JPH04270395A - On-vehicle traffic information providing device - Google Patents

Abstract

PURPOSE:To provide many clear and natural traffic informations to a driver in a voice. CONSTITUTION:From traffic information received by a receiving means, a traffic information message to be outputted is determined by a phrase selecting means, devided into phrases for constituting its traffic information message, its divided phrase is divided into a proper noun clause and other phrase, and other phrase selects a phrase voice waveform from a voice waveform data memory. This device constituting a characteristic feature of obtaining the traffic information by generating the proper noun clause by selecting a syllasble waveform required for constituting the proper clause from waveform data by a proper nuon phrase generating means, and connecting this proper noun clause and other selected by the phrase selecting means so as to become a natural voice. In such a way, the memory is curtailed, the processing time is shortened and the device is miniaturized, and many clear traffic informations can be provided to a driver in a voice.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a traffic information providing device for a motor vehicle that provides a driver with audio traffic information related to the driving of a vehicle.

0002

[Background Art] In recent years, local wireless data communication networks have been used to provide general users with information on constantly changing road congestion, various traffic regulations, parking lot vacancies, etc. for the purpose of traffic safety and smoothness. Experiments have been conducted on this system and good results have been obtained. For example, the new automobile traffic information system (A
MTICS), the summary of which is published in the Journal of the Society of Automotive Technology VOL. 4
2, NO. 2, 1988.

[0003] Traffic information is collected by a traffic information management center, processed, converted into data, and sent to vehicles as radio waves. A method has been considered in which the traffic information communication radio waves are received, demodulated into digital data, and then the necessary traffic information is provided to the driver via a display or an audio output device.

[0004] When providing traffic information in the form of audio, it becomes necessary to generate a large number of proper nouns such as place names and intersection names, and it is semiconductor memory that provides sentences containing a large number of words using the recording and playback method. A large number of is required.

On the other hand, when providing voice messages in a car, it is difficult to hear due to noise, so it is difficult to hear messages with high clarity.
Moreover, high-quality sound is required. The rule synthesis method, which artificially creates and connects speech waveforms corresponding to each character of input text, can output arbitrary speech, making it possible to output a wide variety of sentences, but it is still insufficient in terms of speech quality. It cannot be said that the technical level has been reached.

As a means of synthesizing speech with high clarity, a method has been proposed in which syllable waveforms are registered as a dictionary, and waveforms corresponding to input text are selected and connected to synthesize output speech. For example, there is a speech synthesis device disclosed in Japanese Unexamined Patent Publication No. 1-284898. This device will be explained below.

FIG. 4 is a block diagram showing an example of a conventional speech synthesis device. In FIG. 4, when text to be converted into speech is input from the text input terminal 1, the text analysis means 2 performs morphological analysis such as modulation and part-of-speech analysis, kanji-kana conversion, and accent processing.
Necessary information is sent to the phoneme label string buffer 3 and prosody information generating means 4.

As this information, the phoneme label string buffer 3
The prosodic information generating means 4 includes the number of moras in an exhalation paragraph, the accent type, the utterance speed, etc.

[0009] The prosody information generating means 4 generates a pitch pattern, a time length pattern for each phoneme, and an amplitude pattern according to rules based on this information, and stores a pitch pattern buffer 5, a time length buffer 6, and an amplitude pattern, respectively. Write to buffer 7.

The waveform selection means 8 refers to the phoneme label sequence buffer 3, pitch pattern buffer 5, time length buffer 6, and amplitude buffer 7, and selects an optimal waveform from the waveform dictionary 9.

First, a search phoneme sequence is set, and a search is performed taking phonemes and pitch patterns into consideration. If a waveform candidate is found, the selected waveform is output as is. If a waveform candidate is not found, the waveform generating means 10 generates a voice waveform so as to have a desired waveform.

[0012] Next, a selection is made taking into consideration the time length condition and the amplitude condition in that order. If there is no waveform that satisfies each condition, the waveform candidate closest to the condition is selected, and the waveform modification processing means 11
Deformation processing is performed at. The waveforms from the waveform selection section 8, waveform generation means 10, and waveform modification processing means 11 are transferred to the waveform connection section 1.
Connect at 2.

[0013]

[Problems to be Solved by the Invention] However, conventional speech synthesis devices select appropriate waveforms that take into account prosody, which is particularly important for the naturalness of synthesized speech, and in order to obtain sufficient naturalness. It required a huge amount of dictionary and processing time, making it impossible to apply it to in-vehicle devices.

[0014] This invention was made to solve the above-mentioned problems, and it is possible to reduce the memory, shorten the processing time, and make the device smaller. The purpose of the present invention is to obtain a traffic information providing device for automobiles that can be provided to drivers.

[0015]

[Means for Solving the Problems] An in-vehicle traffic information providing device according to the present invention determines a traffic information message to be output from traffic information data transmitted from outside the vehicle, and classifies the traffic information message into phrases constituting the traffic information message. and divides the divided clauses into proper noun clauses and other clauses, and for other clauses, a clause selection means selects the clause speech waveform from the clause speech waveform data memory, and a clause selection means selects the clause speech waveform from the clause speech waveform data memory, and each clause constituting the proper noun clause A speech waveform data memory that stores speech waveforms corresponding to syllables, a proper noun phrase generation means that selects syllable waveforms necessary to construct a proper noun phrase from the syllable waveform data memory, and a proper noun phrase generation means. A phrase waveform connecting means is provided for generating a traffic information message by making natural sounds from the proper noun phrase and other phrases selected by the phrase selecting means.

[0016]

[Operation] In this invention, the phrase selecting means determines the traffic information message to be output from the traffic information received by the receiving means, classifies the determined traffic information message into phrases, and defines the divided phrases as proper names. It is divided into phrases and other phrases, and for other phrases, a phrase speech waveform is selected from the phrase speech waveform data memory and outputted to the phrase waveform connecting means. Further, the proper noun phrase generating means selects a speech waveform necessary to construct a proper noun phrase from the syllable waveform data memory and outputs it to the phrase waveform connecting means. This phrase waveform connecting means connects the proper noun phrase created by the proper noun phrase generating means and other phrases selected by the phrase selecting means so as to produce a natural sound, and outputs a traffic information message. , for audio output.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the automobile traffic information providing apparatus of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of one embodiment. Reference numeral 101 in the figure is an antenna, which receives traffic information transmitted from outside the vehicle. 1
02 is a receiving means, which receives the radio wave signal received by the antenna 101 and demodulates it into digital data of traffic information.

The phrase selection means 103 processes the traffic information data received by the receiving means 102 and selects phrases constituting a sentence pattern corresponding to the traffic information data. Among the selected phrases, phrases other than some proper nouns such as place names and intersection names are read out from the phrase speech waveform voice data memory 105 with reference to the phrase speech generation dictionary 104 and output.

The phrase speech waveform speech data memory 105
The phrase speech waveform is in the phrase speech generation dictionary 104, and the phrase speech waveform is in the phrase speech generation dictionary 104.
The storage location of each phrase speech waveform is recorded.

On the other hand, proper noun syllable selection means 106 selects proper nouns, such as place names and intersection names, syllable by syllable from among the phrases selected by phrase selection means 103, and configures them into proper nouns.

The proper noun generation dictionary 107 is a string of syllable labels constituting a proper noun, for example, "hi-b" for "Hibiya".
The syllable waveform data memory 108 stores a symbol string indicating the pronunciation "i-ya" and attributes of a pitch pattern unique to proper nouns, and a syllable waveform data memory 108 stores the waveform of each phoneme.

The proper noun syllable selection means 106 refers to the proper noun generation dictionary 107 , reads out the syllable waveform audio data from the memory 108 , and outputs it to the syllable waveform connection processing means 109 .

The phrase waveform connection processing means 109 combines the syllable waveforms output from the proper noun syllable selection means 106, performs processing such as adjusting the time length, and outputs the combined syllable waveforms as a proper noun waveform to the phrase waveform connection means 110.

The phrase waveform connecting means 110 connects each phrase waveform output from the phrase selecting means 103 and the proper noun waveform output from the syllable waveform connecting processing means 109, and outputs it to the audio output means 111 as a traffic information message. do.

The audio output means 110 converts the traffic information message created by the phrase waveform connecting means 110 from audio digital data to an audio waveform signal. The converted audio waveform signal is output as audio by the speaker 112.

Next, the operation will be explained. Traffic information is
It is sent by radio waves from a wireless base station to a vehicle. This radio wave is received by the antenna 101, and the receiving means 1
02 to demodulate into traffic information data. After processing this traffic information data and converting it into the necessary traffic information message, it is output to the driver by voice.

For example, when providing traffic congestion information as shown in FIG. 2, traffic information corresponding to the information type (traffic jam), traffic jam length (1 km), road name (National Route 2), and intersection name (Tamae Bridge) is provided. When the data is sent, a traffic information message such as ``There is a 1km traffic jam at Tamae Bridge intersection on National Route 2'' is composed and output as voice. In addition, “△” in Figure 2
” marks indicate breaks in phrases. At this time, clauses other than intersection names have low diversity, so processing is performed on a clause-by-clause basis.

Each phrase is a phrase speech waveform voice data memory 1
Since the phrase speech waveform is registered in 05 and the storage location of the phrase speech waveform is registered in the phrase speech generation dictionary 104, the phrase selection means 103 selects the phrase based on the traffic information data sent from the receiving means 102. The phrase speech waveform data is read out from the speech waveform speech data memory 105 and output. The text pattern of the output traffic information message often differs depending on the type of traffic information.

[0029] Other types of traffic information include "Parking lot"
, "regulation", "event", etc. In each case, the phrase selection means 103 composes a traffic information message suitable for the traffic information.

However, since it is assumed that there are many proper nouns such as intersection names, they can be arbitrarily synthesized by connecting the original waveforms of phonemes or syllables.

The proper noun generation dictionary 107 includes a string of syllable labels constituting a proper noun (for example, Tamae Bridge in FIG.
-MA-E-BA-SHI'') and a type indicating attributes of a pitch pattern peculiar to proper nouns, which will be described in detail later, are stored.

The proper noun syllable selection means 106 selects the proper noun sent from the clause selection means 103 from the syllable waveform voice data memory 108 by referring to the proper noun generation dictionary 107 based on the pitch pattern and phoneme context. It is read out and output to the syllable waveform connection processing means 109.

The syllable waveform connection processing means 109 combines the syllable waveforms outputted from the proper noun syllable selection means 106, performs processing such as adjusting the amplitude and time length, and outputs them as a proper noun waveform. The syllable waveform audio data can also be transformed in advance so that the time length and amplitude are constant by taking advantage of the characteristics of proper noun synthesis.

Finally, the phrase waveform connecting means 110 completes the sentence of the traffic information message by connecting the proper noun synthesized by the syllable waveform connecting means 109 to each phrase selected by the phrase selecting means 103. . The traffic information message synthesized in this way is output as audio from the speaker 112 by the audio output means 111 and provided to the driver.

Examples of traffic information messages include parking lot information, regulation information, and event information in addition to traffic jams, such as text patterns shown in Table 1 below.

[0036]

[Table 1]

[0037] In these messages, the parking lot name used in the parking lot information, the regulation lamp name used in the regulation information, the event location name used in the event information, etc. are all proper nouns and can change frequently, so these are used as waveforms. It is best to synthesize using an arbitrary synthesis procedure using connections.

Generally, the intonation pattern of proper nouns is as shown in Figure 3.
It is known that it is limited to the two types shown below. For example, "Japanese Phonetics" (written by Yasushi Amanuma et al., P126-P12)
8). Three syllable place names include "Ueno",
``Hakata'' and other cities belong to type 1 (flat type), while ``Nagoya'' and ``Kobe'' belong to type 2 (flat type). A four-syllable place name,
"Osaka" and "Okinawa" are type 1, "Aomori" and "Fukuoka"
etc. belong to type 2. In FIG. 3, the "○" marks indicate the positions of syllables, and the solid lines indicate changes in intonation.

When synthesizing speech, the pitch pattern, which is an important parameter, corresponds to the intonation pattern described above. Therefore, when synthesizing proper nouns by waveform editing, for example, pitch patterns of around 200 Hz, as shown in Table 2 below, Two types of waveforms are prepared: a waveform (a) with a low pitch frequency and a waveform (b) with a high pitch frequency of, for example, around 350 Hz.

[0040]

[Table 2]

[0041] “Ueno” is type 1, so “u” is (a)
, waveforms are selected from (b) for "e" and waveforms for "no" from (b) to synthesize proper nouns. In addition to this, for example, four types of classification methods shown in Table 3 below may be used.

[0042]

[Table 3]

As described above, by using a waveform selection method that focuses on the pitch pattern unique to proper nouns, it becomes easy to select an appropriate waveform that takes into account the prosody, which is particularly important for the naturalness of synthesized speech, and at the same time, it becomes easy to use a dictionary. Easy to maintain and compact.

Furthermore, text analysis such as morphological analysis such as dependency and part-of-speech analysis, kanji-kana conversion, and accent processing is no longer necessary. Therefore, the processing time required for code conversion and waveform selection can be significantly reduced, and it can be put to practical use as an in-vehicle device. Furthermore, since most of the messages provided to the driver are natural voices, sufficient naturalness and clarity can be obtained even inside the vehicle.

[0045]

[Effects of the Invention] As described above, according to the present invention, by focusing on the pitch pattern peculiar to proper nouns, multiple types of fixed phrases in natural speech are prepared, and synthesized sounds of place names, intersection names, etc. are inserted into these. Since the system is configured to create traffic information messages and output them aloud, it is possible to synthesize clear and natural speech without having to analyze text or generate prosodic information, reducing memory and processing time. I can figure it out. Therefore, the device can be miniaturized and can be configured to be sufficiently practical for use in a vehicle.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an automobile traffic information providing device according to an embodiment of the present invention.

FIG. 2 is a schematic diagram showing an example of conversion from traffic information data to a messager applied to the embodiment of FIG. 1;

FIG. 3 is a table diagram showing an example of a sentence pattern applied to the embodiment of FIG. 1;

FIG. 4 is a block diagram showing a conventional speech synthesis device.

[Explanation of symbols]

101 Antenna 102 Receiving means 103 Clause selection means 104 Clause speech generation dictionary 105 Clause speech waveform speech data memory 106
Proper noun syllable selection means 107 Proper noun generation dictionary 108 Speech waveform audio data memory 109
Syllable waveform connection processing means 110 Clause waveform connection means 111 Audio output means 112 Speaker

Claims (1)

[Claims] Claim 1: receiving means for receiving traffic information data transmitted from outside the vehicle; a syllable speech waveform data memory storing prerecorded syllable speech waveforms excluding proper nouns such as place names and intersection names; determining a traffic information message to be output from the traffic information received by the means;
The determined traffic information message is divided into clauses that make up the message, and the divided clauses are distinguished into proper noun clauses and other clauses, and for other clauses, the clause speech waveform is selected from the clause speech waveform data memory. a phrase selection means, a syllable waveform data memory storing syllable waveforms corresponding to each syllable constituting the proper noun phrase, and selecting a syllable waveform necessary for constituting the proper noun phrase from the syllable waveform data memory. and a proper noun phrase generation means that connects the proper noun phrase generation means, and connects the proper noun phrase created by the proper noun phrase generation means and other phrases selected by the phrase selection means so as to create a natural sound, and generates a traffic information message. A traffic information providing device for an automobile, comprising a phrase waveform connecting means for generating a message, and an audio output means for receiving an audio waveform of a traffic information message from the phrase waveform connecting means and outputting it as voice.