JP3595041B2 - Speech synthesis system and speech synthesis method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate such output of a system that makes a user easily grasp the user's situation and the condition of the system and to improve the operation convenience by changing an applicable phoneme/rhythm control regulation in accordance with the user's situation, user's environment and the condition of the system. SOLUTION: The communication state in a computer in which a voice synthesizing system is operated and a communication state between the computer and the outside are monitored by a communication state monitoring part 12 and corresponding information of the communication state is outputted. In a language analysis part 111 in a voice synthesizing part 11, by performing language analysis such as morpheme analysis and syntactic structural analysis of an input text and applying regulations of various levels to the contents of the result of language analysis in a phoneme control part 112 and a rhythm control part 113, phoneme/rhythm control influencing the quality of the synthesized voice accompanied by conversion from language media to voice media is performed. At this time, the regulation applied to the phoneme/rhythm control is changed according to the information of the communication state.

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a speech synthesis system and a speech synthesis method that are preferably used in general for systems that use speech synthesis technology.
[0002]
[Prior art]
In recent years, as the application of speech synthesis technology has expanded and the processing power of computers has improved, the use of speech synthesis has increased more and more, and text-to-speech conversion has become easily available as an application. Was. Iwata et al .: "Software Japanese text-to-speech synthesis for personal computers," Proc. 245-246 (October 1993) is an example.
[0003]
These speech synthesis systems are designed to obtain a unique processing result in principle in any of the language analysis, phonological control, prosodic control, and waveform generation of the input text. The resulting synthesized speech was always the same.
[0004]
Some systems allow the user to select a synthetic voice, such as a male / female voice, a high / low voice, etc., before generating the synthesized voice, and generate a synthesized voice according to the selection. It can be said that the synthesized speech is the same framework in the sense that it is uniquely determined.
[0005]
[Problems to be solved by the invention]
However, generating the same synthesized speech for the input text is not only monotonous and tired, but also does not necessarily mean that the features of the speech media are effectively used.
[0006]
Speech media differs from text media in that prosody and timbre change add additional information, such as emotions and intentions, the situation of the speaker, and the mood of the speaker, in addition to direct messages in linguistic expressions. It is well known that there is a characteristic that conveys information. Fujisaki et al .: "Expression of utterance intention using prosodic features of speech," Proc. 225-226 (March, 1993), Uesoko, et al .: "Analysis and Modeling of Emotional Expression of Voice," IEICE Technical Report, SP92-131, pp. 139-143. 65-72 (January 1993) is an example.
[0007]
All of the above-described current speech synthesis systems and their application applications have a function as a media conversion (text-to-speech conversion) that simply converts linguistic information expressed in the form of character language media into an expression in the form of voice media. However, it is hard to say that they actively use the voice feature of conveying secondary information.
[0008]
Furthermore, when speech synthesis operates not as a single device but in cooperation with another system, or as one application application on a general-purpose computer such as a personal computer or a workstation together with other application applications Even when used, the given text is simply copied regardless of the status of the system or application running together or the status of the system running the speech synthesis application. It just converts it to sound.
[0009]
The present invention has been made in consideration of the above circumstances, and its object is to provide a phoneme according to a system operation status (system status), a user's own status (user status), or an environment where the user is located (user environment).・ A system that allows the user to easily understand the operation status of the system by dynamically changing the prosody control, or by outputting another sound or synthesized sound in addition to the original synthesized sound corresponding to the input text An object of the present invention is to provide a speech synthesis system and a speech synthesis method capable of generating an output and an output of a system suitable for a situation where a user is placed, and improving usability.
[0010]
[Means for Solving the Problems]
In order to solve the above-described problem, a configuration according to a first aspect of the present invention performs linguistic analysis of an input text, applies rules to the analysis result, performs phonological / prosodic control, and generates synthesized speech. A communication state monitoring means for monitoring at least one of a communication state inside the computer and a communication state between the computer and the outside and outputting communication state information, in addition to the voice synthesis means for outputting; Preferably, a rule applied in the phonological / prosodic control is changed in accordance with communication state information output from the communication state monitoring means.
[0011]
The configuration according to the second aspect of the present invention includes at least one of an operation state of computer hardware and an operation state of computer software in addition to the speech synthesis unit corresponding to the speech synthesis unit in the configuration according to the first aspect. Operating state monitoring means for monitoring operation and outputting operating state information, wherein the voice synthesizing means applies rules applied in the phonological / prosodic control according to the operating state information output from the operating state monitoring means. Is changed. Here, at least one process of language analysis, phonological control, prosodic control, and voice waveform generation in the voice synthesis means may be performed according to an operation state indicated by the operation state information. It is preferable to share a plurality of computer hardware connected by a wireless network.
[0012]
The configuration according to the third aspect of the present invention further includes at least one of a user's system usage status and a user's system usage environment, in addition to the voice synthesis unit corresponding to the voice synthesis unit in the configuration according to the first aspect. The voice synthesizing means includes a user status monitoring means for monitoring and outputting user status information, wherein a rule applied in the phonological / prosodic control is determined according to the user status information output from the user status monitoring means. It is characterized in that it is changed. Here, based on the user situation information, a non-natural voice explicit determination means for outputting a determination result as to whether or not to indicate that the voice is not a human voice, and according to the determination result of the non-natural voice explicit determination means, Text changing means for specifying that the voice is not a human voice by partially changing the expression of the input text, and a non-natural output for outputting a sound specifying that the voice is not a human voice along with the output of the synthesized voice according to the determination result It is also possible to further comprise at least one of the voice explicit sound output means.
[0013]
In the configuration according to the first aspect, a well-known linguistic analysis such as a morphological analysis or a syntactic structure analysis is first performed on the input text in the speech synthesizing unit, and the input text is decomposed into a morpheme sequence and “read”. And information such as the part of speech of the morpheme, inflection, accent type, and the strength of the dependency relationship between the morphemes are output.
[0014]
Further, in the speech synthesis means, various levels of rules (phonological / prosodic rules) are applied to the contents of the results of the above-mentioned linguistic analysis, thereby synthesizing with the conversion from linguistic media to speech media. Control that affects the quality of voice, that is, phonemic / prosodic control, is performed, and a voice waveform corresponding to the input text is generated.
[0015]
On the other hand, the communication monitoring means monitors at least one of a communication state in the computer in which the speech synthesis system operates and a communication state between the computer and the outside. Each communication state monitoring result is transmitted to the voice synthesizing means.
[0016]
In the voice synthesizing means, the application contents of the phoneme / prosodic rules are changed according to the result of monitoring the communication state.
Next, also in the configuration according to the second aspect, the speech synthesis means applies various levels of phonological / prosodic rules to the contents of the results of the linguistic analysis of the input text, thereby achieving language Phonological / prosodic control that affects the quality of synthesized speech accompanying the conversion from media to audio media is performed, and an audio waveform corresponding to the input text is generated.
[0017]
On the other hand, the operation state monitoring means monitors at least one of the operation state of the computer hardware on which the speech synthesis system operates and the operation state of the computer software. Each operation state monitoring result is transmitted to the voice synthesizing means.
[0018]
In the voice synthesizing means, the application contents of the phonological / prosodic rules are changed according to the operation state monitoring result.
In addition, at least one process of language analysis, phonological control, prosodic control, and voice waveform generation in the voice synthesis means is connected to a plurality of communicable wired or wireless networks in accordance with the operation state monitoring result. Computer hardware.
[0019]
Next, in the configuration according to the third aspect as well, the speech synthesis means applies various levels of phonological / prosodic rules to the contents of the results of the linguistic analysis of the input text, thereby achieving language Phonological / prosodic control that affects the quality of synthesized speech accompanying the conversion from media to audio media is performed, and an audio waveform corresponding to the input text is generated.
[0020]
On the other hand, the user status monitoring means monitors at least one of the user's system usage status and the user's system usage environment. Each user situation monitoring result is transmitted to the voice synthesis means.
[0021]
In the voice synthesizing means, the application contents of the phonological / prosodic rules are changed according to the result of the user situation monitoring.
Further, in a configuration further including a non-natural voice explicit determination unit and at least one of a text change unit and a non-natural voice explicit sound output unit, a change in a part of the expression of the input text (for example, (Addition), a text change that clearly indicates that the voice is not a human voice, or a sound that clearly indicates that the voice is not a human voice is output together with the output of the synthesized voice. That is, another sound or synthesized sound is output in addition to the original synthesized sound.
[0022]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[First Embodiment]
FIG. 1 is a block diagram showing a first embodiment of the speech synthesis system of the present invention.
[0023]
The speech synthesis unit 11 which is the center of the system in FIG. 1 includes a language analysis unit 111 that performs a language analysis of an input text, a phoneme control unit 112 that performs a phonological control by applying a rule to the analysis result, and It is composed of a prosody control unit 113 that performs prosody control, a waveform generation unit 114 that generates a speech waveform under the control of the phoneme control unit 112 and the prosody control unit 113, and a waveform output unit 115 that outputs the generated waveform. ing. As for the framework of the configuration of the speech synthesis unit 11, a general configuration method of an existing speech synthesis system capable of text-to-speech conversion can be used as it is. As a general construction method of a text-to-speech synthesis system, for example, Sato et al .: "Speech synthesis from Japanese text," Research and Application Report of the Institute of Electro-Communications, Vol. 32, No. 11 pp. 2243-2252 (November 1983).
[0024]
The linguistic analysis unit 111 in the speech synthesis unit 11 performs linguistic analysis such as morphological analysis and syntactic structure analysis on the input text, decomposes the input text into a series of morphemes, and uses a symbol string representing “reading” and a part of speech of the morpheme. , Accent type, and the strength of the dependency relationship between morphemes are also output.
[0025]
The phoneme control unit 112 and the prosody control unit 113 in the speech synthesis unit 11 apply rules at various levels to the contents of the language analysis result of the language analysis unit 111, thereby converting the language media to the speech media. To control the quality of the synthesized speech accompanying the conversion to.
[0026]
Specifically, recognition of accent units (that is, including accenting of unknown words, accent combining, accent division of compound words, addition of secondary accents to accessory word chains, etc.), recognition of reading (that is, unknown words) (Including the addition of reading to words, rendaku processing, conversion from notation to reading, etc.), recognition of a unit of a so-called prosodic word (accent phrase) chain to form one prosodic unit (ie, The assignment corresponds to this), prosody control parameter value determination (that is, determination of parameter values of a model that generates a time-varying pattern of pitch, determination of phoneme / pause timing, determination of power), and accumulation pattern corresponding to reading Search and selection of storage units (that is, conversion to storage units, conversion to search conditions, selection when multiple search results are obtained, etc.) Contains), performs editing of the storage segment (i.e., the connection of the storage element pieces, interpolation processing such as the processing of each step such Included) is by the respective rules.
[0027]
Among them, the phoneme control unit 112 recognizes reading, searches for a storage pattern corresponding to the reading, selects a storage unit, and edits a storage unit. The prosody control unit 113 recognizes accent units, prosody words. (Accent phrase) Responsible for recognizing the unit of the chain and determining the prosody control parameter value. The processes and rules at each of these stages can be arbitrarily categorized, and may differ from the above categorization or may be omitted depending on the implementation of the system. Done. In addition, depending on the implementation of the system, the boundary between the language analysis unit 111 in the preceding stage and the waveform generation unit 114 in the subsequent stage is also various, but here, the phonological control unit 112 is assumed to perform the above phonological control. Similarly, a prosody control unit 113 is defined to perform the above-described prosody control.
[0028]
The prosody control unit 113 applies these rules to the result of the morphological analysis performed by the language analyzing unit 111, determines timings of individual phonemes and pauses corresponding to the reading of the morphological sequence, and determines the timing of the phoneme corresponding to the morphological sequence or the reading. Is divided into prosodic words (accent phrases), which are units of prosody control for adding accents, and forms a unit on the expiration paragraph due to semantic sentence structure and physiological constraints. A parameter value that divides the prosodic word sequence into units for prosodic control to add called tone components, and gives the size of accent and speech components to each prosodic control unit in consideration of timing. To determine the pitch. The prosody control unit 113 further determines a power envelope based on a morphological sequence, a phonemic sequence corresponding to reading, a pitch, or the like.
[0029]
On the other hand, the phoneme control unit 112 stores the speech waveform, the analysis parameter of the speech waveform, or the storage unit in which the both are associated with the partial sequence of the phoneme corresponding to the reading, and stores the stored data storage unit 1121. In consideration of the variation of the storage unit stored in the storage data storage unit 1121, the sequence of the storage unit corresponding to the phoneme partial sequence corresponding to the reading of the morphological sequence is determined.
[0030]
In the present embodiment, the rules applied by the phonological control unit 112 and the prosody control unit 113 are switched according to the communication state inside the computer and the communication state between outside the computer. Will be described later.
[0031]
The waveform generation unit 114 in the speech synthesis unit 11 connects the storage unit sequence output from the phoneme control unit 112 and controls the signal processing level according to the control information output from the prosody control unit 113, that is, the timing, pitch, and power envelope. Is performed to generate a speech waveform.
[0032]
The waveform output unit 115 in the voice synthesis unit 11 outputs the voice waveform generated by the voice synthesis unit 11 from, for example, a speaker, an earphone, or the like.
By the way, in the present embodiment, it is desired to generate a storage unit stored in the storage data storage unit 1121 of the phoneme control unit 112, a rule used by the phoneme control unit 112, and a rule used by the prosody control unit 113. Natural speech data that matches the tone of the synthesized speech is collected and created in advance from the data. For example, if you want to synthesize dialogue-like speech, you can collect simulated dialogue speech, whisper, fast-talking voice, tired sound, cheerful voice, or in a crowd (or in a crowded environment) The collected voices, calm voices, and various human voices are collected as much as possible, and rules and accumulated data corresponding to each voice are derived from the analysis results of each pitch, power, and time length.
[0033]
Previous studies have pointed out that speech in various situations has phonological and prosodic characteristics with different tendencies, and that prosodic control rules derived from various speech data show different tendencies. Is Hirai et al .: "F automatically generated from various speech corpora. ₀ The difference between the control rules is shown together with the actual data in “Acoustic Society of Japan, 2-5-3, pp. 271-272 (October to November 1994)”.
[0034]
There have been many studies on the derivation of rules from audio data. For example, Hirose et al .: "Speech Synthesis and Accent Intonation," IEICE Journal, Vol. 70, No. 4, pp. 378-385 (April 1987), Mimura et al .: "Analysis and Control of Speech Power Using Statistical Method," Journal of the Acoustical Society of Japan, Vol. 49, no. 2, pp. 253-259 (December 1993), Miki et al .: "Analysis of Pause Length Change of Sentence Voice by Speech Rate," Proc. 247-248 (December 1992), etc., and can be used to extract rules.
[0035]
The control rules and accumulated segments extracted for each environment include information about the extracted environment to be used at the time of speech synthesis, such as dialogue, whisper, fast-talking voice, tired voice, and energetic voice. Information on the collection status of voice data such as a good voice, a busy voice, a calm voice, etc. is added.
[0036]
As is well known, phonological and prosodic control rules and stored data of existing speech synthesis systems are essentially linguistic environments (eg, morphemes, parts of speech, inflections, etc.) and phonological and prosodic. Environment (for example, arrangement of phonemes, accent type and accent nucleus, pitch, power envelope, timing, etc.) and control contents (for example, reading symbol string, accent binding information, prosodic parameter value, selection priority of storage unit, etc.) and It can be understood as the correspondence with the voice waveform / analysis parameters.
[0037]
Therefore, in this embodiment, a rule extraction environment is added to this correspondence relationship, and the extracted control rules and accumulated data are converted into a linguistic environment, a phonological / prosodic environment and a rule extraction environment, and control contents and voice waveforms. -Described as correspondence with analysis parameters.
[0038]
As described above, a plurality of rules and accumulated data are provided, and the voice synthesis unit 11 (the language analysis unit 111 and the phoneme control unit 112 therein) appropriately selects and uses them, thereby adjusting the tone of the voice of the synthesized voice. Can be given a variation.
[0039]
Therefore, in the present embodiment, in addition to the above-described speech synthesis unit 11, a communication state monitoring unit 12 is provided as means for providing information for determining conditions for rule selection in the speech synthesis unit 11. The communication status monitoring unit 12 includes an intra-computer communication status monitoring unit 121 that monitors a communication status in a computer on which the speech synthesis system operates, and an external communication status monitoring unit 122 that monitors a communication status between the computer and the outside. And
[0040]
The in-computer communication state monitoring unit 121 in the communication state monitoring unit 12 monitors the communication state or the communication state based on the communication state or the quality of the communication path between software, hardware, or software and hardware operating in the same computer. I do. For the sake of simplicity, here, hardware or software that communicates with each other will be simply referred to as communicator A and communicator B, respectively. That is, it is assumed that communication is performed between the communication party A and the communication party B.
[0041]
The in-computer communication state monitoring unit 121 inquires of software or hardware (hereinafter referred to as a communication medium for convenience) that mediates this communication in order to know the communication state exchanged between them, and Situation (for example, information sender, amount of communication or time change of communication amount, frequency of communication, total amount of data to be sent, amount of data already sent, etc.) and quality of communication path (eg, data transfer speed and error Frequency, etc.). These notifications do not always need to be inquired, and even if there is no inquiry, the communication medium side may notify the in-computer communication state monitoring unit 121 at an appropriate timing.
[0042]
As such a communication medium, an existing function provided by an operating system or an operation system (hereinafter, referred to as an OS) (for example, Windows DDE = Dynamic Data Exchange capable of realizing a messaging function, and data transfer using a clipboard) Existing functions provided by the window system (for example, events and selection buffers in the X Window System, Windows messages, etc. are examples), or various services implemented in a server-client model can be used. Of course, it is possible not only to use the existing system, but also to construct a new system having a similar mechanism.
[0043]
Further, in order to know the communication state exchanged between the correspondent A and the correspondent B, a mechanism may be employed in which the correspondents A and B are directly inquired without using a communication medium. In this case, the in-computer communication state monitoring unit 121 inquires of the parts having the function of performing communication (hereinafter, referred to as communication units) which the communication parties A and B have, respectively. In the same way as above, the communication status is notified. Of course, similarly to the above, even if there is no inquiry, the communication units of the correspondents A and B may notify the in-computer communication state monitoring unit 121 at appropriate timing.
[0044]
The in-computer communication status monitoring unit 121 uses the information on the communication status obtained in this way to determine, for example, the ratio of the data that has already been communicated, such as large / small traffic, large / small data to be sent. Is transmitted to the speech synthesizer 11 as communication state information. These pieces of information may be used as communication state information as they are, or may be used as communication state information in the in-computer communication state monitoring unit 121 after being compared with a threshold and rounded to a discrete level.
[0045]
On the other hand, the external computer communication state monitoring unit 122 in the communication state monitoring unit 12 monitors the communication state with the outside of the computer. The communication status monitoring unit 122 outside the computer can also be configured to acquire the communication status via the communication medium, similarly to the communication status monitoring unit 121 inside the computer. Similarly, as the communication medium, the OS, the existing functions provided by the OS (messaging function), the existing functions provided by the window system (for example, events), or various services implemented in the server-client model In addition to (for example, Network File System, printer daemon, etc.), a device or driver such as a modem capable of data communication with a device outside the computer can be used. Of course, it is possible not only to use the existing system but also to construct a new system having the same mechanism as in the above case.
[0046]
The voice synthesizer 11 receives communication status information from the intra-computer communication status monitoring unit 121 (in the communication status monitoring unit 12) and the off-computer communication status monitoring unit 122, respectively, and responds to the phoneme control unit 112 and the prosody according to the communication status information. The control unit 113 selects a control rule and stored data to be applied, respectively.
[0047]
Here, the correspondence between the communication state information, the control rule to be selected, and the stored data is determined in the phoneme control unit 112 and the prosody control unit 113. For example, if the communication volume is large or the frequency of communication is large, make the voice fast, if it is very large, make a tense voice, or conversely, if the communication volume is small or the frequency of communication is small, increase the pitch dynamic range. Use a calm voice, insert more pauses, or relax. When the remaining communication amount is large, the pitch is raised or the voice is made quicker as the voice becomes more relaxed and the remaining voice becomes smaller. When the quality of the communication channel is poor, the voice quality is changed by superimposing irregular voices on a weak voice or pitch. An example is a correspondence relationship in which a light voice is converted into a light voice when the transfer speed is high, and a heavy voice is generated when the transfer speed is low.
[0048]
As described above, the control rules and the accumulated data applied in the speech synthesis unit 11 (in the language analysis unit 111 and the phoneme control unit 112) to the analysis result in the language analysis unit 111 are transmitted to the communication state monitoring unit 12 (in the communication state monitoring unit 12). By switching (changing) and outputting the synthesized voice in accordance with (the communication state indicated by) the communication state information output from the intra-computer communication state monitoring unit 121 or the off-computer communication state monitoring unit 122, the user can: From the tone of the synthesized voice, it is possible to know the communication state inside the computer at that time or the communication state between the outside of the computer.
[0049]
It should be noted that the above-described correspondence is merely an example, and may be changed according to the user's preference of the speech synthesis system. Further, a correspondence may be independently set for each of the communication state inside the computer monitored by the communication state monitoring unit 121 inside the computer and the communication state outside the computer monitored by the communication state monitoring unit 122 outside the computer. Absent.
[0050]
In the above embodiment, the communication status monitoring unit 12 includes both the in-computer communication status monitoring unit 121 and the out-of-computer communication status monitoring unit 122. However, only one of them is provided. It does not matter even if it is.
[Second embodiment]
FIG. 2 is a block diagram showing a second embodiment of the speech synthesis system according to the present invention. The same parts as those in FIG. 1 are denoted by the same reference numerals.
[0051]
First, the configuration of FIG. 2 is characterized in that, in addition to the speech synthesis unit 11, a hardware status monitoring unit 221 for monitoring the operating status of computer hardware and a software status monitoring unit 222 for monitoring the operating status of computer software An operation state monitoring unit 22 having the following is provided. Accordingly, the functions of (the phoneme control unit 112 and the prosody control unit 113) in the speech synthesis unit 11 in FIG. 2 are also changed as described below. Although different from the control unit 113), the same reference numerals are given for convenience.
[0052]
The hardware state monitoring unit 221 in the operation state monitoring unit 22 directly measures a parameter indicating the operation state of the computer hardware on which the speech synthesis system operates, or inquires of the computer hardware or its software driver about the operation state. Alternatively, the operating state of the computer hardware is monitored by notifying the operating state at an appropriate timing from the computer hardware or its software driver itself.
[0053]
For example, the level and stability of the power supply voltage supplied to the hardware configuring the system, the connection status of devices (peripheral devices) such as a card, a printer, a keyboard, a mouse, and network cables such as network cables. (Whether connected or not, and whether it can be used).
[0054]
Based on the monitoring result regarding the hardware status acquired in this way, the hardware status monitoring unit 221 may, for example, have a sufficiently high / high / slightly low / low / substantially low / sufficiently stable / stable power supply voltage. Power quality information, which is ranked as "active" / "slightly unstable" / "very unstable", or operation status information such as "hardware available / standby / connection lost" To the speech synthesizer 11.
[0055]
Note that the above classification is an example, and any classification is possible as needed. In addition, an appropriate threshold value may be set and compared with the threshold value, the result may be rounded to a discrete level, or the acquired numerical value may be used as the operation state information, and the present invention is not limited to the above classification.
[0056]
The speech synthesis unit 11 receives operation state information from the hardware state monitoring unit 221 (in the operation state monitoring unit 22), and according to the operation state information, applies control rules to be applied to the phoneme control unit 112 and the prosody control unit 113, respectively. Select the stored data.
[0057]
Here, the correspondence between the operation state information and the selected control rule and the stored data is the same as the correspondence between the communication state information and the selected control rule and the stored data in the first embodiment. And the prosody control unit 113. For example, when a high-quality power supply voltage is supplied sufficiently stably, a synthetic voice is generated with normal prosody control or timbre, but when the power supply voltage starts to drop or becomes unstable, Choose a rule that selects stored data that corresponds to a slightly unhealthy voice, switch to a rule that creates a loose tone, select a rule that weakens pitch up / down, or a soft voice For example, a correspondence relationship such as selecting a rule that gives Of course, this correspondence is merely an example, and may be changed according to the user's preference of the speech synthesis system. The change of the correspondence between these rules may be selected so as to give the opposite impression to the above.
[0058]
The prosody control unit 113 and the phonology control unit 112 in the speech synthesis unit 11 control the prosodic and phonological quality of the synthesized speech to be generated and output by using the rules selected according to such correspondence. This allows the user to know the state of the computer hardware at that time from the tone of the synthesized voice.
[0059]
By the way, in a portable system (portable device) represented by a PDA (Personal Digital Assistants), since an area available for display is small, it is inefficient to allocate a large area for presenting operation state information of the system. However, if it is too small, the original purpose of calling the user's attention may not be satisfied. In general, for portable systems, the stability of the supplied power is usually lower than for fixed systems in a maintained environment. Therefore, in a portable system, it is effective to convey such operation state information by controlling prosody and voice quality. Like the power supply voltage, the connection status of the hardware is usually relatively unconscious for general users and tends to be overlooked.However, if the prosody and voice quality are changed according to the change in the connection status, the user will be affected. I can inform you implicitly.
[0060]
On the other hand, the software state monitoring unit 222 in the operation state monitoring unit 22 is a software (process) that uses (targets) computer resources such as a processor (CPU), a memory, and a hard disk of a computer in which the speech synthesis system operates. Monitor the operating state of software due to the limited distribution of computer resources, such as how much it occupies, or from the opposite point of view, how long a piece of software is waiting to process, What kind of input is now accepting, for example, what is valid as the type of input device and type of input content, and what information is presented by certain software now, for example, Operating state of software corresponding to operation mode (scene) such as source of presentation information and type of presentation contents To get.
[0061]
The operating state of such software is acquired by the software state monitoring unit 222 by inquiring and notifying the OS on which the software is operating, or the operating state is directly given to the software itself. It is obtained by adding a notification unit (notification function) that notifies when inquired. Needless to say, it is also possible to prepare a mechanism in which the software itself notifies the software status monitoring unit 222 of the operation status at an appropriate timing even if there is no inquiry.
[0062]
Here, examples of the acquired operation state information of the software include, for example, information such as a memory usage amount, a software state, a CPU occupancy rate, a total occupation time, and an operation priority. These pieces of information can be obtained by using the existing OS system calls and libraries. Further, software having a notification unit for notifying the type of input currently accepted and the type of information being presented may be newly created.
[0063]
In general, the type of input to be accepted dynamically changes according to the operation mode (scene) even in the same application. For example, an e-mail application that sends and receives e-mails displays a list of received e-mails, displays the contents of one of the selected e-mails, edits the text of an e-mail to be transmitted, and edits an e-mail. Is transmitted, and depending on each scene, whether the same key input is valid or ignored or what action is performed when the same key input is valid. In the case of software that accepts voice recognition input, information such as what recognition vocabulary can be input now corresponds to “acceptable input types”. Also dynamically changes according to the operation mode (scene) of the software.
[0064]
On the other hand, an e-mail application acquires information indicating the source and content of information, such as who sent the e-mail or the contents of which information is confidential, by character string collation or linguistic analysis. The source of the presentation information and the type of the presentation content are transmitted to the software operation state monitoring unit 222 as operation information. Here, the mail application is described as an example, but the same can be applied to application software for browsing information from a plurality of information sources, such as an electronic network bulletin board or an information providing system on an electronic network.
[0065]
Based on the information on the operating state of the software acquired in this manner, the software state monitoring unit 222 determines, for example, which combination of the memory occupancy is large / small, the total occupation time of the CPU is long / short, and the combination of the recognized vocabulary. The information indicating the set, the operation mode, the source of the information, and the type of the content of the information is used as the operation state information as the operation state information. To the prosody control unit 113).
[0066]
Upon receiving the operation state information from the software state monitoring unit 222, the speech synthesis unit 11 selects rules and accumulated data to be applied in the phoneme control unit 112 and the prosody control unit 113 according to the operation state information. Thus, for example, when the memory occupation is large or the accumulated time occupied by the CPU is long, an unsound voice or an apologetic voice is generated to inform the user of the system status implicitly, or conversely, It is possible to encourage the user to perform his / her own processing by using the tone of the user. In addition, by selecting rules for changing accents and phrases according to the source of the information and reflecting the regional colors in the voice, it is possible to make the user aware of the difference in the source of the information. Also, if there is stored data of the voice of the information provider, the information provider can be easily identified by using the stored data. In addition, when remote operation is performed by a telephone or the like or when the display area is small on a portable device, depending on what kind of input the certain software is currently accepting (the type of the input device and the type of the input content). By changing the prosody and timbre, the user can know what to input next and the current “scene” from the tone of the synthesized speech to be output.
[0067]
Here, the correspondence between the operation state information (the operation state information of the software), the selected control rule, and the stored data is the same as the above-described case of the correspondence between the operation state information of the computer hardware, the selected control rule, and the stored data. First, the phoneme control unit 112 and the prosody control unit 113 determine this.
[0068]
As described above, the control rules and the accumulated data applied in the speech synthesis unit 11 (the phoneme control unit 112 and the prosody control unit 113 therein) to the analysis result of the language analysis unit 111 are transmitted to the operation state monitoring unit 22 (the inside of the operation state monitoring unit 22). By switching (changing) and outputting the synthesized voice according to the operation status information output from the hardware status monitoring unit 221 or the software status monitoring unit 222, the user can change the tone of the synthesized voice. It is possible to know the state of the computer hardware or the state of the computer software at that time.
[0069]
By the way, in the speech synthesis unit 11 of the present embodiment, the language analysis unit 111, the phoneme control unit 112, the prosody control unit 113, the waveform generation unit 114, and the waveform output unit 115 are modularized so as to operate independently. Even if the format of the exchange of data with each other is via a network, processing can be performed even with the exchange of data within the same execution process. In addition, the entire processing procedure and the partial processing procedure of each unit can be separated as separate processes, and the separated process returns a processing result to the original process. Such a system can be easily mounted on a multitask OS by using a system call such as generation of a child process and socket communication with the child process, and a library.
[0070]
Upon receiving the operating state information from the operating state monitoring unit 32, the voice synthesizing unit 11 determines the remaining amount of the memory, the occupation time and the occupancy of the CPU, and receives a waveform from the language analyzing unit 111 constituting the voice synthesizing unit 11. In proceeding with the speech synthesis processing reaching the output unit 115, it is checked whether or not sufficient computer resources such as memory and CPU capacity are secured. Then, when there is a possibility that the memory may be insufficient, or when it is determined that sufficient computer resources cannot be secured from the load status of the CPU, the voice synthesis unit 11 may perform a process after the stage in which the process is currently performed. Of the above processing is assigned to another computer hardware as another process, and the processing result is received.
[0071]
Here, which process is to be shared is determined based on the CPU capacity and the amount of memory required for the process, and this depends on the type of the speech synthesis method and the scale of the stored data. For example, in the case of the analysis parameter synthesizing method, a lot of resources are used for the signal processing in the waveform generation unit 114 and then for the processing of editing the stored data in the phoneme control unit 112. Retrieving data requires the most CPU power. Also in the analysis parameter synthesizing method, the search time of the stored data becomes longer as the number of types of the stored segments included in the stored data increases. Therefore, which process is to be shared may be determined by assigning an appropriate priority according to the synthesis method or the scale of the stored data, and determining the process to be shared according to the priority.
[0072]
By the way, the application of the rules in the language analysis and phonological / prosodic control of the speech synthesis unit 11 counts the possibility of applying many rules, and the processing is performed by evaluating the result of applying the rules. Proceed. It is not always efficient to execute these processes sequentially on one computer, but it is more efficient to process the applicability of rules and the evaluation in the case of assuming rule application in parallel. If the rules to be applied are fixed, it is possible to make the problem relatively invisible by performing sequential processing by tuning to some extent, but the rules to be applied are dynamic as in the present invention. When the number is changed to a parallel processing, it is more efficient to perform the processing in parallel.
[0073]
Therefore, in the present embodiment, the applicability of the above-described rule and the evaluation when the rule application is assumed are processed in parallel. This parallel processing can be executed as a remote process by another computer on the network to which the computer is connected, or can be shared by a sub-processor on the same computer.
[0074]
As described above, in the present embodiment, when proceeding with the speech synthesis processing from the language analysis unit 111 constituting the speech synthesis unit 11 to the waveform output unit 115, sufficient computer resources are provided by the operation state information from the operation state monitoring unit 32. If it is judged that the process cannot be secured, the appropriate process among the processes after the stage in which the process is progressing up to now is assigned to another computer hardware as a separate process, language analysis, phonological / prosodic The applicability of rules in dynamic control and the evaluation when rules are assumed to be applied are shared by other computers on the network to which the computer is connected and the sub-processors on the same computer, and are processed simultaneously and in parallel. Thus, efficient processing is realized and the waiting time of the user is reduced.
[0075]
In the above embodiment, both the hardware status monitoring unit 221 and the software status monitoring unit 222 are provided in the operation status monitoring unit 22, but only one of them is provided. May be present.
[Third Embodiment]
FIG. 3 is a block diagram showing a third embodiment of the speech synthesis system according to the present invention. The same parts as those in FIG. 1 are denoted by the same reference numerals.
[0076]
First, the configuration of FIG. 3 is characterized in that, in addition to the speech synthesis unit 11, a user status monitoring unit 321 that monitors the system usage status of the user and a user environment monitoring unit 322 that monitors the system usage environment of the user. The point is that a user status monitoring unit 32 is provided. Accordingly, the functions of (the phoneme control unit 112 and the prosody control unit 113) in the speech synthesis unit 11 in FIG. 3 are also changed as described below. Although different from the control unit 113), the same reference numerals are given for convenience.
[0077]
The user status monitoring unit 321 in the user status monitoring unit 32 monitors information from at least one of an input device, a clock, and a usage history for obtaining a usage status (user status) of the user's system. Of the system usage status, such as how concentrated the user is using the system. As the input device, for example, a camera can be used. It is possible to accurately estimate the direction of the user's head captured by the camera, and how long and stably the system is facing the front (front) during a certain period of time (whether it is facing another ) Is evaluated as the degree of concentration of the user. In addition, with respect to an input device such as a pointing device represented by a mouse, a keyboard, and the like for input operation by the user, the operation status of the user (input operation frequency, input operation time, moving speed and distance of the pointing device, etc.) can be monitored. It is possible. In addition, the clock and the usage history are used to improve the estimation accuracy of the usage status by recording the usage status in the same day of the week and the same time zone.
[0078]
The user status monitoring unit 321 uses the information about the system usage status of the user obtained in this way and uses the information indicating the degree of concentration of the user and the operation status of the user as user status information as the voice synthesis unit 11 (in the To the phoneme control unit 112 and the prosody control unit 113).
[0079]
Upon receiving the user status information from the user status monitoring unit 321, the voice synthesis unit 11 selects rules and accumulated data to be applied in the phoneme control unit 112 and the prosody control unit 113 according to the user status information. Accordingly, for example, when the degree of concentration is equal to or less than a predetermined threshold, attention is paid to the user by increasing the power or applying a rule such that the utterance speed is small (slow) at the beginning of the sentence. It is possible to encourage.
[0080]
On the other hand, the user environment monitoring unit 322 in the user status monitoring unit 32 is provided with at least one of an input device, a clock, and a usage history for obtaining an environment (user environment) of a place where the user uses the system. Monitors information and monitors the system usage environment, such as what kind of sound environment (ambient noise environment) the user is in, how bright the user is, and the physical location (position) of the user. Output the result. Examples of such an input device include a microphone that collects ambient noise, a position estimation device such as a GPS, a brightness sensor, a camera, a gas sensor, and a water sensor. In addition, the clock and the usage history are used to improve the estimation accuracy of the usage environment by recording the usage environment in the same day of the week and the same time zone.
[0081]
The user environment monitoring unit 322, based on the information on the user's system usage environment acquired in this manner, displays information indicating the spectral characteristics and level of ambient noise, brightness, the user's whereabouts (position), and the like. The information is sent to the speech synthesis unit 11 (the phoneme control unit 112 and the prosody control unit 113 therein) as information.
[0082]
Upon receiving the user status information from the user environment monitoring unit 322, the voice synthesis unit 11 selects rules and accumulated data to be applied in the phoneme control unit 112 and the prosody control unit 113 according to the user status information. Thus, for example, when there is dominant noise in the high frequency component, the prosody control rules are applied so that the voice has a high pitch so that the voice can be clearly heard, and the predominant storage unit of the high frequency component is selected. Phonological control rules can be applied, or in quiet places with low noise levels, rules can be applied that result in a quiet or calm voice. Also, by using a prosodic rule so that the pitch is higher and the utterance speed is faster when using in a bright place, the utterance rate is slower when using in a dark place and the dynamic range of the pitch is widened. It can also give a relatively calm impression in dark places compared to bright places. Such a correspondence may be changeable according to the user's preference.
[0083]
In the above embodiment, the user status monitoring unit 32 is provided with both the user status monitoring unit 321 and the user environment monitoring unit 322, but only one of them is provided. It may be something.
[Fourth embodiment]
FIG. 4 is a block diagram showing a fourth embodiment of the speech synthesis system according to the present invention. The same parts as those in FIG. 3 are denoted by the same reference numerals.
[0084]
First, the configuration of FIG. 4 is characterized in that, in addition to the configuration of FIG. 3 (the voice synthesis unit 11 and the user status monitoring unit 32), a non-natural voice explicit determination unit 41 and a text change unit 42 are provided. It is. Accordingly, the function of the speech synthesis unit 11 (such as the language analysis unit 111) in FIG. 4 is also different from that of the speech synthesis unit 11 (such as the language analysis unit 111) in FIG. 3, as described below. The same reference numerals are used for convenience.
[0085]
First, based on the user situation information output from the user situation monitoring unit 32, the non-natural speech explicit determination unit 41 determines whether or not it is necessary to clearly indicate that the speech is not a human utterance (non-natural speech). The judgment result (non-natural voice explicit judgment result) is output. For example, if the user situation information indicates that the user is not concentrated, or if it indicates a time zone or place where synthesized speech has not been output so far, human utterances The result of the judgment that it should be clearly indicated that the message should not be output.
[0086]
The text changing unit 42 receives the non-natural voice explicit determination result from the non-natural voice explicit determination unit 41, and if the determination result indicates that it is not a human utterance, it corresponds to the input text. Prior to outputting the synthesized speech (that is, prior to outputting the linguistic analysis result for the input text in the linguistic analysis unit 111 to the phonological control unit 112 and the prosody control unit 113 to generate and output the corresponding audio waveform), Prefix fixed expressions such as "Synthetic sound" or "Notification from the system" to announce the beginning of message output by voice synthesis. The speech synthesizer 11 synthesizes and outputs the vocabulary prefixed by the text changer 42.
[0087]
In this way, for example, when the user is not concentrated, or when the synthesized speech is used in a time zone or a place where the synthesized speech has not been output so far, the message output by the speech synthesis is notified in advance ( Such a situation can be notified to the user by outputting the synthesized speech (which explicitly indicates that the speech is non-natural) before outputting the synthesized speech corresponding to the input text. In particular, in a situation where synthesized voice with high quality and close to real voice is output, in the case of use in a user environment where human voices are used, a synthetic sound indicating that it is a non-natural voice is prefixed. By doing so, it is possible to avoid attracting attention by emphasizing that it is a synthesized sound by not confusing it with the voice of the surrounding people or not explicitly indicating that it is a non-natural sound.
[Fifth Embodiment]
FIG. 5 is a block diagram showing a fifth embodiment of the speech synthesis system according to the present invention. The same parts as those in FIG. 4 are denoted by the same reference numerals.
[0088]
First, the feature of the configuration of FIG. 5 is that, instead of the text changing unit 42 shown in FIG. 4, a non-natural voice explicit sound output unit 43 that outputs a sound (non-natural voice explicit sound) that clearly indicates that it is not a human utterance. Is provided. Along with this, the function of the speech synthesis unit 11 in FIG. 4 also includes, for example, the waveform output unit 115 in the speech synthesis unit 11 and the synthesized voice generated by the waveform generation unit 114 and the non-natural voice explicit sound output unit 43. It differs from the speech synthesis unit 11 (the waveform output unit 115 and the like) in FIG. 4 in that it has a function of mixing the generated non-natural voice explicit sound, but for convenience the same reference numerals are used. is there.
[0089]
First, the non-natural voice explicit sound output unit 43 outputs the input text when the non-natural voice explicit judgment result output from the non-natural voice explicit judgment unit 41 indicates that it is not a human utterance. Prior to the output of the synthesized speech corresponding to (i), for example, a signal sound (non-natural sound explicit sound) such as “P” is output. This signal sound is output by the waveform output unit 115 prior to the output of the synthesized voice generated by the waveform generation unit 114 according to the phonological / prosodic control by the phonological control unit 112 and the prosody control unit 113.
[0090]
In this way, when the user is not concentrated, or in a time zone or place where synthesized voice has not been output so far, for example, a non-natural voice explicit sound such as "Pi" is input. By outputting the synthesized speech corresponding to the text prior to outputting the synthesized speech, it is possible to call attention to the user by clearly indicating that the message is not a human voice but a synthesized speech.
[0091]
In addition, it is also possible to add the text changing unit 42 in FIG. 4 to the configuration in FIG. 5 and to provide a configuration including both the text changing unit 42 and the non-natural voice explicit sound output unit 43.
[Sixth Embodiment]
FIG. 6 is a block diagram showing a sixth embodiment of the speech synthesis system according to the present invention. The same parts as those in FIG. 1 are denoted by the same reference numerals.
[0092]
First, the feature of the configuration of FIG. 6 is that, in addition to the configuration of FIG. 1 (the voice synthesis unit 11 and the communication state monitoring unit 12), a text change unit 42 for changing the input text as shown in FIG. 4 is provided. (However, the content of the text change is different from the example of FIG. 4). Along with this, the functions in (speech analysis section 111 etc.) in the speech synthesis section 11 are also different from the speech synthesis section 11 (language analysis section 111 etc.) in FIG. 1 as described below, but are the same for convenience. The reference numerals are attached.
[0093]
In the speech synthesis system having the configuration shown in FIG. 6, when the language analysis unit 111 in the speech synthesis unit 11 receives communication state information from the communication state monitoring unit 12, the language analysis unit 111 passes the information to the text change unit 42 and starts up.
[0094]
Then, the text changing unit 42 changes the text by inserting a vocabulary of a fixed expression according to the communication state information into the input text that has been subjected to the language analysis by the language analyzing unit 111 while communicating with the language analyzing unit 111. That is, at the beginning of the processing of the phoneme control unit 112 and the prosody control unit 113 in the speech synthesis unit 11 or at the stage where the pause insertion position is determined during the processing of the prosody control unit 113, the text changing unit 42 At the end of a sentence or at a pause insertion position, a vocabulary of a fixed expression determined by the communication state information is inserted. The speech synthesis unit 11 synthesizes and outputs the vocabulary including the vocabulary inserted by the text change unit 42.
[0095]
By the above-described text change processing according to the communication state information in the text change unit 42, for example, when the communication volume is large (when communication is congested), “A”, “Em”, “Em”, “Yes”, etc. Can be inserted at the beginning or end of the sentence or at the position where the pause is inserted, or a message such as "Wait a minute" can be placed at the beginning of the sentence. By inserting such a fixed vocabulary (a vocabulary set in advance), there is an effect that the processing time is shortened and the load due to the processing of the synthesized speech is reduced. Conversely, when the traffic is small, inserting unnecessary words similar to the above has the effect of implicitly indicating to the user that the system is idle.
[0096]
Although the text changing unit 42 in the configuration of FIG. 6 receives the communication status information from the communication status monitoring unit 12 through the voice synthesis unit 11, the text changing unit 42 may directly receive the communication status information from the communication status monitoring unit 12.
[Seventh embodiment]
FIG. 7 is a block diagram showing a seventh embodiment of the speech synthesis system according to the present invention. The same parts as those in FIG. 2 are denoted by the same reference numerals.
[0097]
First, a feature of the configuration of FIG. 7 is that a text changing unit 42 as shown in FIG. 6 is provided in addition to the configuration of FIG. 2 (the speech synthesis unit 11 and the operation state monitoring unit 22). . Along with this, the functions in (the language analysis unit 111 and the like) in the speech synthesis unit 11 in FIG. 7 are also different from those in the speech synthesis unit 11 (the language analysis unit 111 and the like) in FIG. 2 as described below. The same reference numerals are used for convenience.
[0098]
In the speech synthesis system having the configuration shown in FIG. 7, when the language analysis unit 111 in the speech synthesis unit 11 receives the operation state information of the system from the operation state monitoring unit 22, the language analysis unit 111 passes the information to the text change unit 42 and starts up.
[0099]
Then, the text changing unit 42 inserts the vocabulary of the fixed expression according to the operation state information into the input text that has been subjected to the language analysis by the language analysis unit 111 while communicating with the language analysis unit 111. That is, at the beginning of the processing of the phoneme control unit 112 and the prosody control unit 113 in the speech synthesis unit 11 or at the stage where the pause insertion position is determined during the processing of the prosody control unit 113, the text changing unit 42 A vocabulary of a fixed expression determined by the operation state information is inserted at the end of a sentence or at a pause insertion position. The speech synthesis unit 11 synthesizes and outputs the vocabulary including the vocabulary inserted by the text change unit 42.
[0100]
By the above-described text change processing in accordance with the operation state information in the text change unit 42, for example, when the CPU is occupied for a long time, unnecessary words such as “A”, “Eto”, “Eye”, It can be inserted at the end of a sentence or at a pause insertion position. By inserting such a fixed vocabulary, there is an effect that the processing time is increased and the load due to the processing of the synthesized speech is reduced.
[0101]
Although the text change unit 42 in the configuration of FIG. 7 receives the operation state information from the operation state monitoring unit 22 through the speech synthesis unit 11, the text change unit 42 may directly receive the operation state information from the operation state monitoring unit 22.
[Eighth Embodiment]
FIG. 8 is a block diagram showing an eighth embodiment of the speech synthesis system according to the present invention. The same parts as those in FIG. 3 are denoted by the same reference numerals.
[0102]
First, the configuration of FIG. 8 is characterized in that a text changing unit 42 as shown in FIG. 6 is provided in addition to the configuration of FIG. 3 (the speech synthesis unit 11 and the user status monitoring unit 32). . Along with this, the functions in (the language analysis unit 111 and the like) in the speech synthesis unit 11 in FIG. 8 are also different from those in the speech synthesis unit 11 (the language analysis unit 111 and the like) in FIG. 3 as described below. The same reference numerals are used for convenience.
[0103]
In the speech synthesis system having the configuration shown in FIG. 8, when the language analysis unit 111 in the speech synthesis unit 11 receives the user status information from the user status monitoring unit 32, it passes the information to the text change unit 42 and starts up.
[0104]
Then, the text changing unit 42 inserts the vocabulary of the fixed expression according to the user situation information into the input text that has been subjected to the language analysis by the language analysis unit 111, while contacting the language analysis unit 111. That is, at the beginning of the processing of the phoneme control unit 112 and the prosody control unit 113 in the speech synthesis unit 11 or at the stage where the pause insertion position is determined during the processing of the prosody control unit 113, the text changing unit 42 At the end of the sentence or at the position where the pause is inserted, the vocabulary of the fixed expression determined by the user situation information is inserted. The speech synthesis unit 11 synthesizes and outputs the vocabulary including the vocabulary inserted by the text change unit 42.
[0105]
By the text change processing according to the operation state information in the text change unit 42 described above, for example, when the user is not concentrated, a vocabulary that calls out to a person such as “that” is set at the beginning of the sentence, You can call attention.
[0106]
Although the text changing unit 42 in the configuration of FIG. 8 receives the user status information from the user status monitoring unit 32 through the speech synthesizing unit 11, the text changing unit 42 may directly receive the user status information from the user status monitoring unit 32.
[0107]
【The invention's effect】
As described in detail above, according to the present invention, not only the direct communication as a message of the linguistic information, but also an implicit indication of the situation of the entire system including the speech synthesis function, A user-friendly system can be constructed using the information transmission function. In addition, it is possible to output a synthesized voice according to the usage status of the user.
[0108]
In particular, it is important from the viewpoint of a user interface to inform the user of the internal state of the system as an output medium of a computer. It can be said that it is appropriate to use the audio media to inform the user of the operating state of the system at the same time as using it as the main purpose as a language message transmission.
[0109]
Such information can be used not only to enhance its effect by being used together with a visual output such as a screen display unit, but also when the area of the display unit is small like a portable device represented by a PDA. If the message is transmitted mainly by audio media, it is possible to suppress the occupation of the screen area by displaying the message.
[0110]
Furthermore, by controlling the prosody and timbre in consideration of the user's usage situation, a more natural system output becomes possible. Not only does it not output monotonous synthesized speech without considering the situation, but in situations where high-quality synthesized voices close to real voice increase, it is clear that it is synthesized speech, In spite of the unnaturalness, natural communication becomes possible as communication with machines.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a first embodiment of a speech synthesis system according to the present invention.
FIG. 2 is a block diagram showing a second embodiment of the speech synthesis system of the present invention.
FIG. 3 is a block diagram showing a third embodiment of the speech synthesis system according to the present invention.
FIG. 4 is a block diagram showing a fourth embodiment of the speech synthesis system according to the present invention.
FIG. 5 is a block diagram showing a speech synthesis system according to a fifth embodiment of the present invention.
FIG. 6 is a block diagram showing a sixth embodiment of the speech synthesis system of the present invention.
FIG. 7 is a block diagram illustrating a speech synthesis system according to a seventh embodiment of the present invention.
FIG. 8 is a block diagram showing an eighth embodiment of the speech synthesis system according to the present invention.
[Explanation of symbols]
11 ... Speech synthesis unit,
12 ... communication state monitoring unit
22 ... operation state monitoring unit
32: User status monitoring unit
41 ... non-natural voice explicit judgment unit,
42 ... text change section
43 ... Non-natural voice explicit sound output unit,
111 ... Language analysis unit,
112 ... phoneme control unit,
113 ... Prosody control unit,
114 ... waveform generator
115 ... waveform output unit,
121: communication status monitoring unit in the computer
122: communication status monitoring unit outside the computer
221, a hardware status monitoring unit;
222: software state monitoring unit
321, a user state monitoring unit,
322: user environment monitoring unit,
1121... Stored data storage unit.

Claims (9)

Speech synthesis means for performing linguistic analysis of the input text, applying rules to the analysis results, performing phonological / prosodic control, and generating / outputting synthesized speech;
A communication state monitoring unit that monitors at least one of a communication state in the computer and a communication state between the computer and the outside and outputs communication state information,
The speech synthesis system, wherein the speech synthesis unit is configured to change a rule applied in the phonological / prosodic control in accordance with the communication state information output from the communication state monitoring unit. . Speech synthesis means for performing linguistic analysis of the input text, applying rules to the analysis results, performing phonological / prosodic control, and generating / outputting synthesized speech;
Operating state monitoring means for monitoring at least one of the operating state of the computer hardware and the operating state of the computer software and outputting operating state information,
The speech synthesis means is applied in the phonological / prosodic control so that the tone of the synthesized voice corresponds to the operation state in order to notify the user of the operation state in the tone of the synthesized voice. A speech synthesis system configured to change a rule to be performed according to the operation state information output from the operation state monitoring unit. At least one process of language analysis, phonological control, prosodic control, and voice waveform generation in the voice synthesis means is performed by a communicable wired network or wireless network in accordance with the operation state indicated by the operation state information. 3. The speech synthesis system according to claim 2, wherein said plurality of computer hardware are assigned to each other. Speech synthesis means for performing linguistic analysis of the input text, applying rules to the analysis results, performing phonological / prosodic control, and generating / outputting synthesized speech;
User status monitoring means for monitoring at least one of a user's concentration level, a user's operation status of the system, and a position and brightness of a place where the user uses the system as a user status,
The voice synthesis unit changes a rule applied in the phonological / prosodic control according to the user status monitored by the user status monitoring unit so that the tone of the voice of the synthesized voice corresponds to the user status. A speech synthesis system characterized in that the speech synthesis system is configured to The apparatus further comprises a non-natural voice explicit determination means for outputting a determination result as to whether or not the synthetic voice is not a human utterance based on the user situation information, and the non-natural voice explicit determination means In response to the judgment result, a text change unit that specifies that the synthesized voice is not a human voice by changing a part of the expression of the input text, and according to the judgment result of the non-natural voice explicit judgment unit, The speech synthesis system according to claim 4, further comprising at least one of a non-natural voice explicit sound output unit that outputs a sound indicating that the synthesized voice is not a human utterance in addition to the output. A speech synthesis method that performs linguistic analysis of an input text, applies rules to the analysis result, performs phonological and prosodic control, and generates and outputs synthesized speech.
Monitoring at least one of a communication state in the computer and a communication state between the computer and the outside, and changing a rule applied in the phonological / prosodic control according to the communication state according to the monitoring result. A speech synthesis method characterized by the following. A speech synthesis method that performs linguistic analysis of an input text, applies rules to the analysis result, performs phonological and prosodic control, and generates and outputs synthesized speech.
In order to monitor at least one of the operation state of the computer hardware and the operation state of the computer software, and to inform the user of the operation state by the tone of the voice of the synthesized voice, the tone of the voice of the synthesized voice is changed to the operation state. Correspondingly, a rule applied in the phonological / prosodic control is changed according to the result of the monitoring. A speech synthesis method that performs linguistic analysis of an input text, applies rules to the analysis result, performs phonological and prosodic control, and generates and outputs synthesized speech.
At least one of a user's concentration, a user's operation state of the system, and a position and brightness of a place where the user uses the system is monitored as a user state,
A speech synthesis method, wherein rules applied in the phonological / prosodic control are changed in accordance with the result of the monitoring so that the tone of the voice of the synthesized voice corresponds to the user monitoring situation. The speech synthesis method according to any one of claims 6 to 8, further comprising outputting at least one of another sound and a synthesized speech in addition to the synthesized speech corresponding to the input text.

Images