JP2004004952A - Voice synthesizer and voice synthetic method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To surely convey information by voice even under use environment and the user having auditory difficulties that voice synthesis is regarded to be inapplicable under noise in the conventional manner. <P>SOLUTION: The voice synthesizer is provided with a voice synthesizing section which synthesizes the voice according to a text and an emphasis processing section that performs a single or a plurality of phoneme emphasis processing to the voice synthesized by the voice synthetic section. By this invention, the information is surely conveyed even by the user having auditory difficulties and even in use under the noise and its practical effect is great. <P>COPYRIGHT: (C)2004,JPO

Description

{Circle over (1)} The present invention relates to a speech rule synthesis system for converting text to speech, and more particularly to a technique for transmitting speech to a hearing-impaired person or when used under noise.

音 声 Speech rule synthesis technology that converts text to speech is important as one means of transmitting information transmitted in characters in a format that is easy for humans to understand. For example, most of the information sent through an information network is text, and in order to convey a large amount of text information as it is, it is necessary to use a display having a large display capability or print it on paper.

However, when information terminals are downsized and become portable, large-sized displays and printers cannot be used, so converting to voice is most effective. FIG. 55 is a configuration block diagram of a typical device of a conventional speech synthesizer. In FIG. 55, 10 is a text input unit for inputting a target text, 20 is a language processing unit for parsing the text, 30 m is a voice synthesizing unit for synthesizing voice, 40 m is an operating unit for operating voice quality of the synthesized voice, 50 m is voice quality control means for controlling voice quality according to the input of the operation means, and 60 is an electroacoustic transducer. The speech synthesizer 30m controls the speech synthesizer according to the reading information and the prosody information input from the language processing, and stores the speech in a desired synthesis unit such as a vowel / consonant / vowel chain. It has a unit segment database 80 and unit connection means 90m for generating synthesized speech by linking synthesis units.

The operation of the conventional speech synthesizer configured as described above will be described below.

First, the text input means 10 inputs a target text to the language processing means 20. Next, the language processing means 20 performs a syntax analysis on the text input from the text input means 10, generates reading information and prosody information, and outputs the information to the speech synthesis control means 70m. The unit database 80 outputs a synthesis unit to the unit connection unit 90m according to the reading information input from the speech synthesis control unit 70m. The unit connection unit 90m connects the synthesis units input from the unit database 80 in accordance with the prosody information input from the voice synthesis control unit 70m and the control signal input from the voice quality control unit 50m to generate a synthesized voice, and The synthesized speech is output through the converter 60.

(4) Next, a method for creating a speech unit will be described. A speech unit is created by cutting out a waveform of a previously recorded speech in units of CV, VCV, CVC, or the like. Here, C represents a consonant, and V represents a vowel. A combining method using these combining units is called a CV method, a VCV method, a CVC method, or the like.

In the case of the CV method, a unit such as “ka” which is a combination of a consonant k and a vowel a is used. In the case of the VCV method, “aka” is a combination of a vowel a, a consonant k, and a vowel a, and in the case of the CVC method, “kat” is a combination of a consonant k, a vowel a, and a consonant t. Each of them has advantages and disadvantages such as the number of types of segments and the quality of synthesized speech, but in each system, synthesized speech is generated by connecting speech segments one after another.

作成 When creating a speech unit used in such a synthesis method, if the pre-processing is performed so as to easily perform the deformation required at the time of synthesis, the amount of calculation at the time of synthesis can be reduced. For example, at the time of synthesis, it is necessary to correct the pitch so that a desired pitch pattern is obtained, but a method in which a waveform is cut out in advance by windowing in units of a pitch cycle is described in Japanese Patent Application No. 6-302471. The method will be described with reference to the drawings.

FIG. 56 shows a method of cutting out a waveform. As shown in FIG. 56, a mark is provided at the peak position corresponding to the pitch period of the waveform, and a cutout is made around the mark with a window having a length of twice or less the pitch period. The waveform cut out in this way is called a pitch waveform. Unvoiced consonants having no concept of pitch are cut out as a continuous waveform. This is called an initial waveform.

FIG. 57 shows a process at the time of synthesis. As shown in the figure, superimposition is performed so that a desired pitch period is obtained. When increasing the pitch, the overlapping is performed by narrowing the interval between each other, and when decreasing the pitch, the overlapping is performed by increasing the interval.

(4) In such a speech synthesizer and a speech unit creating method, there is a problem that a synthesized speech is difficult to hear when used under noise or when a person with hearing impairment uses. Although current speech synthesis technology cannot achieve sufficient intelligibility even when used in a quiet environment by hearing people, it is more serious when used in noise or when used by hearing-impaired people. There is a decrease in clarity. This is because synthesized speech uses a limited number of speech units, and because of connection processing and deformation processing during synthesis, there is a lot of missing information, and it is susceptible to noise masking and hearing impairment. In the related art, there is a problem that it is difficult to transmit information necessary for voice recognition under noise or hearing impairment.

SUMMARY OF THE INVENTION The present invention is directed to overcoming the above-mentioned conventional problems. A speech synthesis unit that synthesizes speech in accordance with text, and an emphasis processing unit that performs single or multiple phoneme emphasis processing on the speech synthesized by the speech synthesis unit And a speech synthesizer comprising:

Preferably, the emphasis processing is a consonant emphasis processing for performing amplitude emphasis processing of a consonant or a consonant and a subsequent vowel based on phoneme information.

Preferably, the emphasis process is a band emphasis process for emphasizing a consonant frequency band based on phoneme information.

As described above, according to the present invention, information can be reliably transmitted even to a hearing impaired user or use under noisy conditions, and its practical effect is large.

In one embodiment of the present invention, the voice synthesized according to the user's auditory characteristics is subjected to an emphasis process or a process of compressing the dynamic range of the amplitude, or the voice synthesized according to the noise environment of the use scene is subjected to the emphasis process or A process for compressing the dynamic range of the amplitude is performed. In addition, speech is synthesized after applying emphasis processing or processing for compressing the dynamic range of amplitude to the synthesis unit stored in the database according to the hearing characteristics of the user, or stored in the database according to the noise environment of the use scene. After performing the emphasis processing or the processing of compressing the dynamic range of the amplitude on the synthesized unit, the voice is synthesized. Also, speech is synthesized using a synthesis unit that has been subjected to an emphasis process or a process of compressing the dynamic range of the amplitude in advance. When the speech synthesis is interrupted, the speech synthesis is resumed by returning to a point on the text where the content is easy to understand before the stop position based on the result of the language processing. In addition, by setting a portion for performing the emphasis processing based on the language processing, information can be reliably transmitted even to a user with a hearing impairment or use under noise.

According to a first embodiment of the present invention, there is provided a speech synthesizer for synthesizing speech in accordance with a text, and an emphasis processing unit for performing single or plural phoneme emphasis processes on the speech synthesized by the speech synthesizer. It is a synthesis device.

Preferably, the enhancement process is a formant enhancement process.

Preferably, the formant enhancement process is an enhancement process of a peak of a voice spectrum.

Preferably, the formant emphasis process is a process of emphasizing a band including a predetermined formant frequency for each phoneme based on the phoneme information input from the speech synthesis unit to the emphasis processing unit.

音 声 Preferably, the formant enhancement process is an enhancement process of a band including a formant frequency based on the formant information input from the speech synthesis unit to the enhancement processing unit.

Preferably, the emphasis processing is a consonant emphasis processing for performing amplitude emphasis processing of a consonant or a consonant and a subsequent vowel based on phoneme information.

Preferably, the emphasis process is a band emphasis process for emphasizing a consonant frequency band based on phoneme information.

Preferably, a microphone and a control unit for analyzing the environmental sound input from the microphone and controlling the emphasis processing unit based on the physical characteristics of the environmental sound are provided. Further, the control unit analyzes the environmental sound input from the microphone and selects an enhancement processing method used in the enhancement processing unit based on the physical characteristics of the environmental sound.

Preferably, the control device includes an operation unit for a user to adjust a processing method and a degree of emphasis, and a control unit for controlling the emphasis processing unit based on a signal input from the operation unit.

Preferably, there is provided a measuring unit for measuring the auditory characteristics and preferences of the user, and a control unit for controlling the emphasis processing unit based on the auditory characteristics and preferences of the user. Further, the control section selects an emphasis processing method used in the emphasis processing section based on the user's auditory characteristics and preferences input from the measurement section.

Preferably, there are provided storage means for storing the hearing characteristics and preferences of the user, and a control unit for controlling the emphasis processing unit based on the hearing characteristics and preferences of the user. More preferably, the control unit selects an emphasis processing method used in the emphasis processing unit based on the user's auditory characteristics and preferences stored in the storage unit.

Preferably, the apparatus further includes a hearing characteristic reading unit and a control unit, and the control unit controls the emphasis processing unit with reference to a user's hearing characteristics and preferences stored in a recording medium by the hearing characteristic reading unit. Further, the control unit selects an emphasis processing method to be used by the emphasis processing unit based on the user's auditory characteristics and preferences read by the auditory characteristics reader.

A second embodiment of the present invention provides a speech unit database that stores speech in a desired synthesis unit such as a vowel / consonant / vowel chain, a unit transformation unit that emphasizes the synthesis unit, The speech synthesis device includes a speech synthesis unit that synthesizes speech by connecting the synthesis units that have been subjected to the emphasis processing by the unit deformation unit using a target text.

Preferably, the enhancement process is a formant enhancement process.

Preferably, the formant enhancement process is an enhancement process of a peak of a voice spectrum.

Preferably, the formant enhancement process is an enhancement process of a band including a formant frequency predetermined for each phoneme based on the phoneme information.

Preferably, the formant enhancement process is an enhancement process of a band including a formant frequency based on the formant information.

Preferably, the emphasis processing is a consonant emphasis processing for performing amplitude emphasis processing of a consonant or a consonant and a subsequent vowel based on phoneme information.

Preferably, the emphasis process is a band emphasis process for emphasizing a consonant frequency band based on phoneme information.

Preferably, the emphasis process is a closure emphasis process for extending a consonant closure based on linguistic information.

Preferably, the emphasis processing is an extension processing for extending a phoneme length based on linguistic information.

Preferably, a microphone and a control unit for analyzing the environmental sound input from the microphone and controlling the unit deformation unit based on the physical characteristics of the environmental sound are provided. Further, the control unit analyzes the environmental sound input from the microphone, and selects an emphasis processing method used in the unit deformation unit based on the physical characteristics of the environmental sound.

Preferably, the apparatus includes operating means for allowing a user to adjust a processing method and a degree of emphasis, and a control section for controlling a segment deforming section based on a signal input from the operating means.

Preferably, the apparatus further includes a measuring unit for measuring the hearing characteristics and preferences of the user, and a control unit for controlling the segment deformation unit based on the hearing characteristics and preferences of the user. Further, the control unit selects an emphasis processing method to be used in the unit deformation unit based on the user's auditory characteristics and preferences input from the measurement unit.

Preferably, there are provided storage means for storing the hearing characteristics and preferences of the user, and a control unit for controlling the segment deformation unit based on the hearing characteristics and preferences of the user. Further, the control unit selects an emphasis processing method to be used in the unit deformation unit based on the user's auditory characteristics and preferences stored in the storage unit.

Preferably, there is provided a hearing characteristic reading unit and a control unit, and the control unit controls the segment deforming unit with reference to the user's hearing characteristics and preferences stored in a recording medium by the hearing characteristic reading unit. Further, the control unit selects an emphasis processing method to be used in the unit deformation unit based on the user's auditory characteristics and preferences read by the auditory characteristic reading unit.

In a third embodiment of the present invention, a speech unit database for storing speech subjected to phonological enhancement processing in advance in a desired synthesis unit such as a vowel / consonant / vowel chain, and And a voice synthesizing unit for synthesizing voice by connecting the voice synthesizers.

Preferably, a plurality of speech unit databases having different emphasis methods and degrees, a microphone, and an environmental sound input from the microphone are analyzed, and the voice synthesizer uses the voice for voice synthesis based on physical characteristics of the environmental sound. A control unit for selecting the speech unit database.

Preferably, a plurality of speech unit databases having different emphasis methods and degrees, operation means for a user to adjust the state of emphasis, and a speech synthesis unit based on a signal input from the operation means, And a control unit for selecting the speech unit database to be used for the communication.

Preferably, a plurality of speech unit databases having different emphasis methods and degrees, a measuring unit for measuring the auditory characteristics and preferences of the user, and a speech synthesizer for speech synthesis based on the auditory characteristics and preferences of the user. A control unit for selecting the speech unit database to be used.

Preferably, a plurality of speech unit databases having different emphasis methods and degrees, storage means for storing user's auditory characteristics and preferences, and a speech synthesizer for speech synthesis based on the user's auditory characteristics and preferences. A control unit for selecting the speech unit database to be used.

Preferably, there is provided a storage medium storing a plurality of speech unit databases having different emphasis methods and degrees, and speech unit database reading means.

The fourth embodiment of the present invention is a speech synthesizing method in which a speech synthesized by a speech synthesizing unit for synthesizing speech in accordance with a text is subjected to one or more phoneme emphasis processes.

According to a fifth embodiment of the present invention, the speech synthesis unit output from a speech unit database storing speech in a desired synthesis unit such as a vowel / consonant / vowel combination is subjected to emphasis processing. This is a speech synthesizing method for synthesizing speech by connecting the synthesized units subjected to the emphasis processing by a target text.

According to a sixth embodiment of the present invention, a speech synthesis unit output from a speech unit database storing speech subjected to phoneme emphasis processing in advance in a desired synthesis unit such as a chain of vowels / consonants / vowels. Is a speech synthesis method in which the speech is synthesized by connecting to the target text.

(Example 1)
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a first embodiment of the speech synthesizer of the present invention. FIG. 2 shows a flowchart for explaining the operation of the first embodiment, and FIGS. 3, 4, 5, and 6 show a part of the flowchart for explaining the operation. FIGS. 7 and 8 are schematic diagrams of the emphasizing process of the first embodiment. In FIG. 1, the same components or portions as those in FIG. 55 are denoted by the same reference numerals, and thus description thereof will be omitted, and only different portions will be described. 55 is the same as FIG. 55 except that the voice synthesis unit 30m is replaced by the voice synthesis unit 30a, the voice quality control unit 50m is replaced by the voice quality control unit 50a, the operation unit 40m is replaced by the microphone 110, and the auditory characteristic measurement unit 120 is added. Configuration. The speech synthesis unit 30a controls the speech synthesis unit 30a based on the reading information, prosody information, and emphasis unit information input from the language processing unit 20, and converts the voice into a vowel / consonant / vowel chain. The unit database 80 stored in the desired synthesis unit, the phoneme emphasis processing unit 130a that applies the emphasis processing to the synthesis unit stored in the unit database 80, and the synthesis unit processed by the phoneme emphasis unit 130a are connected. It has a unit connecting means 90a for generating a synthesized voice and a compression processing means 140a for performing a compression process for compressing the dynamic range of the amplitude to the synthesized voice generated by the unit connecting means 90a.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 1, 2, 3, 4, 5, and 6.

{Circle around (1)} First, the hearing characteristics of the user are measured by the hearing characteristics measuring means 120 and the measurement result is output to the voice quality control means 50a. (Step 1000). The measuring method is, for example, the measuring method described in Audiology @ Japan Vol. 35, pp. 401-402 in 1992, and 1993, Spring Meeting of the Acoustical Society of Japan, pp. 329-330. The voice quality control unit 50a determines the setting of the emphasis processing based on the measurement result input from the auditory characteristic measurement unit 120 (step 1100). First, the p value indicating the frequency resolution of the user is compared with 15 (step 1110). If the p value is less than 15 in step 1110, the formant emphasis information is set to true (step 1120). If the p value is 15 or more in step 1110, the formant emphasis information is set to false (step 1125). Next, a gap detection threshold value indicating the time resolution of the user is compared with 10 ms (step 1130). If the gap detection threshold is 10 ms or more in step 1130, the consonant emphasis information is set to true (step 1140). If the gap detection threshold is less than 10 ms in step 1130, the consonant emphasis information is set to false (step 1150). Next, the average hearing level of the user less than 2 kHz is compared with the average hearing level of 2 kHz or more (step 1160). When the value obtained by subtracting the average hearing level below 2 kHz from the average hearing level above 2 kHz in Step 1160 is 30 dB or more, the band emphasis information is set to true (Step 1170). If the value obtained by subtracting the average hearing level below 2 kHz from the average hearing level above 2 kHz in step 1170 is less than 30 dB, the band emphasis information is set to false (step 1180). The text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 performs a syntax analysis of the text input from the text input unit 10, generates reading information, prosody information, and emphasis unit information, and outputs the information to the speech synthesis control unit 70a (step 1300). The segment database 80 outputs a synthesis unit to the phoneme enhancement processing unit 130a according to the reading information input from the speech synthesis control unit 70a (step 1400). The phonemic emphasis processing means 130a performs emphasis processing on a synthesis unit according to the emphasis unit information input from the speech synthesis control means 70a and the control signal input from the voice quality control means 50a (step 1500). The phoneme emphasis processing unit 130a determines whether the emphasis unit information input from the speech synthesis control unit 70a is true or false (step 1510). If the emphasis part information is true in step 1510, the time length of the vowel stationary part in the synthesis unit is extended by 20% (step 1520). It is determined whether the formant emphasis information input from the voice quality control means 50a is true or false (step 1530). If the emphasis unit information is false in step 1510, it is determined whether the formant emphasis information input from the voice quality control unit 50a is true or false (step 1530). If the formant emphasis information is true in step 1530, the phoneme emphasis processing means 130a converts the formant as shown in FIG. 7B) according to the formant information corresponding to the synthesis unit stored in the segment database 80 as shown in FIG. The center frequency and bandwidth of each filter in the filter bank are set to selectively pass the included band, and the contrast between the band containing the formant and the band not containing the formant is enhanced as shown in FIG. 7c) (step 1540). ). Next, it is determined whether the consonant emphasis information input from the voice quality control means 50a is true or false (step 1550). If the formant enhancement information is false in step 1530, it is determined whether the consonant enhancement information input from the voice quality control unit 50a is true or false (step 1550). If the consonant emphasis information is true in step 1550, the phoneme emphasis processing means 130a performs processing from the consonants and consonants in the synthesis unit according to the label information corresponding to the synthesis unit stored in the unit database 80 as shown in FIG. The amplitude of the transition is amplified as shown in FIG. 8 (step 1560). Next, it is determined whether the band emphasis information input from the voice quality control means 50a is true or false (step 1570). If the consonant emphasis information is false in step 1560, it is determined whether the band emphasis information input from voice quality control means 50a is true or false (step 1570). If the band emphasis information is true in step 1570, the phoneme emphasis processing means 130a performs high band emphasis processing for emphasizing the band of 2 kHz or more on the consonants in the synthesis unit (step 1580), and connects the synthesis unit to the segment connection means 90a. (Step 1590). If the band emphasis information is false in step 1570, the phoneme emphasis processing means 130a outputs the synthesis unit to the unit connection means 90 (step 1590). The unit connection means 90a synthesizes the synthesis unit input from the phoneme emphasis processing means 130a according to the prosody information and emphasis section information input from the voice synthesis control means 70a to generate synthesized speech (step 1600). First, the segment connection means 90a determines whether the emphasis section information input from the speech synthesis control means 70a is true or false (step 1610). If the emphasis section information is true in step 1610, the segment connection means 90a extends the value of the closure corresponding to the synthesis unit by 20% (step 1620), and synthesizes the synthesized speech in accordance with the prosody information input from the speech synthesis control means 70a. It is generated (step 1630) and output to the compression processing means 140a (step 1640). If the emphasis processing information is false in step 1610, the unit connecting means 90a generates a synthesized voice according to the prosody information input from the voice synthesis control means 70a (step 1630) and outputs it to the compression processing means 140a (step 1640). . The compression processing unit 140a compresses the dynamic range of the amplitude of the synthesized speech generated by the unit connection unit 90a according to the control signal of the voice quality control unit 50a (step 1700). First, the voice quality control unit 50a divides the environmental sound input from the microphone 110 into bands of 1 kHz or less, 1 kHz to 2 kHz, 2 kHz to 4 kHz, and 4 kHz or more, and obtains an average level of 100 ms for each band (step 1710). The average level of the environmental sound of 1 kHz or less is compared with 20 dBSPL / Hz (step 1730). If the average level of the environmental sound of 1 kHz or less is equal to or more than 20 dBSPL / Hz in step 1730, the voice quality control unit 50a determines that the dynamic range of the level of the component of the synthetic voice of 1 kHz or less is the average level value of the environmental sound of 1 kHz or less to 90 dBSPL. The compression processing parameters are set so as to satisfy (step 1740), and the average level of the environmental sound of 1 kHz to 2 kHz is compared with 20 dBSPL / Hz (step 1750). If the environmental sound of 1 kHz or less is less than 20 dBSPL / Hz in step 1730, the average level of the environmental sound of 1 kHz to 2 kHz is compared with 20 dBSPL / Hz (step 1750). If the average level of the environmental sound of 1 kHz to 2 kHz is equal to or higher than 20 dBSPL / Hz in step 1750, the voice quality control means 50 a determines that the dynamic range of the level of the component sound of 1 kHz to 2 kHz of the synthesized voice is the average level of the environmental sound of 1 kHz to 2 kHz. The compression processing parameters are set so as to have a value of up to 90 dBSPL (step 1760), and the average level of the environmental sound at 2 kHz to 4 kHz is compared with 15 dBSPL / Hz (step 1770). If the environmental sound of 1 kHz to 2 kHz is less than 20 dBSPL / Hz in step 1750, the average level of the environmental sound of 2 kHz to 4 kHz is compared with 15 dBSPL / Hz (step 1770). When the average level of the environmental sound of 2 kHz to 4 kHz is equal to or higher than 15 dBSPL / Hz in step 1770, the voice quality control unit 50a determines that the dynamic range of the level of the component of the synthetic voice of 2 kHz to 4 kHz is 2 kHz to 4 kHz. The compression processing parameters are set so as to have a value of up to 80 dBSPL (step 1780), and the average level of the environmental sound of 4 kHz or more is compared with 10 dBSPL / Hz (step 1790). If the environmental sound of 2 kHz to 4 kHz is less than 15 dBSPL / Hz in step 1770, the average level of the environmental sound of 4 kHz or more is compared with 10 dBSPL / Hz (step 1790). If the average level of the environmental sound of 4 kHz or more is 10 dBSPL / Hz or more in step 1790, the voice quality control unit 50a determines that the dynamic range of the level of the component of the synthetic voice of 4 kHz or more is the value of the average level of the environmental sound of 4 kHz or more to 60 dBSPL. The parameters of the compression processing are set so as to satisfy (Step 1800), and a control signal is output to the compression processing means 140a (Step 1810). If the average level of the environmental sound of 4 kHz or more is less than 10 dBSPL / Hz in step 1790, a control signal is output to the compression processing means 140a (step 1810). The compression processing unit 140a performs a compression process on the synthesized speech input from the unit connection unit 90a based on the control signal input from the voice quality control unit 50a (step 1820). The compression processing method is, for example, the processing shown in the Journal of the Acoustical Society of Japan, Vol. 47, pp. 373 to 379. The compression processing means 140a outputs a synthesized voice through the electro-acoustic transducer 60 (step 1900).

(Example 2)
Hereinafter, a second embodiment of the present invention will be described with reference to the drawings.

FIG. 9 is a block diagram showing the configuration of a second embodiment of the speech synthesizer of the present invention. FIG. 10 shows a flowchart for explaining the operation of the second embodiment, and FIG. 11 shows a part of the flowchart for explaining the operation. In FIG. 9, the same components or portions as those in FIG. The voice synthesis unit 30a of FIG. 1 has been replaced by the voice synthesis unit 30b, the voice quality control unit 50a has been replaced by the voice quality control unit 50b, the microphone 110 has been replaced by the operation unit 40b, and the auditory characteristic measurement unit 120 has been replaced by the audio characteristic storage unit 220. Other than that, the configuration is the same as that of FIG. The speech synthesis unit 30b is processed by a speech synthesis control unit 70b, a database unit 200b that stores synthesis units, a compression processing unit 140b that performs compression processing for compressing the dynamic range of the amplitude for each synthesis unit, and a compression processing unit 140b. Unit synthesis means 90b for generating synthesized speech by connecting the synthesized units. The database unit 200b stores a plurality of segment databases 280a to 280n for storing a plurality of different emphasized segments for each emphasized process, a plurality of segment databases 280a to 280n, and a compression processing unit 140b. And a switch 210b for switching the connection with the switch 210b.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 9, 10, and 11.

、 In FIGS. 10 and 11, the same operations as those in FIGS. 2 and 4 are denoted by the same reference numerals, and the description thereof will not be repeated. Only different parts will be described. First, the pre-measured hearing characteristics stored in the hearing characteristics storage means 220 are output to the voice quality control means 50b. (Step 2000). The voice quality control means 50b sets parameters for compression processing based on the auditory characteristics input from the auditory characteristic storage means 220, and outputs the parameters to the compression processing means 140b (step 2100). The parameter setting method of the compression process is, for example, the setting method described in Auditory Study Group Material, Material No. H-95-4, pp. 1-8. The text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 analyzes the syntax of the text input from the text input unit 10, generates reading information, prosody information, and emphasis unit information and outputs the generated information to the speech synthesis control unit 70b (step 1300). The user inputs the type of emphasis and the degree of emphasis to the operation means 40b, and the operation means 40b outputs the input result to the voice quality control means 50b as emphasis selection information (step 2400). The voice quality control unit 50b selects the segment database to which emphasis is applied closest to the emphasis selection information input from the operation unit 40b from the segment databases 280a to 280n, and switches the switch 210b to connect to the compression processing unit 140b ( Step 2500). The segment database 280 connected to the compression processing means 140b in step 2500 outputs a synthesis unit to the compression processing means 140b according to the read information input from the speech synthesis control means 70b (step 2600). The compression processing unit 140b compresses the dynamic range of the amplitude of the synthesis unit input from the unit database 280 according to the compression processing parameter input from the voice quality control unit 50b, and outputs it to the unit connection unit 90b (step 2700). The unit connection means 90b synthesizes the synthesis unit input from the compression processing means 140b according to the prosody information and the emphasis unit information input from the voice synthesis control means 70b to generate a synthesized voice (step 2800). First, the segment connection means 90b determines whether the emphasis section information input from the speech synthesis control means 70b is true or false (step 1610). If the emphasis section information is true in step 1610, the segment connection means 90b extends the time length of the vowel stationary section in the synthesis unit by 20% (step 2920), and further extends the closure value corresponding to the synthesis unit by 20%. Then, a synthesized voice is generated according to the prosody information input from the voice synthesis control unit 70b (step 2930). If the emphasis processing information is false in step 1610, the unit connection means 90b generates a synthesized voice according to the prosody information input from the voice synthesis control means 70b (step 2930). The unit connection means 90b outputs a synthesized voice through the electroacoustic transducer 60 (step 1900).

(Example 3)
Hereinafter, a third embodiment of the present invention will be described with reference to the drawings.

FIG. 12 is a block diagram showing the configuration of a third embodiment of the speech synthesizer of the present invention. FIG. 13 is a flowchart for explaining the operation of the third embodiment. In the configuration of the third embodiment, the description of the same components or portions as those of the second embodiment shown in FIG. 9 will be omitted, and only different portions will be described. The auditory characteristic storage unit 220 in FIG. 9 is replaced by the auditory characteristic reading unit 310, the voice synthesis unit 30b is replaced by the voice synthesis unit 30c, the voice quality control unit 50b is replaced by the voice quality control unit 50c, and the unit databases 380a to 380n, The configuration is the same as that of FIG. 9 except that 320a to 320n are added. The speech synthesis unit 30c has the same configuration as the speech synthesis unit 30b of FIG. 9 except that the speech synthesis control unit 70b of FIG. 9 is replaced by the speech synthesis control unit 70c, and the database unit 200b is replaced by the unit database reading unit 300. It is. Each of the segment databases 380a to 380n is a storage medium that stores a plurality of synthesis units subjected to different types of emphasis and emphasis processing of the degree of emphasis for each emphasis process. The unit database reading unit 300 reads the unit database 380 referenced by the compression processing unit 140b. The hearing characteristics 320a to 320n are storage media in which the previously measured hearing characteristics of a plurality of users are stored for each individual. The auditory characteristic reading unit 310 reads the auditory characteristic referred to by the voice quality control unit 50c.

The operation of the speech synthesizing apparatus of the present embodiment configured as described above will be described below with reference to FIGS.

In FIG. 13, the same operations as those in FIG. 10 are denoted by the same reference numerals, and the description thereof will not be repeated. Only different parts will be described. First, the auditory characteristic reading means 310 reads out the auditory characteristic 320 corresponding to the preset user and outputs it to the voice quality control means 50c. (Step 3000). The voice quality control unit 50c sets parameters for compression processing based on the auditory characteristics input from the auditory characteristic reading unit 310, and outputs the parameters to the compression processing unit 140b (step 2100). The text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 performs syntax analysis on the text input from the text input unit 10, generates reading information, prosody information, and emphasis unit information, and outputs the generated information to the speech synthesis control unit 70c (step 1300). The unit database reading means 300 reads out the synthesis unit from the unit database 380 set in advance according to the user's preference and the scene to be used according to the reading information input from the speech synthesis control means 70c and outputs it to the compression processing means 140b. (Step 3600). The compression processing unit 140b compresses the dynamic range of the amplitude of the synthesis unit input from the unit database 380 according to the compression processing parameter input from the voice quality control unit 50c, and outputs it to the unit connection unit 90b (step 2700). The unit connection means 90b synthesizes the synthesis unit input from the compression processing means 140b according to the prosody information and the emphasis unit information input from the voice synthesis control means 70c to generate a synthesized voice (step 2800). The unit connection means 90b outputs a synthesized voice through the electroacoustic transducer 60 (step 1900).

(Example 4)
Hereinafter, a fourth embodiment of the present invention will be described with reference to the drawings.

FIG. 14 is a block diagram showing the configuration of a fourth embodiment of the speech synthesizer of the present invention. FIG. 15 shows a flowchart for explaining the operation of the fourth embodiment, and FIGS. 16 and 17 show a part of the flowchart for explaining the operation. In FIG. 14, the same components or portions as those in FIG. 1 are denoted by the same reference numerals, and the description thereof will be omitted. The configuration is the same as that of FIG. 1 except that the voice synthesis unit 30a in FIG. 1 is replaced with the voice synthesis unit 30d, the voice quality control unit 50a is replaced with the voice quality control unit 50d, and the auditory characteristic measurement unit 120 is deleted. The speech synthesis unit 30d includes a speech synthesis control unit 70d, a segment database 80 that stores synthesis units, a unit connection unit 90d that connects the synthesis units stored in the unit database 80 to generate synthesized speech, And speech phoneme emphasis processing means 130d for emphasizing the synthesized speech generated by the unit connection means 90d.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 14, 15, 16, 17, and 18. 15, 16, 17, and 18, the same operations as those in FIGS. 2, 4, 5, and 6 are denoted by the same reference numerals, and the description thereof will not be repeated. Only different parts will be described.

First, the text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 performs a syntax analysis on the text input from the text input unit 10, generates reading information, prosody information, and emphasis unit information and outputs the generated information to the speech synthesis control unit 70d (step 1300). The segment database 80 outputs a synthesis unit to the segment connection unit 90d according to the reading information input from the speech synthesis control unit 70d (step 4400). The unit connection unit 90d connects the synthesis units input from the unit database 80 according to the prosody information and the emphasis unit information input from the voice synthesis control unit 70d to generate a synthesized speech, and outputs the synthesized speech to the phoneme emphasis processing unit 130d. (Step 1600). The voice quality control unit 50d sets an emphasis processing method (step 4700). First, the voice quality control means 50d divides the environmental sound input from the microphone 110 into bands of 1 kHz or less, 1 kHz to 2 kHz, 2 kHz to 4 kHz, and 4 kHz or more, and obtains an average level of 100 ms for each band (step 1710). The average level of the environmental sound of 1 kHz or less is compared with the average level of the environmental sound of 1 kHz to 2 kHz and 20 dBSPL / Hz, and the average level of the environmental sound of other bands is compared with 15 dBSPL / Hz (step 4720). When the average level of the environmental sound of 1 kHz or less is 20 dBSPL / Hz or more, the average level of the environmental sound of 1 kHz to 2 kHz is 20 dBSPL / Hz or more, and the average level of the environmental sound of another band is less than 15 dBSPL / Hz, The formant emphasis information is set to true (step 4730), and the consonant emphasis information is set to false (4780). Next, the band emphasis information of all bands is set to false (step 4800), and a control signal is output to the phoneme emphasis processing means 130d (step 4810). If it is determined in step 4720 that the average level of the environmental sound of 1 kHz or less is 20 dBSPL / Hz or more, the average level of the environmental sound of 1 kHz to 2 kHz is 20 dBSPL / Hz or more, and the average level of the environmental sound of another band is 15 dBBSPL / Hz. If not, the formant emphasis information is set to false (step 4740), and the average level of the environmental sound of 1 kHz to 2 kHz is compared with 20 dBSPL / Hz, and the average level of the environmental sound in other bands is compared with 15 dBSPL / Hz (step 4750). . In step 4750, the average level of the environmental sound of 1 kHz to 2 kHz is 20 dBSPL / Hz or more, and the average level of the environmental sound of 2 kHz to 4 kHz is 15 dBSPL / Hz or more and the average level of the environmental sound of 1 kHz or less is less than 20 dBSPL / Hz, and If the average level of the environmental sound of 4 kHz or more is less than 15 dBSPL / Hz, the consonant emphasis information is set to true (step 4760), the band emphasis information of all bands is set to false (step 4800), and the control signal is converted to the phoneme emphasis processing means 130d. (Step 4810). If the average level of the environmental sound of 1 kHz to 2 kHz is 20 dBSPL / Hz or more and the average level of the environmental sound of 2 kHz to 4 kHz is 15 dBSPL / Hz or more and the average level of the environmental sound of 1 kHz or less is less than 20 dBSPL / Hz in step 4750, If the average level of the environmental sound of 4 kHz or more is not less than 15 dBSPL / Hz, the consonant emphasis information is set to false (step 4770), and band emphasis information of each band is set (step 4790). The average level of the environmental sound is compared with 20 dBSPL / Hz (step 1730). If the average level of the environmental sound of 1 kHz or less is 20 dBSPL / Hz or more in step 1730, the band emphasis information of 1 kHz or less is set to true (step 4791), and the average level of the environmental sound of 1 kHz to 2 kHz is compared with 20 dBSPL / Hz. (Step 1750). If the environmental sound of 1 kHz or less is less than 20 dBSPL / Hz in step 1730, the band emphasis information of 1 kHz or less is set to false (step 4792), and the average level of the environmental sound of 1 kHz to 2 kHz is compared with 20 dBSPL / Hz ( Step 1750). If the average level of the environmental sound of 1 kHz to 2 kHz is equal to or more than 20 dBSPL / Hz in step 1750, the band emphasis information of 1 kHz to 2 kHz is set to true (step 4793), and the average level of the environmental sound of 2 kHz to 4 kHz and 15 dBSPL / Hz are set. Are compared (step 1770). If the environmental sound of 1 kHz to 2 kHz is less than 20 dBSPL / Hz in step 1750, the band emphasizing information of 1 kHz to 2 kHz is false (step 4794) and the average level of the environmental sound of 2 kHz to 4 kHz is compared with 15 dBSPL / Hz. (Step 1770). If the average level of the environmental sound of 2 kHz to 4 kHz is 15 dBSPL / Hz or more in step 1770, the band emphasis information of 2 kHz to 4 kHz is set to true (step 4795), and the average level of the environmental sound of 4 kHz or more and 15 dBSPL / Hz are determined. A comparison is made (step 1790). If the environment sound of 2 kHz to 4 kHz is less than 15 dBSPL / Hz in step 1770, the band emphasis information of 2 kHz to 4 kHz is set to false (step 4796), and the average level of the environment sound of 4 kHz or more is compared with 15 dBSPL / Hz. (Step 1790). When the average level of the environmental sound of 4 kHz or more is 15 dBSPL / Hz or more in step 1790, the band emphasis information of 4 kHz or more is set to true (step 4797), and the control signal is output to the phoneme emphasis processing unit 130d (step 4810). If the average level of the environmental sound of 4 kHz or more is less than 15 dBSPL / Hz in step 1790, the band emphasis information of 4 kHz or more is set to false (step 4798), and the control signal is output to the phoneme emphasis processing means 130d (step 4810). . The phonemic emphasis processing means 130d performs emphasis processing according to the emphasis section information input from the speech synthesis control means 70d and the control signal input from the voice quality control means 50d (step 4900). The phoneme emphasis processing unit 130d determines whether the emphasis unit information input from the speech synthesis control unit 70d is true or false (step 1510). If the emphasis part information is true in step 1510, the time length of the vowel stationary part in the synthesis unit is extended by 20% (step 1520). It is determined whether the formant emphasis information input from the voice quality control means 50d is true or false (step 1530). If the emphasis unit information is false in step 1510, it is determined whether the formant emphasis information input from the voice quality control unit 50d is true or false (step 1530). If the formant emphasis information is true in step 1530, the spectrum envelope of the synthesized speech input from the unit connection means 90d is obtained, and the spectrum peak is emphasized (step 4910). As a method of enhancing the spectral peak, for example, a method shown in pp. 285 to 286 of the Spring Meeting of the Acoustical Society of Japan in 1993 is used. Next, it is determined whether the consonant emphasis information input from the voice quality control means 50d is true or false (step 1550). If the formant emphasis information is false in step 1530, it is determined whether the consonant emphasis information input from the voice quality control means 50d is true or false (step 1550). If the consonant emphasis information is true in step 1550, the phoneme emphasis processing means 130d amplifies the consonant in the synthesis unit and the amplitude of the transition from the consonant to the vowel (step 4920). As a method of emphasizing consonants, for example, a method as shown in IEICE Technical Report, Vol. 91, 513, pp. 31-38, 1992 is used. Next, it is determined whether the band emphasis information of 1 kHz or less input from the voice quality control means 50d is true or false (step 4930). If the consonant emphasis information is false in step 1560, it is determined whether the band emphasis information of 1 kHz or less input from voice quality control means 50 is true or false (step 4930). If the band emphasis information of 1 kHz or less is true in step 4930, the phoneme emphasis processing means 130d performs emphasis processing on the band component of 1 kHz or less of the synthesized speech input from the segment connection means 90d (step 4940). It is determined whether the 2 kHz band emphasis information is true or false (step 4950). If the band emphasis information of 1 kHz or less is false in step 4930, it is determined whether the band emphasis information of 1 kHz to 2 kHz is true or false (step 4950). If the band emphasis information of 1 kHz to 2 kHz is true in step 4950, the phoneme emphasis processing means 130 d performs emphasis processing on the band component of 1 kHz to 2 kHz of the synthesized speech input from the unit connection means 90 d (step 4960). It is determined whether the band emphasis information of 2 kHz to 4 kHz is true or false (step 4970). If the band emphasis information of 1 kHz to 2 kHz is false in step 4950, it is determined whether the band emphasis information of 2 kHz to 4 kHz is true or false (step 4970). If the band emphasis information of 2 kHz to 4 kHz is true in step 4970, the phoneme emphasis processing means 130 d performs emphasis processing of the band component of 2 kHz to 4 kHz of the synthesized speech input from the unit connection means 90 d (step 4980). It is determined whether the band emphasis information of 4 kHz or more is true or false (step 4990). If the band emphasis information of 2 kHz to 4 kHz is false in step 4970, it is determined whether the band emphasis information of 4 kHz or more is true or false (step 4990). If the band emphasis information of 4 kHz or more is true in step 4990, the phoneme emphasis processing unit 130d performs emphasis processing on the band component of 4 kHz or more of the synthesized voice input from the unit connection unit 90d (step 5000). The synthesized speech is output through the converter 60 (step 1900). If the band emphasis information of 4 kHz or more is false in step 4990, a synthesized speech is output through the electroacoustic transducer 60 (step 1900).

(Example 5)
Hereinafter, a fifth embodiment of the present invention will be described with reference to the drawings.

FIG. 19 is a configuration block diagram showing a fifth embodiment of the speech synthesizer of the present invention. FIG. 20 is a flowchart for explaining the operation of the fifth embodiment. In FIG. 19, the same reference numerals are given to the same components or portions as those in FIG. The voice synthesizer 30b of FIG. 9 is replaced by the voice synthesizer 30e, the voice quality control means 50b is replaced by the voice quality control means 50e, the operation means 40b is replaced by the operation means 40e, and the voice quality control means 50b is replaced by the voice quality control means 50e. The configuration is the same as that of FIG. 9 except that the characteristic storage unit 220 is deleted. The speech synthesis unit 30e includes a speech synthesis control unit 70e, a database unit 200e that stores synthesis units, and a segment connection unit 90e that connects the synthesis units to generate synthesized speech. The database unit 200e includes a plurality of segment databases 580a to 580n for storing a plurality of segments subjected to a plurality of compression processes using different parameters for each parameter used in the compression process, and a plurality of segment databases 580a to 580a to n and a switch 210e for switching the connection between the unit connection means 90e.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 19 and 20.

20. In FIG. 20, the same operations as those in FIG. 10 are denoted by the same reference numerals, and thus description thereof will be omitted, and only different portions will be described. First, the text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 analyzes the syntax of the text input from the text input unit 10, generates reading information, prosody information, and emphasis unit information and outputs the generated information to the speech synthesis control unit 70e (step 1300). The user inputs the degree of compression to the operation unit 40e, and the operation unit 40e outputs the input result to the voice quality control unit 50e as compression ratio selection information (step 5400). The voice quality control unit 50e selects a segment database compressed from the segment databases 580a to 580n at the compression ratio closest to the compression ratio selection information input from the operation unit 40e, and switches the switch 210e to switch the segment connection unit. 90e (step 5500). The unit database 580 connected to the unit connecting unit 90e in step 5500 outputs a synthesis unit to the unit connecting unit 90e according to the reading information input from the speech synthesis control unit 70e (step 5600). The unit connection unit 90e connects the synthesis units input from the unit database 580 according to the prosody information and the emphasis unit information input from the voice synthesis control unit 70e to generate a synthesized voice (step 2800), and the electroacoustic converter The synthesized speech is output through the step 60 (step 1900).

(Example 6)
Hereinafter, a sixth embodiment of the present invention will be described with reference to the drawings.

FIG. 21 is a configuration block diagram showing a sixth embodiment of the speech synthesizer of the present invention. FIG. 22 is a flowchart for explaining the operation of the sixth embodiment. In the configuration of the sixth embodiment, the description of the same components or portions as those of the third embodiment shown in FIG. 12 will be omitted, and only different portions will be described. The voice synthesis unit 30c in FIG. 12 is replaced by the voice synthesis unit 30f, the unit databases 380a to 380n are replaced by the unit databases 680a to 680n, and the auditory characteristic reading unit 310, the voice quality control unit 50c, the auditory characteristic reading unit 310, and the auditory characteristic The configuration is the same as that of FIG. 12 except that a to n are deleted. The speech synthesis unit 30f is the same as that shown in FIG. 12 except that the speech synthesis control unit 70c in FIG. 12 is replaced by the speech synthesis control unit 70f, the unit connection unit 90b is replaced by the unit connection unit 90f, and the compression processing unit 140b is deleted. Has the same configuration as that of the voice synthesis unit 30c. The segment databases 680a to 680n are storage media in which segments subjected to a plurality of compression processes using different parameters are stored for each parameter used in the compression process. The unit database reading unit 300 reads the unit database 680 referred to by the unit connecting unit 90f.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 21 and 22.

に お い て In FIG. 22, the same operations as those in FIG. 13 are denoted by the same reference numerals, and the description thereof will not be repeated. Only different parts will be described. First, the text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 analyzes the syntax of the text input from the text input unit 10, generates reading information, prosody information, and emphasis unit information and outputs the information to the speech synthesis control unit 70f (step 1300). The unit database reading unit 300 reads out the synthesis unit from the unit database 680 set in advance according to the user's preference and the scene to be used in accordance with the reading information input from the speech synthesis control unit 70f, and sends it to the unit connection unit 90f. Output (Step 6600). The unit connection unit 90f connects the synthesis units input from the unit database reading unit 300 according to the prosody information and the emphasis unit information input from the voice synthesis control unit 70f to generate synthesized speech (step 2800). A synthesized voice is output through the converter 60 (step 1900).

(Example 7)
Hereinafter, a seventh embodiment of the present invention will be described with reference to the drawings.

FIG. 23 is a configuration block diagram showing a seventh embodiment of the speech synthesizer of the present invention. FIG. 24 is a flowchart for explaining the operation of the seventh embodiment, and FIG. 25 is a part of a flowchart for explaining the operation. In FIG. 23, the same components or parts as those in FIG. The configuration is the same as that of FIG. 1 except that the voice synthesis unit 30a in FIG. 1 is replaced with a voice synthesis unit 30g, and the voice quality control unit 50a is replaced with a voice quality control unit 50g. The voice synthesizing unit 30g of the voice synthesizing unit 30a shown in FIG. The configuration is the same as that of the speech synthesis unit 30a in FIG.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 23, 24 and 25.

{Circle around (1)} First, the auditory characteristic measuring unit 120 measures the user's auditory characteristics, and outputs the measurement result to the voice quality control unit 50g. (Step 1000). The text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 performs a syntax analysis on the text input from the text input unit 10, generates reading information, prosody information, and emphasis unit information and outputs the generated information to the speech synthesis control unit 70g (step 1300). The segment database 80 outputs a synthesis unit to the compression processing unit 140g according to the reading information input from the speech synthesis control unit 70g (step 7400). The compression processing unit 140g compresses the dynamic range of the amplitude of the synthesis unit input from the unit database 80 according to the control signal input from the voice quality control unit 50g (step 7500). First, the voice quality control unit 50g divides the environmental sound input from the microphone 110 into bands of 1 kHz or less, 1 kHz to 2 kHz, 2 kHz to 4 kHz, and 4 kHz or more, and obtains an average level of 100 ms for each band (step 1710). The average level of the environmental sound of 1 kHz or less is compared with the user's minimum audible value of 500 Hz input from the auditory characteristic measuring means 120 (step 7720). In step 7720, if the average level of the environmental sound of 1 kHz or less is equal to or higher than the minimum audible value of 500 Hz of the user, the voice quality control unit 50g determines the average of the environmental sound of the level of the component of the synthetic unit of 1 kHz or less whose dynamic range is 1 kHz or less. The compression processing parameter is set to be a value obtained by adding the minimum audible value of the user's 500 Hz input from the auditory characteristic measuring means 120 to the level value to 90 dBSPL (step 7730), and the environmental sound of 1 kHz to 2 kHz is set. The average level is compared with the user's minimum audible value of 1 kHz input from the auditory characteristic measuring means 120 (step 7750). If the environmental sound of 1 kHz or less is less than the minimum audible value of 500 Hz of the user in step 7730, compression processing parameters are set based on the measurement result input from the auditory characteristic measuring means 120 (step 7740). The average level of the environmental sound of 2 kHz is compared with the minimum audible value of 1 kHz of the user input from the auditory characteristic measuring means 120 (step 7750). The setting method of the compression processing parameters is, for example, the same as in the second and third embodiments. If the average level of the environmental sound of 1 kHz to 2 kHz is equal to or greater than the minimum audible value of 1 kHz of the user input from the auditory characteristic measuring means 120 in step 7750, the voice quality control means 50 g sets the level of the component of 1 kHz to 2 kHz of the synthesis unit. The parameters of the compression processing are set so that the dynamic range becomes 90 dBSPL, which is a value obtained by adding the minimum audible value of the user's 1 kHz input from the auditory characteristic measuring means 120 to the value of the average level of the environmental sound of 1 kHz to 2 kHz. (Step 7760) The average level of the environmental sound of 2 kHz to 4 kHz is compared with the minimum audible value of 2 kHz of the user input from the auditory characteristic measuring means 120 (Step 7780). If the environmental sound of 1 kHz to 2 kHz is lower than the minimum audible value of 1 kHz of the user in step 7750, the compression processing parameters are set based on the measurement result input from the auditory characteristic measuring means 120 (step 7770), and 2 kHz. The average level of the environmental sound of .about.4 kHz is compared with the minimum audible value of the user at 2 kHz input from the auditory characteristic measuring means 120 (step 7780). If the average level of the environmental sound of 2 kHz to 4 kHz is equal to or higher than the minimum audible value of 2 kHz of the user input from the auditory characteristic measuring means 120 in step 7780, the voice quality control means 50 g sets the level of the component of 2 kHz to 4 kHz in the synthesis unit. The parameters of the compression processing are set so that the dynamic range becomes a value obtained by adding the minimum audible value of the user's 2 kHz input from the auditory characteristic measuring means 120 to the value of the average level of the environmental sound of 2 kHz to 4 kHz to 90 dBSPL. (Step 7790) The average level of the environmental sound of 4 kHz or more is compared with the user's minimum audible value of 4 kHz input from the auditory characteristic measuring means 120 (Step 7810). If the environment sound of 2 kHz to 4 kHz is less than the minimum audible value of 2 kHz of the user in step 7780, the compression processing parameters are set based on the measurement result input from the auditory characteristic measuring means 120 (step 7800). The average level of the environmental sound is compared with the minimum audible value of 4 kHz of the user input from the auditory characteristic measuring means 120 (step 7810). In step 7810, if the average level of the environmental sound of 4 kHz or more is equal to or higher than the minimum audible value of 4 kHz of the user input from the auditory characteristic measuring means 120, the voice quality control means 50g determines the dynamic of the level of the component of 4 kHz or more in the synthesis unit. The compression processing parameter is set to a value obtained by adding the minimum audible value of the user's 4 kHz input from the auditory characteristic measuring means 120 to the value of the average level of the environmental sound having a range of 4 kHz or more to 90 dBSPL (step 7820). ), And outputs a control signal to the compression processing means 140g (step 1810). If the environmental sound of 4 kHz or more is less than the minimum audible value of 4 kHz of the user in step 7810, compression processing parameters are set based on the measurement result input from the auditory characteristic measuring means 120 (step 7830), and the compression processing is performed. A control signal is output to the means 140g (step 1810). The compression processing unit 140g performs a compression process on the synthesis unit input from the unit database 80 based on the control signal input from the voice quality control unit 50g, and outputs the result to the unit connection unit 90g (step 7840). The unit connection unit 90g connects the synthesis unit input from the compression processing unit 140g according to the prosody information and the emphasis unit information input from the voice synthesis control unit 70g to generate a synthesized voice (step 7900), and the electro-acoustic converter A synthesized speech is output through the step 60 (step 1900).

(Example 8)
Hereinafter, an eighth embodiment of the present invention will be described with reference to the drawings.

FIG. 26 is a block diagram showing the configuration of an eighth embodiment of the speech synthesizer of the present invention. FIG. 27 is a flowchart for explaining the operation of the eighth embodiment, and FIG. 28 is a flowchart for explaining a part of the operation of the eighth embodiment. FIG. 29 shows a schematic diagram of the processing result of the formant enhancement according to the eighth embodiment. In the configuration of the eighth embodiment, the description of the same components or portions as those of the third embodiment shown in FIG. 12 will be omitted, and only different portions will be described. The configuration is the same as that of FIG. 12 except that the voice synthesis unit 30c in FIG. 12 is replaced by the voice synthesis unit 30h, the voice quality control unit 50c is replaced by the voice quality control unit 50h, and the unit databases 380a to 380n are deleted. In the speech synthesis unit 30h, the speech synthesis control unit 70c in FIG. 12 is replaced by the speech synthesis control unit 70h, the unit database reading unit 300 is replaced by the unit database 80, and the compression processing unit 140b is replaced by the phoneme emphasis processing unit 130h. 12, except that an emphasis filter unit 800 is added. The emphasis filter unit 800 includes a formant emphasis filter 810a to 810n which is set in advance so as to emphasize the formant for each phoneme, and a switch 820 for switching the connection between the formant emphasis filter 810 and the phoneme emphasis processing unit 130h.

The operation of the speech synthesizing apparatus of this embodiment having the above-described configuration will be described below with reference to FIGS. 26, 27, 28, and 29.

In FIGS. 27, 28, and 29, the same operations as those in FIGS. 2, 4, and 13 are denoted by the same reference numerals, and description thereof will be omitted. Only different parts will be described. First, the auditory characteristic reading means 310 reads the auditory characteristic corresponding to the preset user and outputs it to the voice quality control means 50h. (Step 3000). The voice quality control unit 50 determines the setting of the emphasizing process based on the auditory characteristics input from the auditory characteristics reading unit 310, and outputs the setting to the phonemic emphasis processing unit 130h (step 1100). The text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 analyzes the syntax of the text input from the text input unit 10, generates reading information, prosody information and emphasis unit information, and outputs them to the speech synthesis control unit 70h (step 1300). The segment database 80 outputs a synthesis unit to the phoneme enhancement processing unit 130h according to the reading information input from the speech synthesis control unit 70h (step 1400). The phoneme emphasis processing unit 130h performs emphasis processing on a synthesis unit according to the emphasis unit information input from the speech synthesis control unit 70h and the control signal input from the voice quality control unit 50h (step 8500). The phoneme emphasis processing unit 130h determines whether the emphasis unit information input from the speech synthesis control unit 70h is true or false (step 1510). If the emphasis part information is true in step 1510, the time length of the vowel stationary part in the synthesis unit is extended by 20% (step 1520). It is determined whether the formant emphasis information input from the voice quality control means 50h is true or false (step 1530). If the emphasis unit information is false in step 1510, it is determined whether the formant emphasis information input from the voice quality control unit 50h is true or false (step 1530). If the formant emphasis information is true in step 1530, the switch 820 is connected to the formant emphasis filter 810 corresponding to the synthesis unit output from the segment database 80 by the control signal output from the voice synthesis control means 70h (step 8510). ). As shown in FIG. 28, the band including the formant is selectively passed using the filter bank set in advance for each phoneme connected in step 8510, and the band including the formant and the formant are connected as shown in FIG. 7c). The contrast with the band not including is emphasized (step 8540). Next, it is determined whether the consonant emphasis information input from the voice quality control means 50 is true or false (step 1550). If the formant emphasis information is false in step 1530, it is determined whether the consonant emphasis information input from the voice quality control means 50h is true or false (step 1550). If the consonant emphasis information is true in step 1550, the amplitude of the consonant in the synthesis unit and the transition from the consonant to the vowel is amplified (step 1560). Next, it is determined whether the band emphasis information input from the voice quality control means 50h is true or false (step 1570). If the consonant emphasis information is false in step 1560, it is determined whether the band emphasis information input from voice quality control means 50h is true or false (step 1570). If the band emphasis information is true in step 1570, high-band emphasis processing for emphasizing the band of 2 kHz or more is performed on the consonants in the synthesis unit (step 1580), and the phoneme emphasis processing means 130h converts the synthesis unit to the segment connection means 90h. (Step 1590). If the band emphasis information is false in step 1570, the phoneme emphasis processing unit 130h outputs the synthesis unit to the unit connection unit 90h (step 1590). The unit connection unit 90h connects the synthesis units input from the phonemic enhancement processing unit 130h according to the prosody information and the emphasis unit information input from the voice synthesis control unit 70h to generate synthesized speech (step 1600), and performs electro-acoustic conversion. The synthesized speech is output through the device 60 (step 1900).

(Example 9)
Hereinafter, a ninth embodiment of the present invention will be described with reference to the drawings.

FIG. 30 is a block diagram showing the configuration of a ninth embodiment of the speech synthesizer of the present invention. FIG. 31 is a flowchart for explaining the operation of the ninth embodiment, and FIG. 32 is a part of a flowchart for explaining the operation. 30, the same components or portions as those in FIG. 42 are denoted by the same reference numerals, and the description thereof will not be repeated. 42 has the same configuration as that of FIG. 42 except that voice quality control means 50m is replaced by voice quality control means 50i and operation means 40m is replaced by microphone 110.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 30, 31, and 32. In FIG. 31, the same operations as those in FIG. 2 are denoted by the same reference numerals, and thus description thereof will be omitted, and only different portions will be described.

The voice quality control means 50i sets the fundamental frequency of the synthesized voice (step 9100). First, the microphone 110 outputs an environmental sound signal to the voice quality control means 50i (step 9110). The voice quality control means 50i compares the level of the environmental sound input from the microphone 110 with 30 dB (A) (step 9120). If the level of the environmental sound is equal to or more than 30 dB (A) in step 9120, the fundamental frequency is set to be 20% higher than a predetermined standard value (step 9130). Input (step 1200). If the level of the environmental sound is less than 30 dB (A) in step 9120, the text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 performs syntax analysis on the text input from the text input unit 10, generates reading information and prosody information, and outputs the information to the speech synthesis control unit 70m (step 1300). The unit database 80 outputs a synthesis unit to the unit connecting unit 90m according to the reading information input from the speech synthesis control unit 70m (step 9400). The unit connection unit 90m connects the synthesis units input from the unit database 80 in accordance with the prosody information input from the voice synthesis control unit 70m and the control signal input from the voice quality control unit 50i to generate a synthesized speech (step). 9500), and outputs a synthesized voice through the electroacoustic transducer 60 (step 1900).

(Example 10)
Hereinafter, a tenth embodiment of the present invention will be described with reference to the drawings.

FIG. 33 is a block diagram showing the configuration of a tenth embodiment of the speech synthesizer of the present invention. FIG. 34 is a flow chart for explaining the operation of the tenth embodiment, and FIG. 35 is a part of a flow chart for explaining the operation. 33, the same components or portions as those in FIG. 30 are denoted by the same reference numerals, and the description thereof will not be repeated. 30 has the same configuration as that of FIG. 30 except that voice quality control means 50i is replaced by voice quality control means 50j and microphone 110 is replaced by auditory characteristic measurement means 120.

動作 The operation of the speech synthesizer of this embodiment configured as described above will be described below with reference to FIGS. 33, 34 and 35. 34 and 35, the same operations as those in FIGS. 31 and 32 are denoted by the same reference numerals, and the description thereof will not be repeated. Only different parts will be described.

(4) The hearing characteristics of the user are measured by the hearing characteristics measuring means 120 (step 10000). The method of measuring the auditory characteristics is, for example, the same as in the first embodiment. The voice quality control means 50j sets the fundamental frequency of the synthesized voice in accordance with the user's auditory characteristics and preferences inputted from the auditory characteristic measuring means 120 (step 10100). The auditory characteristic measuring means 120 outputs the measurement result of the auditory characteristic to the voice quality control means 50j. (Step 10110). The voice quality control means 50j compares the user's average hearing level below 2 kHz with the average hearing level above 2 kHz (step 10120). If the value obtained by subtracting the average hearing level of less than 2 kHz from the average hearing level of 2 kHz or more in step 10120 is 30 dB or more, the fundamental frequency of the synthesized speech is set to be 20% higher than a predetermined standard value (step 9130). The text input means 10 inputs a target text to the language processing means 20 (step 1200). If the value obtained by subtracting the average hearing level below 2 kHz from the average hearing level above 2 kHz in step 10120 is less than 30 dB, the text input unit 10 inputs the target text to the language processing unit 20 (step 1200). Next, the language processing unit 20 analyzes the syntax of the text input from the text input unit 10, generates reading information and prosody information, and outputs the information to the speech synthesis control unit 70m (step 1300). The unit database 80 outputs a synthesis unit to the unit connecting unit 90m according to the reading information input from the speech synthesis control unit 70m (step 9400). The unit connection unit 90m synthesizes the synthesis unit input from the unit database 80 according to the prosody information input from the voice synthesis control unit 70m and the control signal input from the voice quality control unit 50j to generate a synthesized voice (step 9500). ), And outputs synthesized speech through the electroacoustic transducer 60 (step 1900).

(Example 11)
Hereinafter, an eleventh embodiment of the present invention will be described with reference to the drawings.

FIG. 36 is a block diagram showing a configuration of an eleventh embodiment of the speech synthesizer of the present invention. FIG. 37 is a flowchart for explaining the operation of the eleventh embodiment, and FIG. 38 is a part of a flowchart for explaining the operation. 36, the same components or portions as those in FIG. 33 are denoted by the same reference numerals, and the description thereof will be omitted. Only different portions will be described. The configuration is the same as that of FIG. 33 except that the voice quality control means 50j in FIG. 33 is replaced by the voice quality control means 50k, and the auditory characteristic measuring means 120 is replaced by the auditory characteristic storage means 220.

The operation of the speech synthesizing apparatus according to this embodiment having the above-described configuration will be described below with reference to FIGS. 36, 37, and 38. In FIG. 37, the same operations as those in FIG. 34 are denoted by the same reference numerals, and description thereof will be omitted. Only different portions will be described.

The voice quality control means 50k sets the fundamental frequency of the synthesized voice (step 11100). First, the auditory characteristic storage unit 220 outputs the user's auditory characteristics measured in advance to the voice quality control unit 50k. (Step 11110). The voice quality control unit 50k compares the average hearing level of the user with 40 dBHL (step 11120). If the average hearing level of the user is equal to or higher than 40 dBHL in step 11120, the speech speed of the synthesized voice is set to be 10% slower than a predetermined standard value (step 11130). Is entered (step 1200). If the average hearing level of the user is less than 40 dBHL in step 11120, the text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 performs syntax analysis on the text input from the text input unit 10, generates reading information and prosody information, and outputs the information to the speech synthesis control unit 70m (step 1300). The unit database 80 outputs a synthesis unit to the unit connecting unit 90m according to the reading information input from the speech synthesis control unit 70m (step 9400). The unit connection unit 90m synthesizes the synthesis unit input from the unit database 80 according to the prosody information input from the voice synthesis control unit 70m and the control signal input from the voice quality control unit 50k to generate a synthesized voice (step 9500). ), And outputs synthesized speech through the electroacoustic transducer 60 (step 1900).

(Example 12)
Hereinafter, a twelfth embodiment of the present invention will be described with reference to the drawings.

FIG. 39 is a structural block diagram showing a twelfth embodiment of the speech synthesizer of the present invention. FIG. 40 shows a flowchart for explaining the operation of the twelfth embodiment, and FIG. 41 shows a part of a flowchart for explaining the operation. In FIG. 39, the same components or portions as those in FIG. 36 are denoted by the same reference numerals, and therefore description thereof will be omitted, and only different portions will be described. The configuration is the same as that of FIG. 36 except that the voice quality control unit 50k in FIG. 36 is replaced by the voice quality control unit 501, the auditory characteristic storage unit 220 is replaced by the auditory characteristic reading unit 310, and the auditory characteristics 320a to 320n are added.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 39, 40, and 41. In FIGS. 40 and 41, the same operations as those in FIGS. 37 and 38 are denoted by the same reference numerals, and the description thereof will be omitted. Only different parts will be described.

The voice quality control unit 501 sets the fundamental frequency of the synthesized voice (step 12100). First, the auditory characteristic reading means 310 reads the user's auditory characteristic 320 set in advance and outputs it to the voice quality control means 501. (Step 12110). The voice quality control means 50 compares the average hearing level of the user with 40 dBHL (step 11120). If the average hearing level of the user is equal to or higher than 40 dBHL in step 11120, the speech speed of the synthesized voice is set to be 10% slower than a predetermined standard value (step 11130). Is entered (step 1200). If the average hearing level of the user is less than 40 dBHL in step 11120, the text input means 10 inputs a target text to the language processing means 20 (step 1200). Next, the language processing unit 20 analyzes the syntax of the text input from the text input unit 10, generates reading information and prosody information, and outputs the information to the speech synthesis control unit 70m (step 1300). The unit database 80 outputs a synthesis unit to the unit connecting unit 90m according to the reading information input from the speech synthesis control unit 70m (step 9400). The unit connection unit 90m connects the synthesis units input from the unit database 80 in accordance with the prosodic information input from the voice synthesis control unit 70m and the control signal input from the voice quality control unit 50m to generate a synthesized voice (step). 9500), and outputs a synthesized voice through the electroacoustic transducer 60 (step 1900).

(Example 13)
Hereinafter, a thirteenth embodiment of the present invention will be described with reference to the drawings.

FIG. 42 is a configuration block diagram showing a thirteenth embodiment of the speech synthesizer of the present invention. FIG. 43 is a flowchart for explaining the operation of the thirteenth embodiment. 42, the same components or portions as those in FIG. 30 are denoted by the same reference numerals, and the description thereof will not be repeated. 30 except that the language processing unit 20 of FIG. 30 has been replaced with a language processing unit 900, the voice synthesis unit 30m has been replaced with a voice synthesis unit 30n, the voice quality control unit 50i has been deleted, and the microphone 110 has been connected to the voice synthesis control unit 70n. This is the same configuration as 30. The language processing unit 900 includes a syntax analysis unit 910 and a speech synthesis start position determination unit 920. The speech synthesis unit 30n has a speech synthesis control unit 70n, a unit database 80, and a unit connection unit 90n.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 42 and 43. In FIG. 43, the same operations as those in FIG. 31 are denoted by the same reference numerals, and description thereof will be omitted. Only different parts will be described.

First, the text input means 10 inputs a target text to the syntax analysis means 910 (step 13100). Next, the syntactic analysis means 910 analyzes the text input from the text input means 10, generates syntactic information, outputs the syntactic information to the speech synthesis start position determining means 920, generates reading information and prosody information, and generates speech information. 70n (step 13200). Speech synthesis start position determination means 920 determines a speech synthesis start position according to the syntax information input from syntax analysis means 910, and outputs start position information to speech synthesis control means 70n (step 13300). The voice synthesis control unit 70n takes in the environmental sound signal from the microphone 110, and compares the average level of the environmental sound for 100 ms with 70 dB (A) (step 13400). If the average level of the environmental sound is less than 70 dB (A) in step 13400, the segment database 80 outputs a synthesis unit to the segment connection unit 90n according to the reading information input from the speech synthesis control unit 70n (step 9400). If the average level of the environmental sound is equal to or more than 70 dB (A) in step 13400, the speech synthesis control unit 70n outputs a speech synthesis stop signal to the unit connection unit 90n, and stops generation of the synthesized speech (step 13500). . The voice synthesis control unit 70n compares the average level of the environmental sound with 70 dB (A) (step 13600), and repeats step 13600 if the average level of the environmental sound is 70 dB (A) or more. Only when the average level of the environmental sound is less than 70 dB (A) in step 13600, the position on the text before the stop position is the closest to the stop position in accordance with the start position information input from the voice synthesis start position determining means 920. Speech synthesis is restarted from the near speech synthesis start position (step 13700), and the segment database 80 outputs a synthesis unit to the segment connection unit 90n according to the reading information input from the speech synthesis control unit 70n (step 9400). The unit connection unit 90n connects the synthesis units input from the unit database 80 according to the prosody information input from the voice synthesis control unit 70n to generate a synthesized voice (step 9500). Is output (step 1900).

(Example 14)
Hereinafter, a fourteenth embodiment of the present invention will be described with reference to the drawings.

FIG. 44 is a configuration block diagram showing a fourteenth embodiment of the speech synthesizer of the present invention. FIG. 45 is a flowchart for explaining the operation of the fourteenth embodiment. 44, the same components or portions as those in FIG. 42 are denoted by the same reference numerals, and the description thereof will not be repeated. Only different portions will be described. 42 has the same configuration as that of FIG. 42 except that the voice synthesis unit 30n is replaced by the voice synthesis unit 30o and the microphone 110 is replaced by the operation unit 40o. The speech synthesis unit 30o has a speech synthesis control unit 70o, a unit database 80, and a unit connection unit 90n.

The operation of the speech synthesizer according to the present embodiment configured as described above will be described below with reference to FIGS. 44 and 45. In FIG. 45, the same operations as those in FIG. 43 are denoted by the same reference numerals, and the description thereof will be omitted. Only different parts will be described.

First, the text input means 10 inputs a target text to the syntax analysis means 910 (step 13100). Next, the syntactic analysis means 910 analyzes the syntax of the text input from the text input means 10, generates syntax information and outputs it to the speech synthesis start position determination means 920, generates reading information and prosody information, and generates speech information. 70o (Step 13200). Speech synthesis start position determination means 920 determines a speech synthesis start position according to the syntax information input from syntax analysis means 910, and outputs start position information to speech synthesis control means 70o (step 13300). The voice synthesis control unit 70o receives the operation signal from the operation unit 40o, and determines whether or not the user has input a voice synthesis stop signal (step 14400). If the speech synthesis stop signal has not been input in step 14400, the segment database 80 outputs a synthesis unit to the segment connection unit 90n according to the reading information input from the speech synthesis control unit 70n (step 9400). If the speech synthesis stop signal is input in step 14400, the speech synthesis control unit 70o outputs the speech synthesis stop signal to the unit connection unit 90n, and stops generating the synthesized speech (step 13500). The voice synthesis control means 70o fetches an operation signal from the operation device and determines whether or not the user has input a voice synthesis restart signal (step 14600). If the voice synthesis restart signal has not been input, step 14600 is performed. repeat. Only when the speech synthesis restart signal is input in step 14600, the speech synthesis start position whose text position is earlier than the stop position and is closest to the stop position in accordance with the start position information input from the speech synthesis start position determination means 920. , The speech synthesis is restarted (step 13700), and the segment database 80 outputs the synthesis unit to the segment connection unit 90n according to the reading information input from the speech synthesis control unit 70o (step 9400). The unit connection unit 90n connects the synthesis units input from the unit database 80 in accordance with the prosody information input from the voice synthesis control unit 70o to generate a synthesized voice (step 9500). Is output (step 1900).

(Example 15)
Hereinafter, a fifteenth embodiment of the present invention will be described with reference to the drawings.

FIG. 46 is a configuration block diagram showing a fifteenth embodiment of the speech synthesizer of the present invention. FIG. 47 is a flowchart for explaining the operation of the fifteenth embodiment. In FIG. 46, the same components or portions as those in FIG. 42 are denoted by the same reference numerals, and the description thereof will be omitted. Only different portions will be described. The language processing unit 900n in FIG. 42 is replaced with a language processing unit 900p. The language processing unit 900p further includes an emphasized word determination unit that receives a syntax analysis result from the syntax analysis unit 910 and determines a word to be emphasized. On the other hand, the voice synthesizer 30n in FIG. 42 is replaced with a voice synthesizer 30p, and the voice synthesizer 30p receives a clock unit 940 connected to the voice synthesis controller 70p and a unit output of a unit database, and performs voice synthesis. A phoneme emphasis processing unit 130p that performs emphasis processing on a segment based on a control signal from the control unit and outputs the result to the segment connection unit is added. Further, the speech synthesis control unit receives the emphasized word information from the emphasized word determination unit 930. Except for the above changes, the configuration is the same as that of FIG.

The operation of the speech synthesizing apparatus according to this embodiment configured as described above will be described below with reference to FIGS. 46 and 47. In FIG. 47, the same operations as those in FIG. 43 are denoted by the same reference numerals, and description thereof will be omitted. Only different parts will be described. First, the text input means 10 inputs a target text to the syntax analysis means 910 (step 13100). Next, the syntactic analysis means 910 analyzes the syntax of the text input from the text input means 10, generates syntactic information, outputs the syntactic information to the speech synthesis start position determining means 920 and the emphasized word determining means 930, and reads the reading information and the prosody. Information is generated and output to the voice synthesis control means 70p (step 13200). Speech synthesis start position determination means 920 determines a speech synthesis start position according to the syntax information input from syntax analysis means 910, and outputs start position information to speech synthesis control means 70p (step 13300a). At the same time, the emphasized word determination means 930 determines a word to be emphasized according to the syntax information input from the syntax analysis means 910, and outputs the emphasized word information to the speech synthesis control means 70p (step 13300b). The voice synthesis control unit 70p takes in the environmental sound signal from the microphone 110 and compares the average level of the environmental sound for 100 ms with 70 dB (A) (step 13400). If the average level of the environmental sound is less than 70 dB (A) in step 13400, the segment database 80 outputs the synthesis unit to the phoneme emphasis processing means 130p according to the reading information input from the speech synthesis control means 70p, and outputs the synthesis unit. At 130p, the emphasis process is not performed, and the synthesis unit is directly output to the segment connection means 90n (step 9400a). If the average level of the environmental sound is equal to or more than 70 dB (A) in step 13400, the speech synthesis control unit 70n outputs a speech synthesis stop signal to the unit connection unit 90n, and stops generation of the synthesized speech (step 13500). . Then, a signal to start the measurement is sent to the timer 940 to start the time measurement (step 14100). The voice synthesis control unit 70p compares the average level of the environmental sound with 70 dB (A) (step 13600), and repeats step 13600 if the average level of the environmental sound is 70 dB (A) or more. If the average level of the environmental sound is less than 70 dB (A) in step 13600, a measurement end signal is sent to the timer 940 to end the time measurement and capture the elapsed time (step 14200). The voice synthesis control unit 70p sets the rank of the voice synthesis start position to 1 when the elapsed time is less than 1 second from 0, and sets the rank of the voice synthesis start position to 2 when the elapsed time is 1 second or more and less than 2 seconds. If the elapsed time is 2 seconds or more and less than 3 seconds, the rank of the speech synthesis start position is set to 3, and if the elapsed time is 3 seconds or more, the rank of the speech synthesis start position is set to 4 (step). 14300). The voice synthesis control means restarts voice synthesis from a voice synthesis start position closest to the stop position before the voice synthesis stop position and having a rank equal to or greater than the rank value determined in step 14300. If no speech synthesis start position having a rank equal to or greater than the rank value determined in step 14300 is found, speech synthesis is restarted from the beginning of the sentence (step 14400). Further, the number of repetitions of the start position where speech synthesis is restarted is increased by one (step 14500). The segment database 80 outputs a synthesis unit to the phoneme enhancement processing unit 130p according to the reading information input from the speech synthesis control unit 70p (step 14600). The speech synthesis control unit 70p determines whether the number of repetitions of the start position at which speech synthesis is restarted is two or more (step 14700). If the number of repetitions is 2 or more in step 14700, an emphasis control signal is output to the phoneme emphasis processing means 130p in the section from the start position to the stop position, and the phoneme emphasis processing means 130p performs emphasis processing for each synthesis unit (step 14800). . If the number of repetitions is less than 2 in step 14700, the speech synthesis control unit 70p does not output the emphasis control signal, and the phoneme emphasis processing unit 130p does not perform the segment emphasis process. According to the prosody information input from the speech synthesis control unit 70p, the unit connection unit 90p connects the synthesis units input from the unit enhancement processing unit to generate synthesized speech (step 9500). A synthesized voice is output (step 1900).

(Example 16)
Hereinafter, a sixteenth embodiment of the present invention will be described with reference to the drawings.

FIG. 48 shows a configuration diagram of the language processing unit of the speech synthesizer according to one embodiment of the present invention. The syntax analyzer 101 performs a morphological analysis and a syntax analysis on the input sentence, and outputs a syntax analysis result including a word string, a clause string, and a dependency structure between the clauses constituting the input sentence. The speech synthesis start position rule holding unit 103 holds rules describing the clauses before and after the speech synthesis start position to be set in the speech synthesis start position determination unit 102 and the condition of the dependency structure between the clauses. FIG. 49 is a diagram illustrating an example of the speech synthesis start position rule held by the speech synthesis start position rule unit 103. The speech synthesis start position determination unit 102 sets the speech synthesis start position between the phrase strings of the syntax analysis result. In FIG. 49, the previous phrase pattern specifies the condition of the phrase located immediately before the speech synthesis start position. Similarly, the later phrase pattern specifies the condition of a phrase located immediately after the speech synthesis start position. When the format of each phrase pattern is expressed in BNF notation,
<Clause pattern>: = * | (<clause name><morphemesequence>)
<Clause name>: = noun phrase | predicate phrase | adverb phrase | ...
<Morpheme sequence>: = * | (<morpheme>) | (<morpheme><morphemesequence>)
<Morpheme>: = * | + | (<part of speech><notation>)
<Part of speech>: = noun | particle | reading |
<Notation>: = * | is |, |, | “*” Represents an arbitrary phrase, an arbitrary morpheme string, an arbitrary morpheme, or an arbitrary notation. “+” Represents an arbitrary arrangement of morphemes. The rank is a value assigned to the corresponding speech synthesis start position, and the control unit 106 selects the speech synthesis start position based on this value. In this embodiment, the rank value increases as the length of the pause inserted at the speech synthesis start position when speech synthesis of the input text is performed. The first speech synthesis start position rule in FIG. 49 means that a speech synthesis start position of rank 3 is set between a noun phrase ending with the particle "ha" and an arbitrary phrase. The speech synthesis start position determination unit 102 compares the syntax analysis result output by the syntax analysis unit 101 with the speech synthesis start position rule held in the speech synthesis start position rule holding unit 103 and the phrase string included in the syntax analysis result. And sets a speech synthesis start position and a rank at a place where the matching is successful. FIG. 50 is a diagram illustrating a process of the speech synthesis start position determination unit. The input text is processed by the syntax analysis unit 101 to generate a phrase string as shown in FIG. For this phrase string, the speech synthesis start position determination unit 102 sequentially compares the speech synthesis start position rules for the two phrases from the beginning of the phrase string, and the rule is described in the rule between the two successfully matched phrases. Set speech synthesis start position with rank. In the example of FIG. 50, the second rule of FIG. 49 is between the first two clauses, the third rule of FIG. 49 is between the second two clauses, and the fourth rule of FIG. 49 is between the third two clauses. Are compared, and a speech synthesis start position having a rank as shown at the bottom of FIG. 50 is set. No speech synthesis start position is set between two phrases that did not match any of the speech synthesis start position rules.

The emphasized word determination unit 104 compares the syntax analysis result output by the syntax analysis unit 101 with the emphasized word rule held in the emphasized word rule holding unit 105 and a word string included in the syntax analysis result, and emphasizes. To determine the words to pronounce. FIG. 51 is a diagram illustrating an example of the rule of the emphasized word held by the emphasized word holding unit 105. In FIG. 51, the emphasized word condition describes a condition of a word to be emphasized. Expressing the form of the emphasized word in BNF notation,
<Emphasis condition>: = (<part of speech><notation>)
<Part of speech>: = noun | verb | adjective |
<Notation>: = * | “*” Is a symbol representing an arbitrary notation. Emphasis information is assigned to a word that satisfies the emphasis word condition in accordance with the description of emphasis ON / OFF in the right-hand column. FIG. 52 is a diagram illustrating the processing of the emphasized word determination unit. In FIG. 52, the syntax analysis unit 101 processes an input text to generate a word string. The emphasized word determination unit 102 checks the emphasized word rule in order from the beginning of the word string, and if the matching is successful, adds emphasis ON / OFF information. For the word (no adjective) in FIG. 52, both the emphasized word condition (no adjective *) and the emphasized word condition (no adjective) are matched, but the emphasized word condition (no adjective) is more detailed than the notation is specified. This is a condition that is prioritized in matching. As a result of the comparison, the emphasized word information as shown at the bottom of FIG. 52 is obtained. Emphasis on words that did not match any of the emphasis word rules is OFF.

(Example 17)
Hereinafter, a seventeenth embodiment of the present invention will be described with reference to the drawings, taking an example of an emphasis process at the time of speech unit creation.

FIG. 53 is a flowchart showing the operation of speech unit creation according to the seventeenth embodiment, and FIG. 54 is a schematic diagram of the input / output characteristics of the amplitude compression processing.

First, the first waveform is cut out from the target voice waveform (step 15000). Next, the cut-out waveform data cut out in step 15000 is multiplied by a preset gain value G (step 15010), the maximum value of the absolute value of the result is obtained and stored in Amax (step 15020). If Amax is larger than the preset value of Alim (step 15030), the cutout waveform is multiplied by (Alim / Amax) (step 15040). If Amax is smaller than or equal to Alim, nothing is performed. If the waveform cut out this time is the last waveform (step 15050), the processing ends. Otherwise, the next waveform is cut out (step 15060), and the process is repeated from step 15010.

よ う By doing so, there is no problem of a time constant that occurs when a limiter is used for a sound waveform, and ideal amplitude compression can be performed. If the amplitude compression processing shown in FIG. 53 is compared to the input / output characteristics of the limiter, it can be expressed as shown in FIG. Since this curve can be arbitrarily selected as shown in, for example, FIGS. 54B and 54C, various amplitude compression processes can be performed. Further, amplitude compression for each phoneme is also possible, such as selecting a curve according to the type of the target speech unit (unvoiced consonant, voiced consonant, etc.). Further, by giving labels to the start point and end point of the consonant part in advance, the consonant part and the vowel part can be compressed with different curves.

る こ と Since various amplitude compression methods can be selected as described above, it is effective as a phoneme emphasis method for emphasizing a specific part of a specific consonant. In other words, such processing at the time of speech unit creation has a very high degree of freedom as a voice emphasis method, and allows detailed processing. In addition, since such processing is executed completely as preprocessing, there is an advantage that the processing speed during speech synthesis is not affected at all.

Therefore, it is possible to perform any complicated voice emphasis processing. Therefore, by emphasizing the frequency domain such as formant emphasis, dividing the audio waveform to be cut out into multiple bands and applying amplitude compression, etc. It is possible to provide a synthesized speech suitable for use in a personal computer. In particular, it is possible to perform a fine emphasizing process which cannot be performed at the time of extracting a waveform by adding a process corresponding to a signal process of a hearing aid, which has been impossible due to a time constant or a limit of a process for an unknown input.

In the seventeenth embodiment, as the processing on the speech unit, the processing of deforming the amplitude and the processing of changing the frequency characteristic mainly for the purpose of emphasizing the consonants have been described. Various waveform deformation processes may be performed, such as by improving the clarity by adjusting the depth.

In the fifteenth embodiment, the microphone 110 captures the environmental sound signal. However, the microphone 110 may capture the utterance of the user.

In the fifteenth embodiment, the segment emphasizing process is performed. However, the segment database that has undergone the emphasizing process is switched between the segment database that has not undergone the emphasizing process, or the emphasis process is performed on the synthesized speech after the segment connection. It is good.

In the first and eighth embodiments, the emphasis processing is performed for the extension of the vowel part, the extension of the closure, the formant emphasis, the consonant emphasis, and the band emphasis, but other emphasis methods may be used.

In the first and eighth embodiments, the formant emphasis information is set to true when p is smaller than 15, but may be set to any other value.

In the first and eighth embodiments, the consonant emphasis information is set to true when the gap detection area is equal to or longer than 10 ms, but may be set to any other value.

In the first and eighth embodiments, the band emphasis information is determined to be true when the difference between the average hearing level of 2 kHz or more and the average hearing level of less than 2 kHz is 30 dB or more. Is also good. The reference of the difference in the average hearing level between the bands may be a value other than 30 dB.

In the first, second, fourth, and eighth embodiments, the steady vowel part is extended by 20% when the emphasis part information is true, but other values may be used. Further, the time length of the consonant part may be extended.

In the first and second embodiments, the closure is extended by 20% when the emphasis part information is true, but other values may be used.

In the first, fourth, and seventh embodiments, the environmental sound is divided into bands of 1 kHz or less, 1 kHz to 2 kHz, 2 kHz to 4 kHz, and 4 kHz or more, but other division methods may be used.

In the first embodiment, in each band of 1 kHz or less, 1 kHz to 2 kHz, 2 kHz to 4 kHz, and 4 kHz or more, when there is an environment of 20 dBSPL / Hz, 20 dBSPL / Hz, 15 dBSPL / Hz, and 10 dBSPL / Hz or more, the compression parameter is set. Although the setting is made and the compression process is performed, other values may be used.

In the fourth embodiment, the average environmental sound level of 1 kHz or less is 20 dBSPL / Hz or more, the average environmental sound level of 1 kHz to 2 kHz is 20 dBSPL / Hz, and the average environmental sound level of another band is 15 dBSPL / Hz or less. Although the formant emphasis information is set to true, other values may be used.

In the fourth embodiment, the average environmental sound level from 1 kHz to 2 kHz is 20 dBSPL / Hz or higher, the average environmental sound level from 2 kHz to 4 kHz is 15 dBSPL / Hz, and the average environmental sound level below 1 kHz is 20 dBSPL / Hz or lower or 4 kHz or higher. The consonant emphasis information is determined to be true when the average environmental sound level is 15 DBSPL / Hz or less, but may be set to any other value.

In the fourth embodiment, in each band of 1 kHz or less, 1 kHz to 2 kHz, 2 kHz to 4 kHz and 4 kHz or more, when there is an environment of 20 dBSPL / Hz, 20 dBSPL / Hz, 15 dBSPL / Hz, and 10 dBSPL / Hz or more, each band is emphasized. Although the information is assumed to be true, other values may be used.

In the seventh embodiment, compression parameters are set as in step 7500, but other criteria and methods may be used.

In the ninth embodiment, the basic frequency is increased by 20% when the average level of the environmental sound is equal to or higher than 30 dB (A). However, other reference values may be used. The change of the fundamental frequency may be any other value.

In the tenth embodiment, when the difference between the average hearing level of 2 kHz or more and the average hearing level of less than 2 kHz is 30 dB or more, the fundamental frequency is lowered by 20%. However, a frequency other than 2 kHz may be used as a boundary of the band. Further, the reference of the difference value may be another value. Further, the change of the fundamental frequency may be any other value.

In the eleventh and twelfth embodiments, it is assumed that when the average hearing level is 40 dBHL or more, the instantaneousness is delayed by 10%. However, auditory characteristics other than the average hearing level may be used for the determination. The standard of the average hearing level may be any other value. Although the standard of the average hearing level is set to 40 dBHL, other values may be used. In addition, it is assumed that the speech speed is reduced by 10%, but other values may be used.

In the thirteenth and fifteenth embodiments, the speech synthesis is stopped when the average level of the environmental sound exceeds 70 dB (A), but other values may be used. In the second, fifth, and eighth embodiments, the switch is used to switch the segment database or the formant emphasis filter. However, the switch may be switched by software.

As described above, according to the present embodiment, the speech synthesized according to the auditory characteristics of the user is subjected to the emphasis processing or the processing of compressing the dynamic range of the amplitude, or synthesized according to the noise environment of the use scene. The voice is subjected to an emphasis process or a process of compressing the dynamic range of the amplitude. In addition, the synthesis unit stored in the database is subjected to an emphasis process or a process of compressing the dynamic range of the amplitude in accordance with the auditory characteristics of the user, and then synthesized, or stored in the database in accordance with the noise environment of the use scene. After performing the emphasis processing or the processing of compressing the dynamic range of the amplitude on the synthesized unit, the voice is synthesized. Also, speech is synthesized using a synthesis unit that has been subjected to an emphasis process or a process of compressing the dynamic range of the amplitude in advance. When the speech synthesis is interrupted, the speech synthesis is resumed by returning to a point on the text where the content is easy to understand before the stop position based on the result of the language processing. In addition, by setting a portion for performing the emphasis processing based on the language processing, information can be reliably transmitted even to a hearing impaired user or use under noise, and the practical effect is large.

As described above, according to the present invention, information can be reliably transmitted even to a hearing impaired user or use under noisy conditions, and its practical effect is large.

1 is a block diagram showing the configuration of a first embodiment of a speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Schematic diagram of the formant enhancement method of the embodiment Schematic diagram of the consonant emphasis method of the embodiment 2 is a block diagram showing the configuration of a second embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment 3 is a block diagram showing the configuration of a third embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment 4 is a block diagram showing the configuration of a fourth embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment 5 is a block diagram showing the configuration of a fifth embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment Configuration block diagram of a sixth embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment 7 is a block diagram showing the configuration of a seventh embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Configuration block diagram of an eighth embodiment of the speech synthesizer according to the present invention Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Schematic diagram of the formant enhancement method of the embodiment 9 is a block diagram showing the configuration of a ninth embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Configuration block diagram of a tenth embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment Configuration block diagram of an eleventh embodiment of the speech synthesizer according to the present invention Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment 12 is a block diagram showing the configuration of a twelfth embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment Flow chart for explaining the operation of the embodiment 13 is a block diagram showing the configuration of a thirteenth embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment 14 is a block diagram showing the configuration of a fourteenth embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment Configuration block diagram of a fifteenth embodiment of the speech synthesizer according to the present invention. Flow chart for explaining the operation of the embodiment 16 is a block diagram showing the configuration of a sixteenth embodiment of the speech synthesizer according to the present invention. Schematic diagram of the rank determination method of the embodiment Schematic diagram for explaining the operation of the embodiment Schematic diagram of the highlighted portion selection method of the embodiment Schematic diagram for explaining the operation of the embodiment 17 is a flowchart for explaining the operation of the seventeenth embodiment of the speech segment generating method according to the present invention. Schematic diagram of the amplitude compression processing method of the embodiment Configuration block diagram of conventional speech synthesizer Schematic diagram of the method for extracting waveforms in the conventional speech unit creation method Schematic diagram of the connection method of the segments of the conventional speech synthesizer

Explanation of reference numerals

Reference Signs List 10 Text input means 20 Language processing means 30a, 30b, 30c, 30d, 30e, 30f, 30g, 30h, 30m, 30n, 30o, 30p Voice synthesizer 40b, 40e, 40m, 40o Operating means 50a, 50b, 50c, 50d , 50e, 50g, 50h, 50i, 50k, 50l, 50m Voice quality control means 60 Electroacoustic transducers 70a, 70b, 70c, 70d, 70e, 70f, 70g, 70h, 70m, 70n, 70o, 70p Voice synthesis control Means 80, 280a, 280b, 280c, 280d, 280n, 380a, 380b, 380c, 380d, 380n, 580a, 580b, 580c, 580d, 580n, 680a, 680b, 680c, 680d, 680n Unit database 90a, 90b, 9 d, 90e, 90f, 90g, 90h, 90m, 90n Unit connection means 110 Microphone 120 Hearing characteristic measurement means 130a, 130d, 130h, 130p Phoneme emphasis processing means 140a, 140b, 140g Compression processing means 200b, 200e Database section 210b, 210e switch 220 auditory characteristic storage means 300 unit database reading means 310 auditory characteristic reading means 320a, 320b, 320c, 320d, 320n auditory characteristic 800 emphasis filter section 810a, 810b, 810c, 810d, 810n formant emphasis filter 820 switch 900, 900p Language processing section 910 Syntax analysis section 920 Speech synthesis start position determination section 930 Emphasis determination section 940 Clock section 101 Syntax analysis section 1 2 speech synthesis start position determination unit 103 speech synthesis start position rule holding unit 104 intensifiers determining unit 105 intensifiers rule holding unit 106 control unit

Claims (35)

A speech synthesizer comprising: a speech synthesis unit that synthesizes speech in accordance with a text; and an emphasis processing unit that performs single or multiple phoneme emphasis processes on the speech synthesized by the speech synthesis unit. 2. The speech synthesizer according to claim 1, wherein the emphasis processing is a consonant emphasis processing for performing amplitude emphasis processing of a consonant or a consonant and a vowel following the consonant based on the phoneme information. 2. The speech synthesizer according to claim 1, wherein the emphasis process is a band emphasis process for emphasizing a consonant frequency band based on phoneme information. 2. The speech synthesizer according to claim 1, further comprising: a control unit configured to analyze the environmental sound input from the microphone and control the emphasis processing unit based on the physical characteristics of the environmental sound. The speech synthesizer according to claim 4, wherein the control unit analyzes the environmental sound input from the microphone and selects an emphasis processing method used in the emphasis processing unit based on the physical characteristics of the environmental sound. 2. The speech synthesizing apparatus according to claim 1, further comprising an operation unit for allowing a user to adjust a processing method and a degree of emphasis, and a control unit for controlling the emphasis processing unit based on a signal input from the operation unit. The speech synthesizer according to claim 1, further comprising: a measurement unit configured to measure a user's hearing characteristics; and a control unit configured to control an emphasis processing unit based on the user's hearing characteristics. The speech synthesizer according to claim 7, wherein the control unit selects an emphasis processing method used in the emphasis processing unit based on the user's auditory characteristics input from the measurement unit. The voice synthesizing apparatus according to claim 1, further comprising: a storage unit configured to store a user's hearing characteristics; and a control unit configured to control an emphasis processing unit based on the user's hearing characteristics. The speech synthesizer according to claim 9, wherein the control unit selects an emphasis processing method used in the emphasis processing unit based on the user's auditory characteristics stored in the storage unit. 2. A speech synthesizer according to claim 1, further comprising: a hearing characteristic reading unit; and a control unit, wherein the control unit controls the emphasis processing unit by referring to a user's hearing characteristic stored in a recording medium by the hearing characteristic reading unit. apparatus. The speech synthesizer according to claim 11, wherein the control unit selects an emphasis processing method to be used by the emphasis processing unit based on the user's auditory characteristics read by the auditory characteristic reading unit. A speech unit database that stores speech in desired synthesis units such as vowel / consonant / vowel combinations, a unit deformation unit that emphasizes the synthesis unit, and an emphasis process performed by the unit deformation unit. And a speech synthesizer for connecting the synthesized units by a target text to synthesize speech. 14. The speech synthesizer according to claim 13, wherein the emphasis processing is a consonant emphasis processing for performing amplitude emphasis processing of a consonant or a consonant and a subsequent vowel based on the phonemic information. 14. The speech synthesizer according to claim 13, wherein the emphasis process is a band emphasis process for emphasizing a consonant frequency band based on phoneme information. 14. The speech synthesizer according to claim 13, wherein the emphasis processing is a closure emphasis processing for extending a consonant closure based on linguistic information. 14. The speech synthesizer according to claim 13, wherein the emphasis processing is an extension processing for extending a phoneme length based on linguistic information. 14. The speech synthesizer according to claim 13, further comprising: a control unit configured to analyze an environmental sound input from the microphone and control the unit deformation unit based on physical characteristics of the environmental sound. 19. The speech synthesizer according to claim 18, wherein the control unit analyzes the environmental sound input from the microphone and selects an emphasis processing method used in the unit deformation unit based on physical characteristics of the environmental sound. 14. The speech synthesizing apparatus according to claim 13, further comprising: an operation unit for allowing a user to adjust a processing method and a degree of the emphasis; . 14. The speech synthesizer according to claim 13, further comprising: a measuring unit configured to measure a hearing characteristic of the user; and a control unit configured to control a unit deforming unit based on the hearing characteristic of the user. 22. The speech synthesizer according to claim 21, wherein the control unit selects an emphasis processing method used in the unit deformation unit based on the user's auditory characteristics input from the measurement unit. 14. The voice synthesizing device according to claim 13, further comprising: storage means for storing a user's hearing characteristics; and a control unit for controlling a segment deformation unit based on the user's hearing characteristics. 24. The speech synthesizer according to claim 23, wherein the control unit selects an emphasis processing method to be used in the unit deformation unit based on the user's auditory characteristics stored in the storage unit. 14. The voice according to claim 13, further comprising: a hearing characteristic reading unit; and a control unit, wherein the control unit controls the segment deformation unit with reference to a user's hearing characteristic stored in a recording medium by the hearing characteristic reading unit. Synthesizer. 26. The speech synthesizer according to claim 25, wherein the control unit selects an emphasis processing method used in the unit deformation unit based on the user's auditory characteristics read by the auditory characteristic reading unit. A speech unit database for storing speech subjected to phoneme emphasis processing in a desired synthesis unit such as a vowel / consonant / vowel chain, and speech synthesis for synthesizing speech by connecting the synthesis unit with a target text And a speech synthesizer comprising: A plurality of speech unit databases having different emphasis methods and degrees and an environment sound input from a microphone are analyzed, and the speech unit database used by the speech synthesis unit for speech synthesis is selected based on the physical characteristics of the environment sound. 28. The speech synthesizer according to claim 27, further comprising: a control unit that performs the control. A plurality of speech unit databases having different emphasis methods and degrees, operation means for a user to adjust the state of emphasis, and a speech synthesis unit used for speech synthesis based on a signal input from the operation means. The speech synthesizer according to claim 27, further comprising: a control unit that selects the speech unit database. A plurality of speech unit databases having different emphasis methods and degrees; a measuring unit for measuring a user's hearing characteristics; and the speech unit database used by a speech synthesis unit for speech synthesis based on the user's hearing characteristics. 28. The voice synthesizing apparatus according to claim 27, further comprising: a control unit configured to select a voice. A plurality of speech unit databases having different emphasis methods and degrees; storage means for storing user's hearing characteristics; and the speech unit database used by the speech synthesis unit for speech synthesis based on the user's hearing characteristics. 28. The voice synthesizing apparatus according to claim 27, further comprising: a control unit configured to select a voice. 28. The speech synthesis apparatus according to claim 27, further comprising: a storage medium storing a plurality of speech unit databases having different emphasis methods and degrees; and a speech unit database reading unit. A speech synthesis method in which a speech synthesized by a speech synthesis unit that synthesizes speech in accordance with a text is subjected to single or plural phoneme emphasis processes. A speech synthesis unit output from a speech unit database storing speech in a desired synthesis unit such as a vowel / consonant / vowel combination is subjected to emphasis processing, and the synthesized unit subjected to the emphasis processing is used for Speech synthesis method that synthesizes speech by connecting with text. The speech synthesis unit output from the speech unit database which stores speech subjected to phoneme emphasis processing in advance in a desired synthesis unit such as a vowel / consonant / vowel combination is connected by a target text to generate a speech. The speech synthesis method to synthesize.

Images