JP3433868B2 - E-mail communication media conversion system - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an e-mail communication media conversion system capable of reading an e-mail composed of text by adding accurate reading and natural prosody with synthetic voice in e-mail communication. Is.

[0002]

2. Description of the Related Art In a conventional electronic mail communication system, texts, figures, voices, etc. are generally faithfully transmitted and received. Therefore, for example, it was not possible to listen to a text-only mail by telephone while outing.

On the other hand, text-to-speech converters (both hardware and software exist) for converting texts into synthetic speech are on the market, and there are some which allow a user terminal to read a mail with synthetic speech, but under the present circumstances. , The number of vocabularies in the word dictionary for reading and prosody provision is small, the reading and prosody provision rules are insufficient, etc. , It was often difficult to hear.

FIG. 2 shows a general flow of this type of text-to-speech conversion apparatus. Reference numeral 200 is a Japanese input text. The reading / prosody imparting unit 210 has a simple word dictionary and a reading / prosody imparting rule, receives the input text 200 as an input, imparts reading and prosody information to the input text, and outputs the reading / prosody information 230. The reading / prosody information 230 is expressed in a simple text format and has readings and prosody information such as reading, accent phrase division position, accent phrase accent type, pose position, and pose length. The format of the reading / prosodic information is slightly different for each text-to-speech conversion device, but can be easily converted to the format of each speech-to-speech device. Synthetic speech generator 24
0 receives the reading / prosodic information 230 as input, generates a synthetic voice, and outputs a synthetic voice 250.

In general, the text-to-speech conversion apparatus is compatible with both the case where the input is text and the case where the input is reading / prosodic information. When the input is text, the text / 200 is used as the input and the reading / prosody providing unit 21.
0, the processing of the synthetic speech generation unit 240 is performed, and the synthetic speech 25
When 0 is output and the input is reading / prosodic information, the reading / prosodic information 230 is input, the processing of the synthetic speech generation unit 240 is immediately performed, and the synthetic speech 250 is output.

[0006]

In the conventional electronic mail communication system, it was not possible to listen to a text mail (an electronic mail consisting only of text) by voice. Also,
Even when a text-to-speech device is used to read a text mail aloud, there are problems in reading accuracy and naturalness of prosody.

The present invention has been made to solve the above-mentioned problems of the prior art, and does not depend on the reading / prosody imparting processing capability of the text-to-speech conversion device of the terminal used by the user, and An object of the present invention is to provide an electronic mail communication media conversion system that allows a user terminal such as a mobile terminal having a small disk capacity to read a text mail correctly and read it in a natural prosody.

Further, according to the present invention, even when an e-mail is read out by using the voice output device of the user terminal, the text mail is read correctly and read out in a natural prosody with almost no increase in the amount of data received by the user terminal. It is to provide an electronic mail communication media conversion system that enables the above.

[0009]

In order to achieve the above object, the present invention provides a mail communication network for delivering electronic mail, and a mail communication network connected to the mail communication network.
In an e-mail communication system including a plurality of user terminals for users to send and receive e-mails, reading / prosodic information for generating synthetic voice for a text e-mail using a voice output device on the mail communication network. Is connected to a text output server that generates a synthetic voice as a synthetic voice device when the user specifies, and the text conversion server assigns reading / prosodic information to an email composed of text. The voice output server serves as a voice output device and generates a synthesized voice. If there is a synthesized voice device in the user terminal, this synthesized voice device may be used as a voice output device instead of the voice output server, or a synthesized voice device in the voice server on the mail communication network and the user terminal. May be used as needed.

[0010]

1 is a diagram showing a schematic configuration of an embodiment of an electronic mail communication media conversion system of the present invention. Reference numeral 100 denotes a mail communication network, which is assumed to be an arbitrary mail communication network. The mail delivery method uses the network method. Reference numeral 110 denotes a text conversion server that has a large-scale word dictionary and a highly accurate reading / prosody addition rule, and adds reading / prosody information to a text mail. A voice output server 120 is used to output a synthesized voice when the user does not have a voice synthesizer in the user terminal or wants to hear a high quality voice. 130 is a user terminal of a user who uses e-mail, 140 is a voice output setting file of each user,
Reference numeral 150 is a voice synthesizing device which outputs a synthetic voice based on the reading / prosodic information. The speech synthesizer 150 corresponds to the synthesized speech generator 240 of FIG. 2, and there are various ones such as one configured only with software and one configured as a board. It should be noted that the user terminal 130 includes the speech synthesizer 1
Some terminals have 50 and others do not. If the user terminal 130 has the voice synthesizer 150, the voice output server 120 may be omitted.

The voice output setting file 140 is a file for setting whether or not to generate reading / prosodic information for voice-outputting a mail, and for which mail the reading / prosody information is to be given, and the like. Sets the necessary information in advance. An example of this audio output setting file is shown in FIG.

In FIG. 3, 311 is a voice output flag field, 312 is a sender field, 313 is a keyword field, 314 is a date and time field, 315 is a conversion server field, 316 is a voice output device field, and 317 is an automatic reading field. . Each field consists of each field name and field value. The audio output flag field 311 is "on" or "o".
The field value is “ff”. Here, when “on”, the reading / prosodic information is added to the text mail, and when it is “off”, it is not added.

The fields 312 to 317 are read and
Indicates the condition for adding prosody information. Each field can have multiple field values, and if they have multiple field values, they represent an OR condition. Also, if there is no field or no field value, read any email for that field.
It is the target of prosody information addition.

The sender field 312 is a target of reading / prosodic information addition only for an electronic mail from a sender including any character string of the field value. In the example of FIG. 3, abc @ aaa. bbb. ccc. jp or yyy @
ddd. ee. jp or yyy @ ddd. ee. jp
Only emails from

The keyword field 313 targets only e-mails containing any of the field values in the text for reading / prosodic information addition. In the example of FIG. 3, only electronic mail including the characters “urgent” or “urgent” in the body is targeted.

The field 314 reads / reads only the mail that arrives at the date / time (period) indicated by the date / time field value.
It is the target of prosody information addition. In the example of FIG. 3, since there is no field value, there is no limitation by date and time.

The conversion server field 315 uses a text conversion server to which reading / prosodic information is added as a field value. If there is no field or no field value, the user terminal becomes the text conversion server.
In the example of FIG. 3, conv @ ooo. ppp. The text conversion server jp is specified.

The voice output device field 316 has a device (a voice output server on the network or a voice synthesizer of a user terminal) that generates a synthesized voice as a field value. The format of the reading / prosodic information 230 of FIG. 2 is determined by this field value. In the example of FIG. 3, syaberinbo (here, “Shaberinbo H
G) (representing the product of NTT-IT) is used to generate synthetic speech.

When the field value is "on", the automatic reading field 317 indicates that an e-mail with reading / prosody information arrives, or the user logs in to the user terminal and there is an unread e-mail. , Which means that the user terminal automatically reads aloud a synthetic voice. When the field value is "off", the e-mail is read out only when the user executes the e-mail reading command.

In summary, in the example of the voice output setting file of FIG. 3, "mail address is abc@aaa.b.
bb. ccc. jp or ddd. ee. For mails that include the text string "Urgent" or "Urgent" from the sender that contains the text string "jp," con
v @ ooo. ppp. The text conversion server jp adds reading / prosodic information, and the voice synthesizer sibererbo outputs synthetic voice at the same time when the e-mail with reading / prosodic information arrives.

When the e-mail is centrally managed by the central management server, the voice output setting file 140 as shown in FIG. 3 may be centrally managed by the central management server for each user.

FIG. 4 is a diagram showing a flow of processing in which the present electronic mail communication media conversion system adds reading / prosodic information to electronic mail. The process of FIG. 4 is repeated each time a new electronic mail arrives. Hereinafter, the processing of FIG. 4 will be described in detail.

(1) A new electronic mail (sender of electronic mail ≠ text conversion server) consisting only of text arrives at an arbitrary user (here, R). This means that when the mail communication network 100 is a bucket relay system such as the Internet, the electronic mail arrives at the user terminal on the receiving side. In the case of a centralized management system such as personal computer communication, it means that the electronic mail arrives at a centralized management server (not shown in FIG. 1) for the electronic mail. After the e-mail arrives, move to (2).

(2) Check the voice output flag field 311 of the voice output setting file 140. When the field is “on”, the process proceeds to (3).
If it is "off", the process is terminated.

(3) It is determined whether each condition of the sender field 312, the keyword field 313, and the date / time field 314 of the voice output setting file 14 is satisfied. When the condition is satisfied, the process shifts to (4). If the condition is not satisfied, the process ends.

(4) The arrived electronic mail is transferred to the text conversion server 110 designated by the conversion server field 315 of the voice output setting file 140. After this process, the process moves to (5).

(5) The text conversion server 110 adds reading / prosodic information to the mail header section. After this process, the process moves to (6).

(6) The electronic mail to which the reading / prosodic information is added is transferred to the user R. This completes the process of adding the e-mail reading / prosodic information.

Here, when the mail communication network is a centralized management system such as personal computer communication, the electronic mail is stored in a centralized management server in the network, and the user draws the electronic mail from the server to his own user terminal. Will read. In the case of such a network form, the central management server is the text conversion server 11 of FIG.
0, the voice output server 120 may also be used. In this case, the transfer processing of (4) and (6) in FIG. 4 is unnecessary, and the processing ends in the centralized management server.

FIG. 6 shows an example of e-mail reading / prosodic information adding processing when the field example of the voice output setting file of FIG. 3 is used.

FIG. 6A shows an electronic mail before the reading / prosodic information is added. The sender of this email is "abc @ a
aa. bbb. ccc. jp ”and fills the sender field 312 of FIG. Further, since the text of the e-mail contains the character string “urgent”, the keyword field 313 of FIG. 3 is satisfied. And in FIG.
Since the date / time field 314 is not specified, this FIG.
The electronic mail of (A) is a target for reading and prosody information addition.

FIG. 6B shows an electronic mail after the reading / prosodic information is added. Reading / prosodic information as shown by hatching is added to the header portion. The format of this reading / prosodic information is the syaberinbo format specified in the audio output device field 316 of FIG.
The "reading data [sentence separator accent]" is collected into a sentence (specifically, refer to the accented kana sentence format shown in "Shaberinbo HG Operation Manual" (NTT-IT) P72). In this example, "konichiwa", "korewa", "kinkyuno",
"Mail Death" is read data, "." (With long pause), "/" (weak bond without pause), "*"
(Strong bond without pause) etc. is a phrase separator,
"00" and "04" are the reading positions to be accented.

FIG. 5 is a diagram showing a flow of voice output processing of the electronic mail communication media conversion system. This voice output process is targeted for any mail that arrives at the user.

(1) In the user terminal 130, it is determined whether or not the mail reading command is executed. When it is executed, the process proceeds to (2). If not executed, the process ends. In addition, the automatic reading field 3 in FIG.
When 17 is “on”, in (6) of FIG. 4, this mail read command is executed immediately after the mail to which the reading / prosodic information is added is transferred to the user.

(2) It is determined whether or not the reading / prosody addition information is present in the header portion of the electronic mail in which the mail reading command is executed. If it does not exist, the process proceeds to (3). If it exists, the process proceeds to (4).

(3) The reading / prosodic information setting process of (4), (5), and (6) in FIG. 4 is performed. After this processing, the process moves to (4).

(4) From the reading / prosodic information of the electronic mail header, a voice is generated by the voice output device indicated by the voice output device field 316 in FIG. 3, and the synthesized voice is output from the user terminal 130.

[0038]

As described above, according to the present invention,
By having a large-scale word dictionary and a text conversion server equipped with highly accurate rules for reading and prosody,
Since it is not necessary for the user terminal to have a word dictionary, it is possible to obtain reading / prosodic information that is accurate in reading and has a natural prosody even in a terminal with a small disk capacity such as a mobile terminal. It becomes possible to listen with no natural synthetic voice. Further, particularly when using the synthetic speech device of the user terminal, it becomes possible to listen to the e-mail with natural synthetic speech without reading errors, while increasing the amount of data received by the user terminal.

[Brief description of drawings]

FIG. 1 is a diagram showing a schematic configuration of an embodiment of an electronic mail communication media conversion system of the present invention.

FIG. 2 is a diagram showing a processing flow of a general text-to-speech conversion device.

FIG. 3 is a diagram showing an example of an audio output setting file used in the present invention.

FIG. 4 is a diagram showing a flow of reading / prosodic information addition processing of the electronic mail communication media conversion system of the present invention.

FIG. 5 is a diagram showing a flow of voice output processing of the electronic mail communication media conversion system of the present invention.

FIG. 6 is a diagram showing a specific example of reading / prosody addition processing.

[Explanation of symbols]

100 mail communication network 110 Text conversion server 120 voice output server 130 user terminal 140 audio output setting file 150 speech synthesizer 311 Audio output flag field 312 Sender field 313 Keyword field 314 date and time field 315 Conversion server field 316 Audio output device field 317 Automatic reading field

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H04L 12/58 G06F 13/00 G06F 3/16

Claims (4)

(57) [Claims] 1. A mail communication network for delivering electronic mail, and a connection to the mail communication network,
In an e-mail communication system composed of a plurality of user terminals for sending and receiving e-mails by a user, reading / prosodic information for generating synthetic voice for a text e-mail using a voice output device on the mail communication network. Connect a text conversion server that assigns a voice output server to a voice output server that becomes a synthetic voice device and generates a synthetic voice when the user specifies, and becomes a voice output device and generates a synthetic voice when the user specifies in the user terminal. According to the user's setting information, the text conversion server adds reading / prosodic information to the e-mail composed of text, and the voice output server or the synthesized voice device serves as a voice output device to synthesize. E-mail communication media conversion system characterized by generating voice 2. A mail communication network for delivering electronic mail, and a connection to the mail communication network,
In an e-mail communication system composed of a plurality of user terminals for sending and receiving e-mails by a user, reading / prosodic information for generating synthetic voice for a text e-mail using a voice output device on the mail communication network. Connect a text conversion server that assigns a text output to a voice output server that becomes a synthetic voice device and generates synthetic voice when specified by the user, and the text conversion server sends an e-mail composed of text to the text conversion server based on the setting information of the user. An electronic mail communication media conversion system, characterized in that reading / prosodic information is added, and the voice output server serves as a voice output device to generate synthetic voice. 3. A mail communication network for delivering electronic mail, and a connection to the mail communication network,
In an e-mail communication system composed of a plurality of user terminals for sending and receiving e-mails by a user, reading / prosodic information for generating synthetic voice for a text e-mail using a voice output device on the mail communication network. Connect a text conversion server that assigns, and install a synthetic voice device that becomes a voice output device and generates a synthetic voice in the user terminal when the user specifies it. An e-mail communication media conversion system characterized in that reading / prosodic information is added by a text conversion server, and the synthetic speech device serves as a speech output device to generate synthetic speech. 4. The electronic mail communication media conversion system according to claim 1, 2 or 3, and information specifying whether to add reading / prosodic information of electronic mail,
As a condition for adding the reading / prosodic information of the e-mail
Te, information that specifies that to grant read-prosodic information of only a particular e-mail, information that specifies the text conversion server
Broadcast, e-mail communication media conversion system characterized by comprising means for setting information for specifying an audio output device, part or all of the information of the information that specifies whether read aloud by voice at the same time as the arrival of electronic mail.