JPH05260082A - Text reader - Google Patents

Abstract

PURPOSE:To read out a text with the voice quality corresponding to a transmission origin in the case of receiving the text sent from the transmission side in the reception side and reading out the text with a synthetic sound based on the sound synthetic rule and to discriminate a transmitter by means of the voice quality at the time of reading the text out without a receiver becoming tired of listening to the synthetic sound. CONSTITUTION:The text inputted by an input/output device 1 together with a sender ID is transmitted as a mail through a transmission mail system 2, the mail is sent to a reception mail system 5 through a message communication system 4 and received by the system 5. The content of the mail can be outputted as needed, and when the sender ID as well as reception text is sent from an input/output device 7 to a sound synthesizer 6, a voice quality switching section 10 in the synthesizer 6 outputs the text by the synthetic sound with the voice quality corresponding to the transmitter ID.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a text-to-speech device which receives text transmitted from a transmission side of a communication system such as electronic mail on the reception side and reads the received text by synthetic speech by rule-based speech synthesis.

[0002]

2. Description of the Related Art In the rule synthesis of voices, voice parameters obtained by observing or analyzing voices actually uttered by a specific person are consonant + vowel (CV), vowel + consonant + vowel (VCV) and consonant. + Vowel + consonant (CVC) units are prepared in advance as speech units, and these speech units are interpolated and connected according to the input phoneme sequence, and the phonological parameters thus obtained and the pitch generated by the other A prosody parameter consisting of a pattern is sent to a synthesizer to synthesize a voice.

Therefore, even when a text received by the receiving side of a communication system such as an electronic mail is read out by using such a rule synthesizing method for voices, conventionally, the above is prepared in advance in the reading device. I was synthesizing speech using various speech units.

[0004]

As described above, in the conventional text-to-speech apparatus, only the speech units prepared in advance in the apparatus are used to synthesize speech regardless of the sender or the like. The voice quality of the synthetic voice did not change at all times. However, if the text is read aloud constantly with the same voice quality, there is a problem that the user is tired of listening to the synthesized voice and tired of listening.

Therefore, according to the present invention, the voice quality at the time of reading a text is automatically switched according to the source identification code transmitted and received together with the text, so that the receiving side system user gets tired or tired of the synthesized voice. It is a first object of the present invention to provide a text-to-speech device in a communication system such as an electronic mail that does not require the above-described processing and that allows the sender to make a judgment based on the voice quality.

Further, according to the present invention, the transmitting side sends a parameter relating to the voice quality together with the text, and the receiving side synthesizes the voice using the voice quality related parameter, so that the text is read aloud with the voice quality intended by the sender. It is possible,
A second object of the present invention is to provide a text-to-speech device in a communication system such as e-mail that enables the receiving-side system user not to get tired or tired of the synthesized voice and to judge the sender based on the voice quality. And

[0007]

In order to solve the above-mentioned problems, the present invention sends at least a text to a communication medium on the transmitting side of a communication system, receives the text on the receiving side, and synthesizes the received text with a rule of voice. In a text-to-speech device that reads out with a synthetic voice according to, the receiving side is provided with voice quality switching means for switching the voice quality of the synthetic voice that reads out the text received by the receiving side according to the source identification code sent from the transmitting side together with this text. That is the first feature.

Further, the present invention provides a text / voice quality related parameter transmitting means for transmitting at least a parameter related to voice quality among parameters required for rule synthesis of voices when transmitting a text from a transmission side to a reception side. A second feature is that the text is read aloud by rule synthesis using the parameters sent from the text / voice quality related parameter sending means, in preparation for the sending side.

[0009]

According to the above construction, when the text sent from the sending side is received by the receiving side and the text is read aloud by rule-based speech synthesis, the text is not constantly read as a synthetic voice with the same voice quality, but is read aloud. The voice quality of the sound is switched by the voice quality switching means provided on the receiving side according to the source identification code sent from the transmitting side together with the text. This allows the receiving-side system user to know the sender by voice quality without getting tired or tired of the synthesized voice.

Further, according to the above configuration, when the text / voice quality related parameter transmitting means of the transmitting side sends the text to the receiving side, it relates to at least the voice quality among the parameters necessary for the rule synthesis of voice. The parameters (voice quality related parameters) are sent with this text. The receiving side receives the text sent from the sending side, and when reading the text aloud, the voice is regularly synthesized based on the voice quality-related parameter sent together with the text.

As a result, the system user on the transmission side can read the text on the reception side with the voice quality intended by the user. Further, since the voice quality of the voice synthesized on the receiving side varies depending on the sender, the system user on the receiving side can know the sender by the voice quality without getting tired or tired of the synthesized voice.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. (First Embodiment) FIG. 1 is a system configuration diagram showing a first embodiment of an electronic mail system to which the text-to-speech apparatus of the present invention is applied.

In FIG. 1, reference numerals 1 and 7 are input / output devices for inputting / outputting texts, and 2 and 5 are mail systems connected to the input / output devices 1 and 7. The mail systems 2 and 5 control mail processing.

Reference numeral 4 is a message communication system which controls mail transfer between the mail systems 2 and 5, and 3 and 6 are voice synthesizing devices for text reading aloud which are connected to the input / output devices 1 and 7. The voice synthesizers 3 and 6 are provided to switch the voice quality of the synthetic voice read out from the text in accordance with a sender identification code (hereinafter referred to as sender ID) indicating the sender sent together with the text from the sender. Voice quality switching unit 1
0 is provided.

FIG. 2 shows the speech synthesizer 6 (3) in FIG.
It is a block configuration diagram of. The speech synthesizer 6 (3) has, for example, four speech element storage units 11-14. The voice unit storage unit 11 stores a consonant unit Cv including a voice parameter of a consonant including a halfway through a vowel, and a vowel transition unit unit vc when a vowel changes to a consonant. In addition, the voice unit storage units 12 to 14 store vowel units made up of voice parameters of vowels having different voice qualities.

The voice synthesizer 6 (3) also includes a voice unit switching unit 15 for selecting one of the voice unit storage units 12 to 14, a sender ID, and a sender indicated by the sender ID. Sender ID / segment storage unit number correspondence table 16 in which the correspondence with the number (segment storage unit number) assigned to the speech unit storage unit storing the vowel segment unique to
And a segment switching control unit 17. This segment switching control unit 17 responds by referring to the sender ID / segment storage unit number correspondence table 16 based on the sender ID sent from the transmission side from the input / output device 7 (1) in FIG. The phoneme unit switching unit 15 is controlled by the phoneme unit storage unit number to be obtained. The voice unit switching unit 15, the sender ID / unit storage unit number correspondence table 16, and the unit switching control unit 17 configure the voice quality switching unit 10 in FIG.

The voice synthesizer 6 (3) further generates a phoneme sequence based on the language analysis of the text sent from the input / output device 7 (1) of FIG. A linguistic analysis / phoneme length determination unit 18, a prosody parameter generation unit 19 that generates prosodic parameters based on the accent information and phoneme length generated by the language analysis / phoneme length determination unit 18, and a phonological parameter generation unit 20. , Synthesizer filter 21.

The phonological parameter generator 20 includes a language analysis /
The vowel unit storage unit 1 selected by the voice unit and voice unit switching unit 15 stored in the voice unit storage unit 11 according to the phoneme sequence and the phoneme length generated by the phoneme length determination unit 18.
A phoneme parameter stored in one of 2 to 14 is used to generate a phoneme parameter. Also, the synthesizer filter 21
Generates synthetic speech based on the prosody parameter from the prosody parameter generation unit 19 and the phonology parameter from the phonology parameter generation unit 20.

Here, the operation of the configuration shown in FIGS. 1 and 2 is transmitted from the input / output device 1 and the mail system 2 side, received by the mail system 5 and the input / output device 7 side, and sent to the voice synthesizer 6. An example will be described in which text is read aloud.

First, when the text input by the input / output device 1 is sent as a mail through the sending mail system 2, the mail is sent to the receiving mail system 5 through the message communication system 4. At the time of transmission, the mail is sent with a text (text information) and a sender ID, which is an identification code indicating the sender, in the mail.

The mail sent to the receiving mail system 5 is received by the mail system 5. The contents of the mail received by the system 5 can be output by the input / output device 7 when necessary. In the configuration of FIG. 1, at this time, the sender / receiver ID is passed from the input / output device 7 to the voice synthesizer 6 together with the received text, so that the voice quality switching unit 10 in the synthesizer 6 is transferred.
By the voice quality switching operation of, the text can be output as a synthesized voice having a voice quality corresponding to the sender ID.

The voice synthesizing process in the voice synthesizing device 6 will be described below. First, a method of creating the speech units stored in the speech unit storage unit 11 and the speech unit storage units 12 to 14 will be described in detail.

When creating a voice segment, first, voice data produced by an announcer or the like is prepared according to a voice production list. Then, this sound data is multiplied by a time window having a fixed time length of about 20 msec, and cepstrum analysis is performed in each window while shifting the fixed time of about 10 msec.

Next, while looking at the power spectrum and voice power of each frame, the cepstrum parameter corresponding to the frame range to be cut out as a segment is extracted,
Let it be a speech unit. FIG. 3A shows an example in which Cv and vc segments are cut out from one phoneme of voice data. In this way, the transitional section is cut out in a relatively wide range (the number of frames is large). The Cv and vc units are stored in the voice unit storage unit 11 as described above.

On the other hand, the vowel part, which is a stationary section, cuts out only one frame of cepstrum parameter. here,
Vowels in the voice data of three speakers (in Japanese,
The cepstrum parameters are cut out from / a /, / i /, / u /, / e /, / o /) as described above, and stored in the speech unit storage units 12 to 14, respectively. Specifically, the speech unit storage units 12 and 13 store speech units (vowel units) created from the voices uttered by two specific senders, and the speech unit storage units. In 14, a voice unit (vowel unit) created from the voice uttered by the announcer is stored.

The information indicating the relationship between the specific two persons and the corresponding speech segment storage sections 12 and 13 includes sender IDs (for example, sender IDs # 1 and # 2) unique to the two speakers. Yes) and the corresponding numbers assigned to the voice unit storage units 12 and 13 are registered in the sender ID / unit storage unit number correspondence table 16.

Now, returning to FIG. 2, the input / output device 7 of FIG.
The text in the mail passed from the voice synthesizer 6 to
It is guided to the language analysis / phoneme length determination unit 18 in the speech synthesizer 6. The linguistic analysis / phoneme length determination unit 18 linguistically analyzes this text to generate a phoneme sequence and accent information, and determines each phoneme length thereof.

The sender ID in the mail delivered from the input / output device 7 of FIG. 1 to the voice synthesizing device 6 has a segment switching control unit 17 in the voice quality switching unit 10 provided in the voice synthesizing device 6. Be led to. The element switching control unit 17 determines that the sender I
Sender ID / segment storage unit number correspondence table 1 based on D
6, the unit storage unit number registered in pair with this sender ID is obtained.

As a result, if the sender ID is ID # 1, the unit number indicating the voice unit storage unit 12 is obtained, and if the sender ID is ID # 2, the unit indicating the voice unit storage unit 13 is obtained. One-sided storage unit number is required. The segment switching control unit 17 gives the obtained segment storage unit number to the segment switching control unit 17.

On the other hand, when the sender ID is other than ID # 1 or ID # 2, that is, the sender ID passed from the input / output device 7 is the sender ID / element storage unit number correspondence table 1
If not registered in No. 6, the segment switching control unit 17
Gives the segment switching control unit 17 a segment storage unit number indicating the speech unit storage unit 14 prepared in advance in the control unit 17.

The speech unit switching unit 15 is a unit switching control unit 1.
Based on the phoneme unit storage unit number given from 7, one of the phoneme unit storage units 12 to 14 designated by the same number is selected and connected to the phoneme parameter generation unit 20 by switching. As a result, the sender ID passed from the input / output device 7 is ID # 1.
If so, the speech unit storage unit 12 is used. If ID # 2, the speech unit storage unit 13 is used. Otherwise, the speech unit storage unit 1 is used.
4 will be connected to the phoneme parameter generation unit 20.

The accent information generated by the linguistic analysis in the linguistic analysis / phoneme length determination unit 18 is passed to the prosody parameter generation unit 19 together with the phoneme length, and the phoneme sequence (information regarding reading) is generated together with the phoneme length. Part 2
Passed to 0.

The prosody parameter generator 19 is for language analysis /
The accent information and the phoneme length passed from the phoneme length determination unit 18, and a basic pitch irrelevant to the sender, which is held by the phoneme length determination unit 18 (for example, becomes a source of a vowel unit stored in the voice unit storage unit 14). A prosody parameter consisting of a pitch pattern is generated according to the basic pitch of the speaker of the voice.

On the other hand, the phonological parameter generation unit 20 determines the necessary speech units for the vowel units based on the phoneme sequence (phonetic sequence of reading) passed from the language analysis / phoneme length determination unit 18. From one of the voice unit storage units 12 to 14 that are switched and connected by the unit switching unit 15, from the voice unit storage unit 11 for consonant units (here, Cv unit and vc unit). , And each of them is read out and interpolated according to each phoneme length passed from the language analysis / phoneme length determination unit 18 to generate a phoneme parameter.

Therefore, if the sender ID passed from the input / output device 7 is ID # 1, the voice unit storage unit 11 and (the vowel unit created from the voice uttered by the sender of the ID # 1 (Stored) from the speech unit storage unit 12,
If it is ID # 2, the speech unit storage unit 11 and (the ID #
2 is stored from the voice unit storage unit 13 (where the vowel units generated from the voices uttered by the sender are stored), and the voice unit storage unit 11 and the voice unit voices generated by the announcer (otherwise) are stored. The necessary speech units are read out from the speech unit storage unit 14 (in which the vowel units are stored) and used to generate phonological parameters.

Here, the connection between the pieces for the phoneme parameter generation by the phoneme parameter generator 20 is shown in FIG.
It is performed as shown in (b). That is, by inserting and adjusting the repeating section and the interpolation section of the frame, the phonemes are connected while matching the phoneme lengths determined by the language analysis / phoneme length determination unit 18. Here, linear interpolation of each degree of the cepstrum parameter is used as the interpolation method.

The phoneme parameters generated by the phoneme parameter generator 20 are supplied to the synthesizer filter 21. The prosody parameter generated by the prosody parameter generating unit 19 is also supplied to the synthesizer filter 21. The synthesizer filter 21 is, for example, an LMA filter (logarithmic amplitude characteristic approximation filter), generates a sound source pulse based on the prosody parameter from the prosody parameter generation unit 19, and uses the phonology parameter from the phonology parameter generation unit 20 as filter coefficients. Produces a synthetic voice as.

In the voice synthesized in this way, only the vowel part is selected (ie, the voice quality switching section 10).
The voice quality is obtained from the vowel unit read out from one of the voice unit storage units 12 to 14 which is switched and connected to the phoneme parameter generation unit 20 by the voice unit switching unit 15 therein. However, the vowel part is the part that most affects the voice quality.

Therefore, when reading a mail (text inside) sent from a specific sender whose sender ID is ID # 1 or ID # 2, in addition to the vowel part, a person unrelated to the sender (announcer) ) The voice unit created from the voice uttered by () (the information of the voice unit storage unit 11 in which is stored)
Can be read aloud with the voice quality of the sender, even if is commonly used. That is, according to the present embodiment, except for the vowel part, the speech unit stored is a speech unit created from a voice uttered by a specific sender whose sender ID is ID # 1 or ID # 2. It is possible to read aloud with the voice quality of the specific sender even if no section is prepared.

On the other hand, the mail is ID # 1 or ID
If it is from a sender having a sender ID other than # 2, the voice unit storage unit 11 and the voice unit storage unit 14 storing the voice unit created from the voice of the announcer as described above are stored. It is used to read aloud with a certain voice quality.

The means for switching the voice quality in the voice synthesizing device 6 (3) described in the first embodiment is not limited to that of the same embodiment. For example, in order to improve the quality of the synthesized voice. The sound source residual signal may be held at the same time as each of the speech units in the speech unit storage units 11 to 14 and passed to the synthesizer filter 21 at the time of synthesis.

Further, in the first embodiment, the basic pitch irrelevant to the sender is used for the prosody parameter generation in the prosody parameter generation unit 19, but the present invention is not limited to this. For example, the basic pitches of the three utterers of the voice that is the source of the vowel element stored in the speech element storage sections 12 to 14 (three basic pitch storage sections in which are stored)
May be prepared, and the basic pitch corresponding to the voice unit storage unit selected by the voice unit switching unit 15 may be used. In such a case, the voice quality of the synthesized voice becomes closer to the voice quality of the speaker of the voice that is the source of the vowel segment.

(Second Embodiment) Next, a second embodiment of the present invention will be described. FIG. 4 is a system configuration diagram showing a second embodiment of an electronic mail system to which the text-to-speech apparatus of the present invention is applied. The same parts as those in FIG.

In FIG. 4, reference numerals 41 and 47 denote input / output devices for inputting / outputting text and the like. This input / output device 41,
47 is different from the input / output devices 1 and 7 of FIG. 1 in that at the time of sending a text, at least a parameter related to voice quality among parameters necessary for rule synthesis of voices, for example, a vowel phoneme peculiar to the sender. On the other hand, it has a text / voice quality related parameter sending unit 50 that sends together the basic pitch and the residual signal (sound source residual signal) of each vowel. 43,
Reference numeral 46 is a voice synthesizer connected to the input / output devices 41 and 47. The voice synthesizers 43 and 46 read the text sent from the sender side by rule synthesis using the voice quality related parameters sent together with the text.

FIG. 5 shows the voice synthesizer 46 (4 in FIG. 4).
It is a block block diagram of 3). Speech synthesizer 46 (4
3) is a basic pitch storage unit 51 for storing the basic pitch from the transmitting side delivered from the input / output device 47 (41) in FIG. 1 and an external speech unit storage unit for similarly storing vowel units. 52, a sound source residual storage unit 53 for similarly storing the sound source residual signal of each vowel, the consonant element Cv and the vowel transient section element vc described in the first embodiment (FIG. It has a voice unit storage unit 54 (corresponding to the second voice unit storage unit 11) and a control unit 55. The control unit 55
Information storage in the storage units 51 to 53 is controlled.

The speech synthesizer 46 (43) also produces a phoneme sequence and accent information based on the linguistic analysis of the text sent from the input / output device 47 (41) from the sender, and determines the phoneme length. An analysis / phoneme length determination unit 58,
A prosodic parameter generation unit 59 that generates prosodic parameters based on the accent information and the phoneme length generated by the language analysis / phoneme length determination unit 58 and the basic pitch stored in the basic pitch storage unit 51, and a phoneme parameter generation unit 60. , Synthesizer filter 61. The phoneme parameter generation unit 60 uses the speech units stored in the external speech unit storage unit 52 and the speech unit storage unit 54 according to the phoneme sequence and the phoneme length generated by the language analysis / phoneme length determination unit 58. To generate phonological parameters. The synthesizer filter 61 is, for example, an LMA filter, and the prosody parameter generating unit 59.
Based on the prosody parameter from the sound source, the phoneme parameter from the phoneme parameter generation unit 60, and the sound source residual signal from the sound source residual storage unit 53, synthetic speech is generated.

Here, the operations of the configurations shown in FIGS. 4 and 5 are transmitted from the input / output device 41 and the mail system 2 side and received by the mail system 5 and the input / output device 47 side.
An example will be described in which the voice synthesizer 46 reads a text.

First, when the text input by the input / output device 41 is transmitted, the text / voice quality related parameter sending section 50 in the same device 41, together with the text, the voice quality related parameters unique to the sender of the text, For example, information including the vowel element created by the voice uttered by the sender, the basic pitch of the voice of the sender, and the residual signal (sound source residual signal) of each vowel is sent to the mail system 2.

Text / voice quality related parameter sending unit 50
Information including the text and voice quality-related parameters (vowel element, basic pitch, and sound source residual signal of each vowel) sent from the sender is sent as mail via the sender mail system 2 and received via the message communication system 4 to the receiver. It is sent to the mail system 5.

The mail sent to the receiving mail system 5 is received by the mail system 5. The contents of the mail received by the system 5 can be output by the input / output device 47 when necessary. In the configuration of FIG. 4, at this time, the input / output device 47 instructs the voice synthesizer 46 to receive the voice-related parameters (vowel segment,
By passing the basic pitch and the sound source residual signal of each vowel,
The synthesizer 46 can output the text as a synthesized voice having a voice quality corresponding to the sender.

The voice synthesizing process in the voice synthesizing device 46 will be described below. First, the voice quality-related parameters (vowel element, basic pitch and sound source residual signal of each vowel) in the mail passed from the input / output device 47 of FIG. 4 to the voice synthesizer 46 are provided in the voice synthesizer 46. 5 is led to the control unit 55 shown in FIG. The controller 55 controls the input / output device 47.
Among the voice quality-related parameters (unique to the sender) passed from the vowel segment, the vowel segment is stored in the external speech segment storage unit 52, the basic pitch is stored in the basic pitch storage unit 51, and the sound source residual signal of each vowel is the sound source residual signal. The difference is stored in the difference storage unit 53.

The text in the mail passed from the input / output device 47 of FIG. 4 to the voice synthesizer 46 is guided to the language analysis / phoneme length determination unit 58 shown in FIG. Get burned. The language analysis / phoneme length determination unit 58
A linguistic analysis is performed on this text, and a phoneme sequence,
Accent information and phoneme lengths are obtained. The accent information obtained by the language analysis / phoneme length determination unit 58 is transferred to the prosody parameter generation unit 59 together with the phoneme length, and the phoneme sequence is transferred to the phoneme parameter generation unit 60 together with the phoneme length.

The prosody parameter generating section 59 is for language analysis /
A prosody parameter including a pitch pattern is generated according to the accent information and the phoneme length passed from the phoneme length determination unit 58 and the basic pitch unique to the sender stored in the basic pitch storage unit 51.

On the other hand, the phonological parameter generation unit 60, according to the phonological sequence (phonetic sequence of reading) passed from the language analysis / phonological length determination unit 58, outputs necessary speech units and, for vowel units, external speech units. From the storage unit 52, the consonant units (here, the Cv unit and the vc unit) are read from the voice unit storage unit 54, and these phoneme lengths are passed from the language analysis / phoneme length determination unit 58. To interpolate and generate phonological parameters. The connection between the pieces is performed as described with reference to FIG. 3B in the first embodiment.

The phoneme parameters generated by the phoneme parameter generator 60 are supplied to the synthesizer filter 61. The prosody parameter generated by the prosody parameter generating unit 59 is also supplied to the synthesizer filter 61. The synthesizer filter 61 includes a prosody parameter generation unit 59.
Based on the prosody parameter from the sound source residual storage unit 53 and the sound source residual signal of each vowel stored in the sound source residual storage unit 53, and the synthesized speech is generated using the phonological parameter from the phonological parameter generation unit 60 as a filter coefficient. To produce.

As described above, according to the present embodiment (second embodiment), the vowel part voice having the greatest influence on the voice quality of the voice is sent from the transmitting side and stored in the external voice unit storage part 52. It is created based on a segment, that is, a vowel segment created from the voice data of the sender himself. Therefore, in addition to the vowel part, even if a voice unit (information of the voice unit storage unit 54 in which the voice unit is stored) created from a voice uttered by a person (announcer) irrelevant to the sender is commonly used, On the receiving side, the voice can be synthesized with the voice quality of the mail sender.

In the second embodiment, as the voice quality-related parameters, three kinds of vowel pieces, the basic pitch and the sound source residual signal are used. However, for example, only one kind of vowel pieces or vowel pieces are used. It is possible to use only two types including the above, and further to add another parameter such as the speaking rate.

Although the first and second embodiments of the present invention have been described above, the present invention is not limited to these embodiments. For example, there is no limitation on the type of synthesis parameter or the method of connecting speech units, and other synthesis parameters such as LPC (Linear Predictive Coding) other than the cepstrum parameter may be used. In short, the present invention can be variously modified and implemented without departing from the scope of the invention.

[0059]

As described above in detail, according to the text-to-speech apparatus of the present invention, when the receiving side receives the text and reads it out with the synthetic voice, the synthetic voice to be read out is transmitted by the sender identification code sent together with the text. Since the voice quality is switched, the user on the receiving side can judge the sender based on the voice quality without getting tired or tired of the synthesized voice.

Further, according to the present invention, when the text is sent from the sending side to the receiving side, the parameter relating to the voice quality necessary for the rule synthesis of voice is also sent, and the parameter is rule-synthesized at the receiving side. Since it is configured to be used when reading, the text can be read aloud with the voice quality intended by the sender, and the voice quality of the reading differs depending on the sender of the text, so that the receiving user does not get tired or tired of the synthesized voice. Moreover, the sender can be judged by the voice quality.

[Brief description of drawings]

FIG. 1 is a system configuration diagram showing a first embodiment of an electronic mail system to which a text-to-speech device of the present invention is applied.

FIG. 2 is a block diagram showing the configuration of a speech synthesizer 6 (3) in FIG.

FIG. 3 is a diagram for explaining clipping of voice units and connection of voice units.

FIG. 4 is a system configuration diagram showing a second embodiment of an electronic mail system to which the text-to-speech device of the present invention is applied.

5 is a block diagram showing a configuration of a speech synthesizer 46 (43) in FIG.

[Explanation of symbols]

1, 7, 41, 47 ... Input / output device, 2, 5 ... Mail system, 3, 6, 43, 46 ... Voice synthesizing device, 4 ... Message communication system, 10 ... Voice quality switching unit, 11-14, 5
4 ... Speech element storage unit, 15 ... Speech element switching unit, 16 ... Sender ID / unit storage unit number correspondence table, 17 ... Element switching control unit, 18, 58 ... Language analysis / phoneme length determination unit, 1
9, 59 ... Prosodic parameter generation unit, 20, 60 ... Phonological parameter generation unit 21, 61 ... Synthesizer filter, 50 ...
Text / voice quality related parameter sending unit, 51 ... Basic pitch storage unit, 52 ... External speech unit storage unit, 53 ... Sound source residual storage unit.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁵ Identification code Office reference number FI technical display location G10L 5/04 F 8946-5H

Claims (2)

[Claims] 1. A transmission side of a communication system sends at least a text and a transmission source identification code to a communication medium, the reception side receives the text and the transmission source identification code, and the received text is read aloud by synthetic speech by rule synthesis of voice. In the text reading device, the text reading device is provided with a voice quality switching unit for switching the voice quality of the read synthetic speech according to the received transmission source identification code. 2. A text-to-speech device in which at least a text is sent to a communication medium at a transmission side of a communication system, the text is received at a reception side, and the received text is read aloud by a synthetic voice by rule-based synthesis of speech, from a transmission side to a reception side. When sending the text to, the text / voice quality related parameter sending means for sending at least a parameter related to the voice quality among the parameters necessary for the rule synthesis of the voice is provided, and the text / voice quality related parameter is sent to the receiving side. A text-to-speech device, which reads out the text by rule composition using the parameter sent from the sending means.