step S32: YES), the process is terminated. The fixed form data is transferred to the fixed form data conversion unit 5 and converted into intermediate notation (step S34). Both the sentence example and the fixed form data are passed to the sentence synthesizing unit 3 in an intermediate notation format, and the sentence is synthesized (step S35). Based on the synthesized text, the speech synthesizer synthesizes speech and reads the text (step S36).
[0053]
As described above, according to the present embodiment, since the sentence example has an intermediate notation from the beginning, it is only necessary to convert the data to the intermediate notation when reading the fixed form data. This eliminates the need for a program for sentence analysis, and only requires data for standard data to be prepared in the morpheme dictionary. This makes it easy to change to intermediate notation, requiring less memory and storage capacity, and processing Speed can be improved.
[0054]
Embodiment 3
Next, an embodiment that reads out overlapping portions of the standard data collectively will be described with reference to FIG. 1 because the configuration diagram is the same as that of the first embodiment. Explanation of overlapping parts will be omitted as appropriate.
[0055]
In the present embodiment, it is premised that there is duplicate data that can be expressed together in fixed data. For example, taking the weekly weather forecast as an example, the typical data consists of the following data: district name, date, weather, precipitation probability 1, precipitation probability 2, precipitation probability 3, precipitation probability 4, predicted minimum temperature, predicted maximum temperature The precipitation probabilities 1 to 4 are the precipitation probabilities from 0 to 6 o'clock, precipitation probabilities from 6 o'clock to 12 o'clock, precipitation probabilities from 12 o'clock to 18 o'clock, and precipitation probabilities from 18 o'clock to 24 o'clock, respectively. It shall represent. In the case of this data, regarding the precipitation probabilities 1 to 4, if the same precipitation probabilities continue, the portions can be expressed together. For example, “The weather of << 1 >><< 2 >> is << 3 >>, the probability of precipitation from 0 o'clock to 6 o'clock is << 4 >>, from 6 o'clock to 12 o'clock is << 5 >>, from 12 o'clock to 18 o'clock is << 6 ”,« 7 »from 18:00 to 24:00, the expected minimum temperature is« 8 », and the predicted maximum temperature is« 9 »." The probability of precipitation from «4» to «7» Furthermore, since the following seven sentence examples can be considered according to the data duplication method, a total of eight sentence examples can be created.
▲ 1 ▼ The weather of << 1 >><< 2 >> is << 3 >>, the probability of precipitation from 0:00 to 12:00 is << 4 >>, from 12:00 to 18:00 << 6 >>, from 18:00 to 24:00 << 7 >>
The expected minimum temperature is «8» and the expected maximum temperature is «9».
▲ 2 ▼ The weather of << 1 >><< 2 >> is << 3 >>, the probability of precipitation from 0am to 6am is << 4 >>, from 6am to 6pm is << 5 >>, from 18:00 to 24:00 is << 7 >>
The expected minimum temperature is «8» and the expected maximum temperature is «9».
▲ 3 ▼ The weather of << 1 >><< 2 >> is << 3 >>, the probability of precipitation from 0:00 to 6:00 is «4», from 6:00 to 12:00 «5», from 12:00 to 24:00 « 6 >>
The expected minimum temperature is «8» and the expected maximum temperature is «9».
▲ 4 ▼ The weather of << 1 >><< 2 >> is << 3 >>, the precipitation probability from 0:00 to 12:00 is << 4 >>, from 12:00 to 24:00 is << 6 >>,
The expected minimum temperature is «8» and the expected maximum temperature is «9».
▲ 5 ▼ The weather of << 1 >><< 2 >> is << 3 >>, the precipitation probability from 0:00 to 18:00 is << 4 >>, from 18:00 to 24:00 is << 7 >>,
The expected minimum temperature is «8» and the expected maximum temperature is «9».
▲ 6 ▼ The weather of << 1 >><< 2 >> is << 3 >>, the probability of precipitation from 0 o'clock to 6 o'clock is << 4 >>, from 6 o'clock to 24 o'clock is << 5 >>,
The expected minimum temperature is «8» and the expected maximum temperature is «9».
▲ 7 ▼ The weather of << 1 >><< 2 >> is << 3 >>, the probability of precipitation from 0:00 to 24:00 is << 4 >>, the predicted minimum temperature is << 8 >>, and the predicted maximum temperature is << 9 >>.
[0056]
Although it is possible to omit the data as described above within one standard data, it is not necessary to limit the duplication check of data within one standard data, and a part common to a plurality of standard data is omitted. Is also possible. For example, in the case of the above-mentioned standard data for the weekly weather forecast, there is standard data from Sunday to Saturday. In this case, the district name << 1 >> is the same in any standard data, and it may be read out once at the beginning, and may be omitted at the second and subsequent readings. In other words, each sentence example in which the district name << 1 >> is omitted is added to the pattern of the above eight sentence examples, so that a total of 16 sentence examples can be created. In this case, since duplication of data is checked over a plurality of fixed data, the read fixed data must be stored in a memory or the like.
[0057]
Next, a method for checking duplicate data will be described. The problem with the checking method is how to determine which data and which data should be checked for duplication.
[0058]
As a first method for discriminating data to be checked for duplication, there is a method for checking duplication by simply determining whether or not each data of the fixed data is continuously the same. With this method, there is no need to instruct which data and which data should be checked, and it is only necessary to check whether or not the same data is continuous for all the data. In this case, if the same data continues, it must be premised that one of the duplicate data may be left unconditionally and omitted. That is, it is premised that there is no data that may be duplicated in the same format even though they are different types of data.
[0059]
Taking the example sentence described above as an example, an example of pattern data in question is given. The probability of precipitation from 18:00 to 24:00 is assumed to be 10% and the predicted minimum temperature is assumed to be 10 degrees. In the above sentence example,% and degree are not fixed characters in the sentence example and are included in the fixed form data, so it will not be duplicated. However, if% and degree are fixed characters, both are as data 10 and is regarded as duplicate data. As a means for avoiding this, it is possible to determine that omission is not possible by checking that there is no example sentence in which << 8 >> representing the lowest expected temperature is omitted in the example data. However, in this case, it can be avoided by this method, but it is not always possible to avoid in any case, so the data format of the different types of data is the same, so there is a possibility of misjudging that it is duplicated. It is necessary to create fixed data and sentence examples so that they are not arranged or arranged as examples.
[0060]
As another method for determining data to be checked for duplication, there is a method for designating data to be checked for duplication. As a specific designation method, there is a method of providing a duplication check flag for each data in the fixed form data, a separate file for duplication check instruction, or data for duplication check as a program parameter of the sentence example storage unit 1 There is a way to specify. As described above, the check includes the duplication check of data in one fixed data and the duplication check of data over a plurality of fixed data, and there are a plurality of data groups to be checked in the check in one fixed data. there's a possibility that. Therefore, it is necessary to set a flag in the data so that they can be distinguished.
[0061]
As an example of setting a flag for each data of regular data, in the above sentence example, the percentage of precipitation probability and the degree of expected temperature are fixed characters in the sentence example. Here is an example in which the distinction is not made. FIG. 7A shows an example in which a flag is set for each piece of regular data. The flag “*” in front of the district name indicates that if the data is duplicated across multiple regular data, it will be omitted for the second and subsequent times. The data of the precipitation probability 1 to 4 and the predicted minimum temperature and the predicted maximum temperature are numbers, and it cannot be determined unless the distinction of the data is instructed. The flags of the precipitation probabilities 1 to 4 are all set to 1, and indicate the same group of data that can be omitted in one regular data. When there are a plurality of groups that can be omitted in one regular data, different numbers or characters may be set for each group. The portion where the flag is blank is data that cannot be omitted.
[0062]
When not specifying the duplicate flag for each data in the fixed form data and specifying the data to be checked for duplication by storing it in a separate file or by specifying with a parameter, "* ,, 1, 1, 1" The duplication check instruction such as “, 1,...” Is input to specify what items should be checked for duplication.
[0063]
In this way, if data to be checked for duplication is specified, it becomes clear which duplication check should be performed for which data and which data, and duplication check can be executed without making a mistake.
[0064]
Once the duplication check has been completed and it is determined which data to omit, it is necessary to select a corresponding sentence example. The sentence example to be selected is determined by which data is omitted, but the target sentence example is not necessarily one. In that case, the sentence example to be used is determined from the sentence example to be used by the selection method used in the first embodiment. In the following, a method for identifying a sentence example after the data to be omitted is determined will be described.
[0065]
First, the first method for identifying a target sentence example is a method for reading all sentence examples every time the standard data is read, and for identifying all sentence examples not including a data name to be omitted in the sentence examples. This is the simplest method, but the most inefficient method, because it checks all the sentence examples every time the fixed form data is read and identifies the target sentence examples. However, if all the sentence examples are checked only once and a table is created that shows what data is omitted from the sentence example and the sentence example, the second and subsequent times will refer to this table. There is no need to repeatedly determine which data is omitted, and the sentence example itself is already stored in the table, so there is no need to read it from the sentence example storage unit 1-2 again, so that it can be processed efficiently. become.
[0066]
Next, the second method of identifying the target sentence example is to store an example sentence with an index indicating what data is omitted from the sentence example. This is a method of creating an index by omitting it and searching for an example sentence of interest. Here are some examples:
[0067]
For example, when the first data and the sixth data are omitted, the index calculation method is determined in advance so that the index is 16. Since there may be a plurality of target sentence examples, the sentence example is an indexed file having a duplicate key. Since there is a sentence example with a duplicate key, first, the value of 16 is positioned as the first sentence example of the target sentence example of the file containing the sentence example, and then reading is sequentially performed until the index is no longer the value of 16. Then, all the target sentence examples are identified. The index may not be an omitted data number, but may be a sentence number that is not omitted, an omitted data name, or the like, but the omitted data number is more preferable as an index because the index becomes shorter.
[0068]
When checking for duplication only within one standard data, it is only necessary to read the standard data one by one and process it, so the flowchart is the same as in FIG. 2, but checks for duplication of data across multiple standard data. In this case, if all the standard data is read at once so that it can be easily checked, the flow of processing changes, so the flowchart of FIG. 2 changes as follows. First, in order to read all the regular data, it is necessary to read all the standard data in step S1, store it in the memory, and move to step S3 when the standard data has been read. Therefore, while the reading of the standard data is not completed in step S2, the process returns from step S2 to step S1, and when the reading of the standard data is completed in step S2, the process proceeds to step S3. From step S4 to step S6, all the read standard data is collectively processed, and when the process of step S6 is completed, the entire process is completed.
[0069]
As described above, according to the present embodiment, not only a sentence example in which a fixed character is simplified can be created, but also a sentence example in which the duplicated data itself is omitted can be created. Can be realized.
[0070]
Embodiment 4
Next, FIG. 1 is a configuration diagram of the first embodiment regarding an embodiment in which information that is desired to be quickly obtained, important information, etc. are first read out collectively instead of reading out sentences in fixed data units. Will be described. Description of the same parts as those in Embodiment 1 is omitted as appropriate.
[0071]
In the present embodiment, each data of the fixed data stored in the fixed data storage unit 1-1 is provided with a flag indicating which data is regarded as a data set, as shown in FIG. 7B. Yes. In this example, the fixed form data is divided into two data sets. The first data set is a set of data in which “*” or “#” is specified as a flag, and the other data set is a set of data for which no flag is set. Data for which the flag “*” is set indicates data of a data set to be read out only at the first time. The data for which the flag “#” is set is one of the data of the data set to be read out first, but indicates that it belongs not only to the first data set but also to the other set data. . That is, it represents common data that is read out both when the first data set is read out and when another data set is read out.
[0072]
Assume that the sentence example as shown in FIG. 4B is stored in the sentence example storage unit 1-2, and when reading the first three cases of the standard data in FIG. 7B,
“The weather on February 1 in eastern Kanagawa Prefecture is cloudy after clear. It is sunny on February 2. It is sunny and rainy on February 3.
Precipitation probability on February 1, 0% from 0:00 to 6:00, 10% from 6:00 to 12:00, 30% from 12:00 to 18:00, 30% from 18:00 to 24:00,
The expected minimum temperature is 0 degrees and the expected maximum temperature is 10 degrees.
Precipitation probability on February 2, 0% from 0:00 to 6:00, 0% from 6:00 to 12:00, 0% from 12:00 to 18:00, 0% from 18:00 to 24:00,
The expected minimum temperature is 0 degrees and the expected maximum temperature is 10 degrees.
Precipitation probability on February 3, 10% from 0:00 to 6:00, 10% from 6:00 to 12:00, 60% from 12:00 to 18:00, 60% from 18:00 to 24:00,
The expected minimum temperature is 0 degrees and the expected maximum temperature is 10 degrees. "
[0073]
In this embodiment, which data of the standard data is collected is provided by providing a flag for each data, but this can be stored in a separate file or specified by a program parameter of the sentence selection unit 1 Also good. To stop specifying the standard data flag in (b) of Fig. 7 and specify only the flag in a separate file or parameter, for example, specify as "*, #, * ,,,,," To do. This may be shortened to “*, #, *”.
[0074]
When the functions of the third embodiment are combined with the present embodiment, a flag indicating a set of data may overlap with a flag indicating data to be checked for duplication, so it is necessary to have a separate flag for each data. Therefore, the standard data is as shown in FIG. When a sentence example that collectively expresses the precipitation probability portion of the third sentence example in FIG. 4B is added and the first three cases in FIG. 7C are read out, the result is as follows.
“The weather on February 1 in eastern Kanagawa Prefecture is cloudy after clear. It is sunny on February 2. It is sunny and rainy on February 3.
Precipitation probability on February 1, 0% from 0:00 to 6:00, 10% from 6:00 to 12:00, 30% from 12:00 to 24:00, the predicted minimum temperature is 0 degrees, and the predicted maximum temperature is 10 It will be degrees.
The probability of precipitation on February 2 is 0% from midnight to 24:00, the expected minimum temperature is 0 degrees, and the predicted maximum temperature is 10 degrees.
The probability of precipitation on February 3 is 10% from 0:00 to 12:00, 60% from 12:00 to 24:00, the expected minimum temperature is 0 degrees, and the predicted maximum temperature is 10 degrees. "
In this way, the sentence can be further shortened.
[0075]
According to this embodiment, it is possible to first output high-importance data in a plurality of fixed data in a voice and then output the subsequent data as detailed data, so that only necessary information can be output. You can listen first, and later information if you don't need it.
[0076]
As another method, there is a method of having a flag in a sentence example instead of having a flag in fixed form data. When this method is used to combine the sentence example designation method of the first embodiment as shown in FIG. 4A, for example, the sentence example shown in FIG. First, a flag of 1 is set in the sentence example used to read out the first data set, and a flag of 2 is set in the sentence example used to read out the second data set. Specify 2 in the second byte of the second flag so that there are 2 sentence examples with a flag of 1 and the first sentence example is used only once and the second sentence example is used repeatedly. . And it reads out about all the regular data using the sentence example for each flag. If this method is used, it is possible to group data by checking data used in a sentence example without setting a flag for each data and grouping.
[0077]
In the method of flagging data or specifying a set of data with separate files or parameters, the data to be used across multiple data sets must be specified as long as it is not essential data for all data sets. Although it is difficult to set a flag, in the case of a method of grouping data in a sentence example, it is specified because the data set is specified by the data used in the sentence example without specifying the data set. Is very easy.
[0078]
[Embodiment 5]
Next, based on the standard data and sentence examples, not only reading the data, but also creating a table and displaying it on the display device, so that information can be obtained not only by hearing but also by looking The embodiment will be described with reference to the configuration diagram of FIG. 8, taking a car navigation used as an in-vehicle information service terminal as an example.
[0079]
In the present embodiment, it is assumed that the standard data is the weekly weather forecast data, and the standard data and the sentence example are transmitted in the form of a character broadcast transmitted by FM broadcasting from the broadcasting station. In addition, a case where road traffic information and weather forecast data for each region are received from a special service station as part of a so-called ITS system is included.
[0080]
The broadcast station broadcasts the fixed form data and the sentence example by FM radio waves by using the data transmission part 6 as input to the fixed form data storage part 6-1 and the sentence example storage part 6-2 in the broadcast station. In car navigation, the standard data and the example sentence are received by the data receiving unit 7 and stored in the standard data storage unit 7-1 and the example sentence storage unit 7-2 on the car navigation side. The display data creation unit 8 obtains standard data and sentence examples from the standard data storage part 12-1 and the sentence example storage part 12-2, creates table data, and displays the table data on the display device 15 such as a display. The configuration of the voice data creation unit 9 is the same as the configuration diagram of the embodiment described above, in which the standard data storage unit 1-1 is in the standard data storage unit 7-1 and the sentence example storage unit 1-2 is in the sentence example storage unit 7-2. Since the replacement is the same as that except that the sound output unit 11 such as a speaker is used as a sound output part, the description is omitted.
[0081]
An example of how the display data creation unit 8 creates the table data will be described below.
[0082]
When generating table data, the first sentence example that expresses the most politely is used as input data, and variables are used not by numbers but by variable names. The typical data is (c) in Fig. 7, and the example sentence is "The weather of << Month day >> of << District name >> will be << Weather >>. Precipitation probability from 0:00 to 6:00 is << Precipitation probability 1 >>% Precipitation probability from 6 o'clock to 12 o'clock will be << precipitation probability 2 >>% Precipitation probability from 12 o'clock to 18 o'clock will be << precipitation probability 3 >>% Precipitation probability from 18 o'clock to 24 o'clock Will be «Precipitation probability 4». Expected minimum temperature will be «Expected minimum temperature» degrees. Expected maximum temperature will be «Expected maximum temperature» degrees. "
[0083]
First, a two-dimensional table having the same format as the table is provided on the memory. Since it is necessary to put an item name for indicating what data is at the top of the table, one line is left in order to put the item name at the top of the table of the two-dimensional table. Next, a variable name indicating what data each column of the table is is acquired as an item name from the sentence example. In the case of the above sentence example, the item name is the district name, month, day, weather, precipitation probability 1, precipitation probability 2, precipitation probability 3, precipitation probability 4, predicted minimum temperature, and predicted maximum temperature. All that remains is to read the standard data and store it in the table in order. When the variable name is simply used as an item name representing each data, the weekly weather forecast table as shown in FIG.
[0084]
Here, in order to make the table easier to see, functions are added next. Since the district name in FIG. 7A is data for performing a duplicate check across a plurality of fixed data, the district names after the first are omitted. As for the month and day, the month and day are disassembled.
[0085]
Next, it is understood that the precipitation probabilities 1 to 4 in the standard data are data for performing a duplication check in the standard data as shown in FIG. In other words, the duplication check means that the same type of data means that when the fixed character immediately before the variable in the sentence example is checked, the fixed character immediately before the precipitation probability 1 to 4 is “0 o'clock”. The probability of precipitation from 18:00 to 16:00 "," The probability of precipitation from 6:00 to 12:00 "," The probability of precipitation from 12:00 to 18:00 ", and" The probability of precipitation from 18:00 to 24:00 " . If you remove the particle “ha” from each fixed character, you can see that it consists of the word “probability of precipitation” and a word that modifies it. Therefore, it is determined that the “precipitation probability” is a word that summarizes the precipitation probabilities 1 to 4 as data. For the item name “Precipitation Probability”, the word “Precipitation Probability”, which is a common part, may be extracted from the precipitation probabilities 1 to 4 which are variable names in the sentence examples.
[0086]
Since it is the part that modifies the variable name in the sentence example that distinguishes the precipitation probabilities 1-4, the precipitation probabilities 1-4 are “from 0:00 to 6:00” and “6”, respectively, excluding the particle. "From 12:00 to 12:00", "12:00 to 18:00", and "18:00 to 24:00", which are regarded as words that distinguish the precipitation probabilities 1 to 4. If it is used as an item name in the table as it is, it is long, so “to” and “to” are converted to “to”, and in order to match the length of each item, a two-digit number is converted to half-width. The precipitation probabilities 1 to 4 are data to be checked for duplication within one standard data, but are numbers and are not preferable to omit when making a table. In this case, the area for entering the item name is the top two rows of the table.
[0087]
For each data, the regular data is not displayed as it is, but the word immediately after the variable name in the sentence example is checked, and the word immediately after the variable name is a word indicating a unit such as “%” or “degree”. In some cases, the table is included in the table for easy understanding. Regarding the number of digits of each data, the maximum data length of the corresponding data and the length of the item name are checked, and the longer one is selected. The item name is centered so that it is in the middle of the number of digits. For numbers, right justify. In this way, when the item name and fixed form data are edited or the table is formed by positioning, a weekly weather forecast table as shown in FIG. 9B is obtained.
[0088]
Furthermore, the predicted minimum temperature and the predicted maximum temperature are decomposed into words, and the variable name is checked to see if the word at the end of the variable name and the word at the beginning of the variable are the same. If the words are the same, they are regarded as common items and can be extracted as common item names. For example, since the first word “expected” and the last word “temperature” are the same, the predicted temperature is taken as a common word, and the item names of each data are “lowest” and “highest”, respectively.
[0089]
As described above, when creating and displaying table data, the table data is displayed with ruled lines. As the ruled line, a format-format character that is a character representing the ruled line may be used, or a graphic ruled line may be used. When using format format characters, secure areas for the format format characters at the top, bottom, left, and right of each data. Can be displayed.
[0090]
According to the present embodiment, when the standard data can be output not only to voice but also to the display device, necessary information can be heard by voice and can be confirmed visually on the screen. If it can be confirmed, necessary information can be acquired sooner. For example, in the case of car navigation, it is dangerous to watch the screen while driving in a car, so it is dangerous to listen to it by voice, and when you are stopped by a signal, etc. Become.
[0091]
Embodiment 6
Next, an embodiment in which a sentence including a table is input and the standard data represented in the table is read out will be described with reference to FIG.
[0092]
A device that reads out a data portion of a table in a sentence by voice is disclosed in "Document Reading Device (Japanese Patent Laid-Open No. 5-46607)". In the document reading apparatus, a ruled line is checked for a ruled line segment (corresponding to a format-format character), a table is detected, a data position on the table is calculated from the ruled line information, an item name and data are extracted, and if necessary Then, only necessary information can be read out by interacting with the user and specifying the table reading-out information. When the item name is the brand and the data is the company A, the reading method is simply to read “brand, company A”.
[0093]
The function of the text / table discriminating unit 12 for discriminating the table portion in the present embodiment is to discriminate the table by detecting the format-format characters. The document and table classification function of the invention of the document reading device Since it is the same, detailed description is omitted. The table structure analysis unit 13 analyzes from what byte to how many bytes of each table, and extracts item names and data. Up to this point, the function is the same as that of the document reading apparatus. In the document reading apparatus, for example, if there is a table as shown in FIG. 9 (a), “district name, eastern Kanagawa prefecture. Precipitation probability 2 and 10. Precipitation probability 3 and 30. Precipitation probability 4 and 30. Expected minimum temperature 0. Expected maximum temperature 10. District name, eastern Kanagawa Prefecture. Read out like "...". In the document reading apparatus, by specifying the table reading information, it is possible to give a percentage to the precipitation probability 1 to 4 or to measure the predicted minimum temperature / expected maximum temperature, or to read out only necessary information. However, you cannot change the way you read out by checking for duplication of data. In addition, when the item name column extends over a plurality of lines as shown in FIG. 9B, or when there is a large item for collecting detailed items, the table is easy to see, such as the district name and date. Therefore, it is not possible to cope with the case where data is omitted.
[0094]
In the present embodiment, the table structure analysis unit 13 further duplicates the function of extracting the table item name as shown in FIG. 9B, the function of supplementing the part where data is omitted, and the way of writing the table. It has a function to identify how to check and items to be checked for duplication, and to automatically create fixed data and sentence examples.
[0095]
First, regarding the extraction of the item name, when there is a table as shown in FIG. 9B, the column with the largest number of rows is checked in the top column of the table. Here, since the district name / month / day / weather spans three lines, the top three lines in the table are regarded as item names. For the district name, date, weather forecast minimum temperature, and forecast maximum temperature, the data for the three lines in the column is extracted as a single continuous data, and the item name is “ ”,“ Monday ”,“ Weather ”,“ Weather Expected Minimum Temperature ”, and“ Expected Maximum Temperature ”. There is also a space between the item names “weather”, but this is a problem when converted to the intermediate notation, so it is narrowed down.
[0096]
Regarding the “Precipitation Probability” column, “Precipitation Probability” is one column and there are four columns below it. The four items are similar data, and it is determined that duplication check is possible. As a result, it is possible to create a sentence example that summarizes the duplicate check and to generate fixed form data with a duplicate check flag. Regarding the four item names representing the probability of precipitation, when creating a sentence example, “˜” is converted from “to” to generate a fixed character of the sentence example.
[0097]
Next, looking at the standard data, regarding the data of the district name, if the data is checked, the second and subsequent items are omitted, so it can be considered that a duplicate check across the district data between the standard data is possible. . As for the date data, February of the second and subsequent data is omitted, but items such as date are generally used, so they are omitted in the table structure analysis unit 13. It is desirable to incorporate a process for restoring “February” in general.
[0098]
Also, units such as the probability of precipitation and the% and degree of predicted temperature should be taken out as a sentence example to reduce the amount of standard data. When the data is restored and generated in this way and the duplication flag is set, the table in FIG. 9B can be converted into the data in FIG.
[0099]
In this way, based on the extracted item name and the unit that was taken out, supplement the minimum particle, etc. to create a sentence example, “District name is« District name ”. ≫.The weather is «weather» .The probability of precipitation, from 0 to 6 o'clock, «0 o'clock to 6 o'clock»%, from 6 o'clock to 12 o'clock, «6 o'clock to 12 o'clock»%, from 12 o'clock to 18 o'clock , «12:00 to 18:00»%, 18:00 to 24:00, «18:00 to 24:00»%, the expected minimum temperature is the «expected minimum temperature» degree, the expected maximum temperature is the «expected maximum temperature» degree. ”As the most basic sentence examples, it is possible to create 16 kinds of sentence examples by summarizing the probability of precipitation or omitting the district name.
[0100]
The subsequent steps are the same as those in the first embodiment except that the sentence portion in the document is transferred from the document / table discriminating section 12 to the sentence analyzing section 3, and the description thereof will be omitted.
[0101]
According to the present embodiment, since the table portion existing in the document can also be generated from the table by generating sentence examples and fixed form data and reading the table data in various sentences, the listener is not bored. At the same time you can listen to the information on the table, you can also shorten the information.
[0102]
Next, a recording medium describing a program for realizing the data reading device according to the first to sixth embodiments of the present invention is a CD-ROM 15-1 or a floppy disk 15-, as in the example of the recording medium shown in FIG. Not only the portable recording medium 15 such as 2 but also any other storage device 14 stored at the end of the communication line, or a recording medium 17 such as a hard disk or RAM of a computer. Is read by the computer 16 when using.
[0103]
【The invention's effect】
According to the data reading device of the present invention, when reading the fixed form data represented by a table or the like, it does not read the same type of sentence repeatedly, but reads out different sentences for each fixed form data, and collects duplicate data. Therefore, the listener can listen to the necessary information in a short time without getting bored. Therefore, it is extremely effective to provide voice information to a user by applying it to an in-vehicle information service terminal device or the like.
[Brief description of the drawings]
FIG. 1 is a basic configuration diagram of the present invention.
FIG. 2 is a flowchart showing the basic processing flow of the present invention.
FIG. 3 is a configuration diagram of the data reading device according to the first embodiment of the present invention;
[Figure 4] Data content of sentence example
FIG. 5 is a configuration diagram of a data reading device according to the second embodiment of the present invention;
FIG. 6 is a flowchart showing a flow of processing according to the second embodiment of the present invention.
Fig. 7 Example of fixed data
FIG. 8 is a configuration diagram of a data reading device according to a fifth embodiment of the present invention;
[Figure 9] Weekly weather forecast table
FIG. 10 is a configuration diagram of a data reading device according to a fifth embodiment of the present invention;
FIG. 11 illustrates an example of a storage medium.
[Explanation of symbols]
1 sentence selection part
1-1 Standard data storage
1-2 Example sentence storage
2 Text composition part
3 sentence analysis department
3-1 Morphological Dictionary
4 Speech synthesis unit
4-1 Speech segment dictionary
5 Standard data converter
5-1 Morphological dictionary for fixed data
6 Data transmitter
6-1 Standard data storage on the broadcasting station side
6-2 Example sentence storage on the broadcasting station side
7 Data receiver
7-1 Standard data storage on the car navigation side
7-2 Example sentence storage on the car navigation side
8 Display data creation part
9 Voice data creation part
10 Display devices such as displays
11 Audio output devices such as speakers
12 Sentence / table discriminator
12-1 Text storage
13 Table Structure Analysis Department
14 Other storage devices stored at the end of the communication line
15 Portable storage media such as CD-ROM and floppy disk
15-1 CD-ROM
15-2 Floppy disk
16 computers
17 Storage media such as RAM / hard disk on computer

Claims (7)

In a data-to-speech device that synthesizes text based on fixed data with a specific format consisting of one or more data and outputs it as a voice,
Viewed including the position for embedding the data in the fixed data, a phrase storage unit that stores a plurality of different text example of the contents of the fixed character portion of the sentence,
The read routine data, a phrase embedded before each Symbol read type data selected by changing in a predetermined manner, embedding each data in the standard data in a predetermined position, to have a sentence synthesis unit for synthesizing sentences A featured data reading device. In a data-to-speech device that synthesizes text based on fixed data with a specific format consisting of one or more data and outputs it as a voice,
A fixed data storage unit for storing a plurality of the fixed data;
Viewed including the position for embedding the data in the fixed data, a phrase storage unit that stores a plurality of different text example of the contents of the fixed character portion of the sentence,
A phrase selecting unit for selecting from said phrase storage unit, by changing the pre-SL read text example be embedded in every type data in a predetermined manner,
Based on the fixed form data and the selected sentence example, each data in the fixed form data is embedded in a predetermined position of the selected sentence example, and a sentence composition unit that synthesizes a document;
A sentence analysis unit that analyzes the synthesized document and converts it into an intermediate notation for expressing speech;
A data reading device comprising a speech synthesis unit that synthesizes and outputs speech based on the intermediate notation. In a data-to-speech device that synthesizes text based on fixed data with a specific format consisting of one or more data and outputs it as a voice,
A fixed data storage unit for storing a plurality of the fixed data;
Viewed including the position for embedding the data in the fixed data, the contents of different text example of fixed character part of a sentence, a phrase storage unit that stores plural in the form of intermediate representation for representing speech,
A sentence example selection unit that selects a sentence example to be embedded for each of the read fixed data from the sentence example storage unit by a predetermined method ;
A fixed data conversion unit that converts each data in the fixed data into an intermediate representation;
Based on each data in the fixed form data after conversion and the selected sentence example, each data in the fixed form data is embedded in a predetermined position of the selected sentence example, and a sentence synthesis unit that synthesizes a document;
A data reading device comprising a speech synthesis unit that synthesizes and outputs speech based on a synthesized sentence. Each data of the standard data includes a duplication check flag indicating whether or not duplication check is performed,
The sentence example storage unit stores a sentence example that expresses overlapping portions of the fixed form data together,
The data reading unit according to claim 2 or 3, wherein the sentence example selection unit has a function of checking an overlapping portion of the fixed form data based on the duplication check flag and selecting a sentence example for collecting the corresponding overlapping portions. apparatus. Each data of the fixed form data includes a data set flag indicating a set of data to be read out collectively in the fixed form data,
The sentence example storage unit stores a plurality of sentence examples including only data of the data set in a data set unit,
5. The preceding sentence example selection unit has a function of reading all standard data at once and selecting a corresponding sentence example for a specified data set unit. Data reading device. A computer-readable recording medium recording a program for causing a computer to execute a step of synthesizing a sentence based on fixed data having a specific format composed of one or more data, and a step of outputting the sentence by voice,
A step of said saw including a position for embedding the data in the fixed data, stores a plurality of different text example of the contents of the fixed character portion of the sentence,
Reading the standard data, selecting a sentence example to be embedded for each of the read standard data in a predetermined method, embedding each data in the standard data in a predetermined position, and synthesizing a sentence,
A computer-readable recording medium storing a program to be executed by a computer. A computer-readable recording medium recording a program for causing a computer to execute a step of synthesizing a sentence based on fixed data having a specific format composed of one or more data, and a step of outputting the sentence by voice,
Storing a plurality of the fixed data;
A step of said saw including a position for embedding the data in the fixed data, stores a plurality of different text example of the contents of the fixed character portion of the sentence,
Selecting a sentence example to be embedded for each of the read fixed data from a stored sentence example by a predetermined method ; and
Embedding each data in the fixed form data at a predetermined position of the selected sentence example based on the fixed form data and the selected sentence example, and synthesizing a document;
Analyzing the synthesized document and converting it into an intermediate notation for expressing speech;
Synthesizing and outputting speech based on the intermediate notation,
A computer-readable recording medium storing a program to be executed by a computer.

Images