JPH04177299A - Sound responding device - Google Patents

Abstract

PURPOSE:To enable a service to be easily understood by a user by a method wherein a fixed part (a guidance) and a non-fixed part (a male) part are outputted through their changed sounds. CONSTITUTION:As a call sound from a telephone set 4 is detected by a network control unit(NCU) 5, it is informed to a host computer 1 that a signal is received. Then, the host computer 1 gives a Chinese character and kana-character sentence coded into a fixed sentence to a sound restricting and synthesizing part 6. A female sound element selecting code for use in selecting a female sound element is given to a sound restriction synthesizing part 6 in advance. In this case, the sound restricting and synthesizing part 6 analyzes the sentence in view of language and converts it into a sound echo code and a melody information so as to generate a synthesized sound. In turn, when a part of the non- fixed sentence is replaced, a male voice element selection code is given to the sound restricting and synthesizing part 6, the sound element is set to the male voice, thereafter the synthesized sound is generated and given to a user. With such an arrangement, the user can listen and understand the sound of the guidance and the male sound easily.

Description

DETAILED DESCRIPTION OF THE INVENTION [Purpose of the Invention (Field of Industrial Application) The present invention relates to a voice response device that responds and outputs information in response to information input from, for example, a telephone or a dedicated terminal device.

(Prior art) Recently, input character code strings are analyzed to obtain phonological sequences and prosodic information, phonological parameters and prosodic parameter strings are generated using rules from that information, and synthesized speech is generated based on these parameter strings. Various speech synthesis devices have been developed that generate .

This type of rule-based speech synthesis device has the advantage of being able to easily generate synthesized speech representing arbitrary words or sentences, compared to conventional speech synthesis devices using a recording/editing method. For this reason, it is attracting attention as an important technology for realizing highly natural man-machine interfaces in combination with voice recognition technology.

On the other hand, currently, there is a service called a PC network that connects personal computers (hereinafter simply referred to as personal computers) or word processors (hereinafter simply referred to as word processors) to networks via telephone lines to provide email communications and various information services. It is.

By combining these two technologies, a system is being constructed in which the contents of mail sent to users of a personal computer network are transmitted by voice via a telephone. This type of device has a regular part (guidance) and an atypical part (guidance).
There are two types of audio output parts (email), but among them,
The guidance section uses a recording and editing method with relatively good sound quality. However, there were drawbacks such as the need for a certain amount of memory and the complexity of controlling the device.

Therefore, along with the improvement in the sound quality of the rule synthesis method, the rule synthesis method is being introduced not only to the email section but also to the guidance section.

Furthermore, by using a rule synthesis method for the guidance section, it becomes possible to easily change the form of the service, but there is a problem in that it is difficult to distinguish between the guidance section and the email section.

Furthermore, in the past, standard sentences were converted from kanji/kana mixed sentences to audio. However, in order to obtain prosodic information from a sentence containing kanji and kana, it is necessary to check it with an accent dictionary and perform language analysis processing, and it may take some time to generate speech after the sentence is input. In some cases, the sound was unnatural. Therefore, there has been a problem that the response of the voice in the guidance part is delayed.

(Problem to be solved by the invention) As mentioned above, in the past, when a user listens to the contents of an e-mail etc. via a telephone, it is difficult to determine whether the content being spoken is the guidance part or the e-mail part. It was difficult to judge.

Furthermore, as described above, in the conventional system, there have been problems such as the voice response in the guidance portion being delayed and the voice often sounding somewhat unnatural.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a voice response device that allows a user to easily distinguish between the voice of the guidance part and the voice of the mail part.

Another object of the present invention is to provide a voice response device that can shorten the response time and improve the naturalness of the voice of the guidance portion.

U Structure of the Invention] (Means for Solving the Problems) A voice response device according to the first invention outputs a voice response to information input from a telephone, and includes a character code in the response part. A voice response device using a rule synthesis method for generating synthesized speech from sequences is provided with means for changing the voice quality of the synthesized speech when the text to be outputted is a fixed form or an unstructured text.

A voice response device according to a second aspect of the present invention is a voice response device that uses a rule synthesis method for generating synthesized voice from a character code string in a response part, and when a sentence to be outputted as a voice is a fixed form,
Equipped with means for generating synthesized speech from pre-registered code strings representing phonemes and information representing prosody, and means for generating synthetic speech from sentences containing Kanji and kana when the sentences to be outputted are atypical. are doing.

(Function) According to the voice response device according to the first invention, by outputting the voice of the standard part (guidance) and the non-standard part (email) with different voice quality, the voice of the guidance part and the voice of the mail part are output. It is possible to easily tell whether a person is a user, and to provide services that are easy for the user to understand.

According to the voice response device according to the second invention, for a non-standard sentence (email), a voice is generated from a sentence containing kanji and kana,
By generating speech for fixed phrases (guidance) from pre-registered phonetic codes and prosodic information, the response time and naturalness of the guidance portion (fixed phrase) can be shortened and naturalness improved. It can be measured.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

First, a first example will be described. FIG. 1 schematically shows the components of a voice response device according to the present invention. That is, the host computer 1 is connected to a personal computer 2 via a telephone line, recognizes the user number, password, etc. sent from the personal computer 2, and connects to the network. The personal computer 2 connected to the network can send and receive information to and from the host computer 1. Furthermore, information can be exchanged with other personal computers 3 via the host computer 1. The speech rule synthesis unit 6 linguistically analyzes the character code sent from the host computer 1 and converts it into a phonetic code containing prosody information. Then, synthesized speech is generated based on that information. This synthesized voice is sent to the NCU (
network control unit) section 5.

On the other hand, the NCU unit 5 is connected to the telephone line, and notifies the host computer 1 of incoming calls, disconnections, PB detection, and BT detection, and sends analog signals given from the voice rule synthesis unit 6 to the telephone line. It is supposed to be done.

In such a configuration, the operation of the first embodiment is similar to that of the second embodiment.
Referring to the main flowchart shown in the figure, explanation will be given by taking as an example a case in which a call is received from a user. First, when the NCU unit 5 detects a ringing sound from the telephone 4, it notifies the host computer 1 that the call has arrived. Then, host computer 1 prints the registered prefix message [This is the network service center]. ” ``Please give me your user number.'' A coded sentence containing kanji and kana is given to the speech rule synthesis unit 6, but since the sentence to be synthesized into speech is a fixed phrase (step S1 in Fig. 2), A female voice segment selection code for selecting a female voice segment is provided to the voice rule synthesis unit 16.

The voice rule synthesis unit 6 receives the given female voice segment selection code and a sentence containing Kanji and kana, sets the voice segment to a female voice (step S2 in FIG. 2), and then linguistically analyzes the sentence to generate a phonological code. is converted into prosody information, and a synthesized speech is generated (step S3 in FIG. 2).

Note that once a speech segment file is selected, the previous state is maintained unless a new segment selection code is input.

The synthesized speech generated in this way is given to the NCU section 5 and output to the telephone set 4 via the telephone line. Here, when the user inputs a user number from the telephone 4, the NCU section 5 detects the PB of the tone signal, encodes it, and transfers it to the host computer 1. The host computer 1 transfers a fixed phrase "Please enter your password" to prompt the input of the password to the voice rule synthesis section 6, and outputs the message as a voice.

Here, when the user enters the PIN number manually from the telephone 4,
The host computer 1 checks whether the previously input user number is valid or not. If the result of this check is that the email is legitimate, and if the user has two emails with the content shown in Figure 3 stored in the host computer 1, the non-standard message ``The email is OO
0 items have been received. ”, replace the 00 part with “2” and make “
I have received 2 emails. ” is given to the speech rule synthesis unit 6. However, since the content to be outputted is an atypical sentence (step S1 in FIG. 2), a male voice unit selection code is given to the speech rule synthesis unit 6. give. The speech rule synthesis unit 6 receives the given male voice segment selection code and a sentence containing kanji and kana, sets the voice segment to a male voice (step S2 in Figure 2), and then linguistically analyzes the sentence. convert it into phonological code and prosodic information, generate synthesized speech (
Step S3 in FIG. 2), informs the user.

Next, if the buzz tone signal from the user is "1", the first email shown in Figure 3 will be sent to you saying, "There will be a patent conference in conference room 4A from 3:00 pm on July 21st. Please come and see us." Please join us.'', and in the case of ``2'', the second email ``July 31st Kosada's travel party has been cancelled.'' is given to the voice rule synthesis unit 6 and output as voice. In addition, if the PIN is not valid for the user number, the female voice segment selection code and ``The PIN is incorrect.'' ``Please enter the user number again.'' are synthesized using voice rules. Part 6
It outputs audio and prompts you to re-enter.

In this way, according to the first embodiment, by generating speech with male voice segments for non-standard sentences and using female voice segments for fixed sentences, ) and fixed phrases (guidance), and this has great practical effects, such as being able to provide services that are easy for users to understand.

Note that the present invention is not limited to the first embodiment described above. For example, the flow of the service in the first embodiment and the content of the message shown in FIG. 3 are not limited to the example described above. In addition, in the first embodiment, in order to distinguish between non-standard sentences and fixed sentences, male voice elements and female voice elements were used, or only female voice elements were used to change the pitch and speaking speed. It may be differentiated by

Next, a second example will be described. The configuration of the second embodiment is the same as that shown in FIG. 1, and the operation of the second embodiment will be explained below, taking as an example a case where a call is received from a user. First, when the NC0 unit 5 detects a ringing tone from the telephone 4, it notifies the host computer 1 that the call has arrived.

Then, host computer 1 says, ``This is the network service center.'' ``Please give me your user number.''
In order to utter the voice, as shown in FIG. wo/do 6
-zo” (message 1) is given to the voice rule synthesis unit 6. The speech rule synthesis unit 6 generates synthesized speech based on these codes.

The synthesized speech generated in this way is given to the NC0 unit 5 and output to the telephone 4 via the telephone line. Here, when the user inputs a user number from the telephone 84, the NC0 section 5 detects the PB of the tone signal, encodes it, and transfers it to the host computer 1. The host computer 1 enters the code "Anshoba'nko°-wo/
"Do°-zo" (message 2) is transferred to the voice rule synthesis section 6, and the message is output as voice. Here, when the user enters the PIN number from the telephone 4, the host computer 1
checks whether the password is valid for the previously input user number. If the result of this check is that the email is valid, and if the user has two emails with the content shown in Figure 3 stored in the host computer 1, the 00 part will be replaced with ``2''. ``Two emails have arrived.'' (Message 3) is given to the speech rule synthesis unit 6, which is a sentence containing kanji and kana. The speech rule synthesis unit 6 receives a given sentence containing kanji and kana, linguistically analyzes the sentence, converts it into a phonetic code and prosody information, generates synthesized speech, and conveys it to the user.

Next, if the buzz tone signal from the user is "1", the first email shown in Figure 3 will be sent to you saying, "There will be a patent conference in conference room 4A from 3:00 pm on July 21st. Please come and see us." Please join us.'', and in the case of ``2'', the second email ``July 31st Kosada's travel party has been cancelled.'' is given to the voice rule synthesis unit 6 and output as voice. Also, if the PIN is not valid for the user number, "Ansho style 6 ngoka 0 / Chigaimahesu 0""Moichito, / Ryoshanojinoko 0-kara, / New ryokushi 0 tekdasai" (message 4) is given to the voice rule synthesis unit 6 to output the voice and prompt for re-input.

In this way, according to the second embodiment, for non-standard sentences, speech is generated from sentences containing kanji and kana, and for fixed sentences, speech is generated from phonological codes and prosody information that have been preliminarily registered. By generating this, it is possible to shorten response time and improve naturalness for fixed phrases (guidance).

Note that the present invention is not limited to the second embodiment described above. For example, the flow of the service in the second embodiment and the content of the message shown in FIG. 4 are not limited to the example described above. In addition, the present invention can be implemented with various modifications without departing from the gist thereof.

[Effects of the Invention] As detailed above, according to the voice response device of the present invention, by outputting the voices of the standard part (guidance) and the non-standard part (email) with different voice quality, This has great practical effects, such as allowing the user to easily distinguish between the two voices and providing services that are easy for the user to understand.

Further, according to the voice response device of the present invention, when using the rule synthesis method for the response part, the voice is generated from a sentence mixed with Kanji and kana for a non-standard sentence (email), and the voice is generated from a sentence mixed with kanji and kana for a non-standard sentence (email), and for a standard sentence (guidance). By generating speech from phonological codes and prosodic information, rather than from sentences containing kanji and kana, it is possible to shorten the time to start speaking and create a natural sound for the speech of fixed sentences (guidance). It has great practical effects, such as being able to provide audio.

[Brief explanation of drawings]

1 to 3 show a first embodiment of the present invention, in which FIG. 1 is a schematic configuration diagram, and FIG. 2 is a flowchart of main parts showing processing for fixed and non-fixed sentences. FIG. 3 is a diagram showing the contents of the mail, and FIG. 4 is a diagram showing the contents of the response message in the second embodiment of the present invention. 1...Host computer (host device), 2.3...Balacon (terminal device), 4...telephone, 5.NCU section, 6
...Speech rule synthesis section. Applicant's representative Patent attorney Takehiko Suzue Figure 1 Figure 2 However, VJ+1, the symbol "/" indicating the accent position, and the symbol "" indicating the accent break, the following 7
It is a symbol that represents the silent interval between the rest phrases, and each one represents 100m5.
""S'J is "ノJ""Su's voiceless blue figure 4

Claims (2)

[Claims] (1) In a voice response device that outputs a voice response to information input from a telephone, and uses a rule synthesis method that generates synthesized voice from a character code string in the response part, the text to be voice output 1. A voice response device comprising means for changing the voice quality of synthesized speech when the voice is a typical one and when the voice is an atypical voice. (2) In a voice response device that uses a rule synthesis method that generates synthesized speech from a character code string in the response part, if the sentence to be output is a fixed form, a pre-registered code string representing phoneme and information representing prosody are used. What is claimed is: 1. A voice response device comprising: a means for generating a synthesized voice from a sentence; and a means for generating a synthesized voice from a sentence containing kanji and kana when the sentence to be outputted is an atypical sentence.