JP3706495B2 - Speech synthesizer - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a speech synthesizer, and more particularly to a speech synthesizer effective for an in-vehicle type voice guidance device such as a car navigation system.
[0002]
[Prior art]
VICS (Vehicle Information Communication System) that detects the current position of a vehicle, particularly a vehicle, and receives and informs traffic information in the vicinity of the vehicle from the outside has been put into practical use. In addition, navigation systems that perform route guidance by collating a map database with a current position and a destination are already widely used.
[0003]
Further, in recent years, from the viewpoint of safety, in the above-described system, a configuration in which various guides are notified by voice has become common.
[0004]
[Problems to be solved by the invention]
By the way, since the conventional car navigation system that performs voice guidance as described above is notified by voice in real time every time traffic information is generated in the system, the first part of the guidance is displayed when the driver notices it. I often miss it. For example, if the two guidance sentences “National Route 1 and Moriguchi direction are 2km traffic from Dainichi intersection. Even if you can hear "The route is going along the road for a while", you may miss the first guidance text, and it is 2km from the Dainichi intersection in the latter half of the first guidance text. If you miss the part of "National highway No. 1 towards Moriguchi" in the first half, you may have a traffic jam ahead of your direction of travel, or a traffic jam in the opposite direction I don't know what happens.
[0005]
For this reason, it is conceivable that voice guidance data is temporarily stored in the buffer memory so that it can be generated again by the operation of simple and safe means such as a push switch. All the accumulated data will be generated again. Since the buffer memory is usually a FIFO (First In First Out) method, if the buffer memory capacity is large, it was performed before the target text was generated again Guidance will be generated. Further, when the capacity of the buffer memory is small, there is a possibility that not all of the target guidance is buffered.
[0006]
By the way, the information that the user really wants is limited in the voice guidance. For example, in the case of traffic information of a car navigation system, the location, direction, and distance where traffic occurs. Therefore, when the voice guidance is generated again, it is not always necessary to generate the voice guidance again from the head of the guidance sentence, as long as the voice guidance can be generated from the portion desired by the user.
[0007]
SUMMARY OF THE INVENTION The present invention has been made in view of such circumstances, and speech synthesis that enables synthesized speech to be generated again over words of specific types of attributes in synthesized speech generated by speech synthesis. The purpose is to provide a device.
[0008]
Another object of the present invention is to provide a speech synthesizer suitable for a car navigation system, which can be generated again from a part necessary for a user, for example, a place name, a distance, etc., in a sentence notified by synthesized speech. To do.
[0009]
[Means for Solving the Problems]
The speech synthesizer according to the present invention includes a sentence data storage unit that stores sentence data to be generated as a synthesized speech, and a speech that is generated as a synthesized speech by performing language analysis on the data stored in the sentence data storage unit. A speech synthesizer including a synthesis unit, wherein the speech synthesizer detects a start address of a word having a predetermined type of attribute by performing linguistic analysis on the sentence data stored in the sentence data storage unit. Analyzing means for synthesizing and synthesizing voice uttering means for uttering the sentence data stored in the sentence data storage means as synthesized speech from the address designated as the voice synthesis start address, the sentence data storage means storing The start address of the word detected by the analyzing means in the currently stored data is stored, and the first address of the last stored sentence data is stored. Address storage means for storing the address as a speech synthesis start address, and words in the sentence data previously stored with the speech synthesis start address stored in the address storage means when an appropriate operation is performed Voice synthesis start address changing means for changing the voice synthesis start address to the text data storage means when the voice synthesis start address is changed by an operation of the voice synthesis start address changing means. The stored sentence data is generated by synthesized speech from the changed address.
[0010]
In such a speech synthesizer according to the present invention, sentence data stored in the sentence data storage means is linguistically analyzed by the analysis means to detect a start address of a word having a predetermined attribute, and the sentence storage data is detected by the address storage means. The address of a word having a predetermined attribute detected by the analysis means in the data stored in the storage means is stored, and the head address of the last stored sentence data is stored as a speech synthesis start address. . Then, for example, by operating an address changing unit such as a push switch, the speech synthesis start address stored in the address storage unit is changed to the start address of a word in the sentence data stored before that, Text data stored in the text data storage means is generated by synthesized speech from the changed address.
[0011]
The speech synthesizer according to the present invention further includes a map data storage means storing map data, a fixed guidance text data storage means storing fixed guide text data relating to road traffic information, a position detection means, in the above-described configuration. The standard guidance text data should be read out from the standard guidance text data storage means according to the position detected by the position detection means in the map data stored in the map data storage means and generated as synthesized speech A guide sentence creating means for creating a sentence is provided.
[0012]
In such a speech synthesizer according to the present invention, the data of the standard guidance text is stored from the standard guidance text data storage means according to the position detected by the position detection means in the map data stored in the map data storage means. Since the text to be read and generated by the synthesized voice is created, it can be easily generated again when the user misses the voice guidance in the car navigation system.
[0013]
Furthermore, the speech synthesizer according to the present invention is characterized in that, in the above configuration, the word having the predetermined attribute is a place name and / or a number.
[0014]
In such a speech synthesizer according to the present invention, when a user misses a voice guidance in a car navigation system, it can be easily generated again from a place name, a traffic jam distance, a distance to an intersection, etc. Become.
[0015]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the present invention will be described in detail with reference to the drawings illustrating embodiments thereof. FIG. 1 is a block diagram showing a configuration example when the speech synthesizer according to the present invention is applied to a car navigation system.
[0016]
In FIG. 1, reference numeral 9 indicates a control unit of the speech synthesizer according to the present invention, and outside thereof, via a VICS tuner 11, an antenna 10, a GPS receiver 12, and a CD-ROM drive 13. The push switch 14 and the speaker 15 are connected.
[0017]
The tuner 11 for VICS is in the buffer memory 2 for VICS data in the control unit 9, the GPS receiver 12 is in the vehicle position detection unit 4, the CD-ROM drive 13 is in the buffer memory 7 for map data, and the push switch 14 Are connected to the controller 9 and the speaker 15 is connected to the speech synthesizer 8.
[0018]
In addition to the above-described components, the control unit 9 includes a memory 1 in which standard guidance text data is stored in advance, an address movement width calculation unit 3, and a guidance text creation unit 5.
[0019]
In the memory 1, standard guide sentence data is stored in advance. This standard guidance text data is, for example, data such as “Direction is from 2 to 2 km, traffic jam, route along the road for a while”, and such standard guidance text data provides guidance. When given to the text creation unit 5, place name data is inserted before “to” and “from”.
[0020]
The buffer memory 2 for VICS data buffers the VICS data received by the tuner 11 and supplies it to the guide sentence creation unit 5.
[0021]
The vehicle position detection unit 4 detects a vehicle position (current position) from a GPS (Groval Positioning System) signal received by the GPS receiver 12 and supplies the vehicle position (current position) to the guide text creation unit 5 and the map data buffer memory 7.
[0022]
The buffer memory 7 reads from the CD-ROM drive 13 and buffers the map data corresponding to an area within a predetermined range centered on the vehicle position detected by the vehicle position detection unit 4. The map data buffered in the buffer memory 7 is given to the guide sentence creation unit 5.
[0023]
As described above, the guidance sentence creation unit 5 inputs data from the VICS data buffer memory 2, the standard guidance sentence data memory 1, the vehicle position detection unit 4, and the map data buffer memory 7, which will be described later. Create a guide sentence as you do. The created guidance text is output to the voice synthesizer 8 and is generated as a synthesized voice from the speaker 15, and is also sent to the FIFO memory 6 for buffering.
[0024]
The voice synthesizing unit 8 is a speech analysis unit 81 that performs language analysis on the data of the guidance sentence given from the above-mentioned guidance sentence creation unit 5 and a voice that is given from the address movement width calculation unit 3 to the data buffered in the FIFO memory 6. It includes two functions of the synthesized voice generating unit 82 that generates from the speaker 15 as synthesized voice from the synthesis start address.
[0025]
The analysis unit 81 of the speech synthesizer 8 detects various kinds of information necessary for speech synthesis by performing language analysis on the data of the guide sentence given from the guide sentence creation unit 5, and analyzes the analysis data as a result. Store in the FIFO memory 6. Specifically, the analysis unit 81 analyzes the part-of-speech information and the dependency state of each word constituting the sentence to determine the reading, accent, and pose length. For example, the analysis unit 81 uses the model shown in FIGS. 2 (a) and 2 (b). The analysis result as shown in the figure is stored in the FIFO memory 6. In the figure, “S3” is a pause symbol and is set to 100 msec in this embodiment. “↑” and “↓” indicate accent marks that specify the raising or lowering of the accent. “\ Plc \” indicates the start start label of the place name data, and “\ num \” indicates the start start label of the numeric data.
[0026]
The address movement width calculation unit 3 includes the start start address of the place name data as described above and the start of the numeric data in each unit of guidance text (range delimited by “.”) Buffered in the FIFO memory 6. The start address is detected and stored. Finally, the head address of the guidance text data stored in the FIFO memory 6 is stored as the speech synthesis start address P.
[0027]
The synthesized speech generator 82 of the speech synthesizer 8 corresponds to the guidance text data stored in the FIFO memory 6 when the speech synthesis start address P stored in the address movement width calculator 3 is given. By synthesizing speech from the address and outputting it to the speaker 15, it is generated as synthesized speech.
[0028]
When the push switch 14 is operated n times, the speech synthesis start address P stored in the address movement width calculation unit 3 is the place name n times before the guidance sentence data previously stored in the FIFO memory 6. Change to the start start address of data or numeric data. As a result, the voice synthesizer 8 reads out the data buffered in the FIFO memory 6 from the voice synthesis start address P given from the address movement width calculator 3 and outputs it to the speaker 15.
[0029]
Regarding the operation of the speech synthesizer of the present invention having the above-described configuration, the guidance text creation unit 5 makes two guidance texts “National Route 1 and Moriguchi direction are 2 km from the Dainichi intersection.” The following is an example of a case in which guidance sentences such as “The route to travel to” are continuously created.
[0030]
It is found from the VICS data received by the antenna 10 that a 2 km traffic jam has occurred at the intersection in front of the direction of travel of the automobile on which the speech synthesizer of the present invention is mounted. Suppose that it turns out that it is running in the direction of “Moriguchi” in front of “Daiichi Intersection” on National Route 1.
[0031]
In this case, the guide text creation unit 5 reads the standard guide text data stored in the memory 1 in advance, “Direction is from 2, km, traffic jam, route along the road for a while”, CD-ROM drive Guide text from "National Highway 1", "Daichi Intersection" and "Moriguchi" read from the CD-ROM by 13 "National Highway 1 and Moriguchi direction are 2km from the Dainichi Intersection." And the navigation function to create a guidance sentence “A route along the road for a while.” Based on the vehicle position detected by the vehicle position detection unit 4, the control unit 9 transfers the above-described guidance text from the guidance text creation unit 5 to the speech synthesis unit 8 at an appropriate timing. No sound is generated, and the analysis unit 81 performs the following language analysis.
[0032]
First, the analysis unit 81 analyzes the part-of-speech information and the dependency state of each word, determines the reading, accent, and pose length, and the analysis result as shown in the schematic diagram of FIG. To store. In FIG. 2A, “S3” is a pause symbol, and is set to 100 msec in this embodiment. “↑” and “↓” indicate accent marks that specify the raising or lowering of the accent. “\ Plc \” indicates the start start label of the place name data, and “\ num \” indicates the start start label of the numeric data.
[0033]
When the analysis result data obtained by the analysis unit 81 of the speech synthesis unit 8 is stored in the FIFO memory 6 in this way, the address movement width calculation unit 3 stores the speech synthesis start address P, the location name data, and the numeric data. First start addresses A, B, C, and D are stored. However, in the first guide sentence, the address A matches the speech synthesis start address P. Each time the guidance text data is newly stored in the FIFO memory 6, the address movement width calculation unit 3 updates and stores the voice synthesis start address of each guidance text, the local name data, and the head start address of the numeral data, as described above. .
[0034]
As described above, the data from the speech synthesis start address P of the analysis data once stored in the FIFO memory 6 to the head address of the FIFO memory 6 (the end of the analysis data of the guide sentence stored in the FIFO memory 6 at the end) It is transferred to the speech synthesizer 8 and synthesized by the synthesized speech generator 82 in accordance with accents, pose symbols, etc., and the synthesized speech from the speaker 15 is “National Route 1 and Moriguchi direction is 2 km from the Dainichi intersection”. Is generated.
[0035]
Next, the second guidance sentence is transferred from the guidance sentence creation unit 5 to the voice synthesis unit 8 and is analyzed by the analysis unit 81 in the same manner as the first guidance sentence. The result is shown in FIG. It is shown in the schematic diagram of b).
[0036]
When the analysis data of the second guidance text is transferred from the voice synthesizer 8 to the FIFO memory 6 and stored, each address of the analysis data of the first guidance text previously stored in the FIFO memory 6 is stored. Moves by an address width corresponding to the data length of the entire newly stored analysis data. Accordingly, the speech synthesis start address P of the analysis data of the first guidance sentence shown in FIG. 2 (a) and the speech synthesis start address of the analysis data of the second guidance sentence shown in FIG. 2 (b). The addresses A, B, C, and D of the first analysis data are updated and stored in the address movement width calculation unit 3 as the moved addresses A ′, B ′, C ′, and D ′ by the difference in address width from P, respectively. Is done.
[0037]
Then, the second analysis data is transferred to the speech synthesizer 8, synthesized by the synthesized speech generator 82, and generated as synthesized speech from the speaker 15, “This is a route along the way for a while”.
[0038]
By the way, if the user notices that he / she has missed the first guidance information when the second guidance information is generated from the speaker 15, the user operates the push switch. As a result, the control unit 9 causes the synthesized speech generation unit 82 of the speech synthesis unit 8 to immediately stop speech synthesis, and also provides the address movement width calculation unit 3 with the voice synthesis start address P for the second guidance at that time. Move from the beginning of the sentence to the start address of the previous place name data or numeric data. As a result, the speech synthesis start address P moves to D ′, which is the last address of the first start address of the place name data or numerical data of the first guidance sentence, and the data after that address is voiced from the FIFO memory 6. Output to the combining unit 8. As a result, the guidance sentence after the address D ′, that is, “2 km traffic jam. This is a route to follow along for a while” is synthesized by the synthesized voice generating unit 82 and generated again from the speaker 15 with synthesized voice.
[0039]
By the way, in this embodiment, when the push switch 14 is operated n times, the address movement width calculation unit 3 is configured to move the speech synthesis start address P to the head start address of the n-th place name data or numerical data. ing. Therefore, when the user operates the push switch 14 twice when the second guidance information similar to the above is started to be generated from the speaker 15, the speech synthesis start address P is the address C 'of the first guidance sentence. As a result, the guidance message “There is a 2km traffic jam from the Dainichi intersection.
[0040]
FIG. 3 is a flowchart showing an operation procedure when the speech synthesizer of the present invention as described above is applied to a car navigation system.
[0041]
It is first checked whether or not the push switch 14 has been operated (step S11). If “NO”, the guide text generator 5 next checks whether or not a guide text has been created (step S12). If no guidance text has been created, the process returns to step S11.
[0042]
When the guidance text is created by the guidance text creation section 5 (“YES” in step S12), the language analysis is performed by the analysis section 81 of the speech synthesis section 8 (step S13), and the analysis data obtained as a result is obtained. It is stored in the FIFO memory 6 (step S14). At this time, the address movement width calculation unit 3 updates the head of the analysis data of the guidance text newly stored in the FIFO memory 6 to the voice synthesis start address P, and the guidance text previously stored in the FIFO memory 6. Each address of the data is changed according to the data length of the analysis data of the guide sentence newly stored in the FIFO memory 6 (step S15).
[0043]
Next, data from the voice synthesis start address P of the data stored in the FIFO memory 6 to the head address is sent to the voice synthesizer 8 and generated by the synthesized voice generator 82 from the speaker 15 as synthesized voice (step) S16).
[0044]
When the data from the voice synthesis start address P to the head address in step S16 is sent to the voice synthesizer 8 and generation of synthesized voice is started from the speaker 15, the push switch 14 is operated (in step S11). “YES”), the speech synthesis start address P is returned to the previous address (the start start address of the place name data or the numerical data) by the number of operations n (step S21). Occurrence resumes.
[0045]
In the present embodiment, 128 kbit SRAM is used for the FIFO memory 6, it is possible to buffer the guidance text of about 8000 characters, and the address held in the address movement width calculation unit 3. Is up to 5 in addition to the speech synthesis start address P, but is sufficiently practical when general traffic jam information is targeted as in the above-described example of guidance text.
[0046]
Furthermore, in this embodiment, the address held by the address movement width calculation unit 3, in other words, the address detected by the speech synthesizer 8 is the head address of the place name data and the numeric data, but may be either one. Furthermore, in the present embodiment, the address held by the address movement width calculation unit 3 is applied to the car navigation system, and the start address of the place name data and the numeral data is used. However, the start address of other data is used depending on the purpose of use. Needless to say, may be stored in the address movement width calculation unit 3.
[0047]
【The invention's effect】
As described in detail above, according to the speech synthesizer according to the present invention, the sentence data stored in the sentence data storage means is linguistically analyzed by the analysis means, and the start address of the word having a predetermined attribute is detected, The address storage means stores the address of a word having a predetermined attribute detected by the analysis means in the data stored in the sentence data storage means, and the beginning address of the last stored sentence data is speech synthesis start It is stored as an address. Then, for example, by operating an address changing unit such as a push switch, the speech synthesis start address stored in the address storage unit is changed to the start address of a word in the sentence data stored before that, Since the text data stored in the text data storage means is generated by synthesized voice from the changed address, it can be generated without waste when it occurs again when the voice guidance is missed. .
[0048]
Further, according to the speech synthesizer according to the present invention, the data of the standard guidance text from the standard guidance text data storage means according to the position detected by the position detection means in the map data stored in the map data storage means. Is read and a guidance sentence is created, so that it can be easily generated again when the user misses the voice guidance in the car navigation system.
[0049]
Furthermore, according to the speech synthesizer according to the present invention, when the user misses the voice guidance in the car navigation system, it can be easily generated again from the place name, the traffic jam distance, the distance to the intersection, etc. become.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration example when a speech synthesizer according to the present invention is applied to a car navigation system.
FIG. 2 is a schematic diagram showing a result of language analysis by the speech synthesizer according to the present invention.
FIG. 3 is a flowchart showing an operation procedure when the speech synthesizer of the present invention is applied to a car navigation system.
[Explanation of symbols]
1 Memory (for standard guidance text data)
3 Address movement width calculation unit 4 Vehicle position detection unit 5 Guide text creation unit 6 FIFO memory 7 Buffer memory (for map data)
8 Speech synthesis unit
13 CD-ROM drive
14 Push switch
15 Speaker
81 Analysis unit
82 Synthetic voice generator

Claims (3)

A speech synthesizer comprising: sentence data storage means for storing sentence data to be generated as synthesized speech; and speech synthesis means for generating synthesized speech by performing language analysis on data stored in the sentence data storage means The speech synthesizer includes: analyzing means for detecting a start address of a word having a predetermined type attribute by linguistic analysis of sentence data stored in the sentence data storage means; and the sentence data storage means. Synthesized speech utterance means for uttering the stored sentence data as synthesized speech from the address designated as the speech synthesis start address;
Address storage for storing the start address of the word detected by the analysis means in the data stored in the sentence data storage means and storing the start address of the last stored sentence data as a speech synthesis start address Means,
Speech synthesis start address changing means for changing the speech synthesis start address stored in the address storage means to the start address of a word in the previously stored sentence data when an appropriate operation is performed; Prepared,
The synthesized speech generation means, when the speech synthesis start address is changed by the operation of the speech synthesis start address changing means, the text data stored in the text data storage means by the synthesized voice from the changed address A speech synthesizer characterized by being generated. Map data storage means for storing map data;
A fixed guide sentence data storage means for storing fixed guide sentence data relating to road traffic information;
Position detecting means;
Sentence to be generated as synthesized speech by reading out the data of the standard guidance text from the standard guidance text data storage means according to the position detected by the position detection means in the map data stored in the map data storage means The speech synthesizer according to claim 1, further comprising: a guidance sentence creating means for creating a sentence. The speech synthesis apparatus according to claim 2, wherein the word having the predetermined attribute is a place name and / or a number.

Images