CONVERSION DEVICE, CONVERSION METHOD, AND PROGRAM

Abstract

Converts voices to natural voice quality regardless of the original voice quality. [Solution] The conversion device 1 includes an acquisition unit 21 that acquires the range of a voice conversion model and the speaker's range, and a conversion unit 23 that converts the range of speaker P's speech data 14 into the range of the voice conversion model and outputs the speech data 17 after the range conversion to the voice conversion model.

Description

[Technical field]

[0001] The present invention relates to a conversion device, a conversion method, and a program. [Background technology]

[0002] With the recent development of AI (Artificial Intelligence) technology, there is a voice conversion technology that can imitate voice quality with good quality. The voice conversion model used for voice conversion technology is trained with the voice of a character such as a voice actor. When the voice quality of the source voice of the voice conversion technology and the voice quality of the character of the voice conversion model are similar, the voice conversion technology can output a natural voice.

[0003] In Patent Document 1, when a characteristic voice is input, the voice reflecting the characteristic is converted into a voice corresponding to the speaker information of the conversion destination. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2024-018197 A Summary of the Invention [Problem to be solved by the invention]

[0005] When the source voice quality and the converted speaker's voice quality are not similar, the voice conversion technology may output an unnatural voice quality.

[0006] The present invention has been made in consideration of the above circumstances, and an object of the present invention is to provide a technique capable of converting a voice into a natural voice quality regardless of the voice quality of the source voice. [Means for solving the problem]

[0007] A conversion device of one embodiment of the present invention includes an acquisition unit that acquires the range of a voice conversion model and the range of a speaker, and a conversion unit that converts the range of the speaker's speech data into the range of the voice conversion model and outputs the speech data after range conversion to the voice conversion model.

[0008] In one embodiment of the conversion method of the present invention, a computer obtains the range of a voice conversion model and the range of a speaker, converts the range of the speaker's speech data to the range of the voice conversion model, and outputs the speech data after range conversion to the voice conversion model.

[0009] A program of one embodiment of the present invention causes a computer to function as an acquisition unit that acquires the range of a voice conversion model and the range of a speaker, and a conversion unit that converts the range of the speaker's speech data into the range of the voice conversion model and outputs the speech data after range conversion to the voice conversion model. Effect of the Invention

[0010] According to the present invention, it is possible to provide a technology capable of converting a voice into a natural voice quality regardless of the voice quality of the source voice. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a diagram illustrating the system configuration of the conversion system of the present disclosure. [Diagram 2] FIG. 2 is a diagram illustrating a conversion method in the conversion system of the present disclosure. [Diagram 3] FIG. 3 is a diagram illustrating functional blocks of the conversion device. [Figure 4] FIG. 4 is a diagram illustrating an example of the data structure of the range data. [Diagram 5] FIG. 5 is a diagram illustrating an example of the data structure of character range data. [Figure 6] FIG. 6 is a diagram illustrating an example of compression of a range by a conversion device. [Figure 7] FIG. 7 is a flowchart illustrating the conversion process performed by the conversion device. [Figure 8] FIG. 8 is a diagram illustrating a system configuration of a conversion system according to a modified example. [Figure 9] FIG. 9 is a diagram illustrating the functional blocks of the training device. [Figure 10] FIG. 10 is a diagram for explaining a conversion method in the conversion system of the modified example. [Figure 11] FIG. 11 is a diagram for explaining the hardware configuration of a computer used in a device such as a conversion device. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the description of the drawings, the same parts are given the same reference numerals and the description will be omitted.

[0013] (Conversion System) 1, the conversion system 7 according to the present disclosure inputs the speech of a speaker P to a voice conversion model, converts the speech of the speaker P into a predetermined voice quality using the voice conversion model, and outputs the converted speech. The voice conversion model may be called, for example, an AI voice changer.

[0014] In the present disclosure, range is a range of pitches. Quality is a characteristic or sound of a voice. Quality, in the present disclosure, may exclude range or may include range.

[0015] The voice quality after conversion of the voice conversion model is any voice quality, such as the voice quality of a character played by a voice actor, a voice quality generated programmatically, etc. The voice conversion model is a model that has learned any voice quality, such as the voice quality of a character played by a voice actor, a voice quality generated by an experiment, etc.

[0016] In the present disclosure, the voice conversion model is a model learned from the voice quality of a character. A voice conversion model is prepared for each character identifier. The characters may include virtual characters for voice qualities generated through experiments or the like.

[0017] As shown in FIG. 1, the conversion system 7 includes a conversion device 1 and a voice quality conversion device 3.

[0018] The voice conversion device 3 converts the voice quality of the input speech data into a predetermined voice quality using a voice conversion model. The voice conversion device 3 may have multiple voice conversion models and convert the input voice quality using a specified voice conversion model for output.

[0019] In the present disclosure, for each of the voice conversion models used by the voice conversion device 3, the range of the voice quality after conversion of the voice conversion model is determined in advance. The range of the voice conversion model is the range of the character etc. trained by the voice conversion model, and is a range that can ensure the quality of the voice conversion.

[0020] The range of the voice conversion model is shared with the conversion device 1. The conversion device 1 holds the range of the converted voice quality for each voice conversion model used by the voice conversion device 3 as character range data 12.

[0021] In the present disclosure, the conversion device 1 processes the speech data of the speaker P so that the voice conversion device 3 can convert the voice into a natural voice quality using a voice conversion model. The conversion device 1 converts the range of the speech data input to the voice conversion device 3 to match the range of the voice conversion model. The conversion device 1 first identifies the range of the character of the voice conversion model and the range of the speaker P. The conversion device 1 converts the range of the speech data 14 of the speaker P into the range of the character of the voice conversion model, and inputs the speech data after the range conversion to the voice conversion device 3.

[0022] Data in which the range of speaker P's speech data 14 has been converted to the range of the character is input to the voice conversion device 3. The voice conversion device 3 uses a voice conversion model to convert the input data into the character's voice quality, and outputs the converted data. The voice conversion device 3 outputs data in which the speech of speaker P has been converted to the voice quality of the voice conversion model, with the same content, speed, and pitch of speaker P's speech.

[0023] In this way, the conversion device 1 inputs the speech data converted into the range of the voice quality to be converted to the voice conversion device 3. As a result, even if the voice quality of the speaker P is not similar to the voice quality converted by the voice conversion model, the conversion system 7 can output the speech in which the voice quality of the speaker P's speech has been naturally converted into the voice quality of the voice conversion model.

[0024] The conversion method performed by the conversion system 7 will be described with reference to FIG.

[0025] In step S1, the conversion device 1 acquires the vocal range of the character of the voice conversion model used in the voice conversion device 3 and the vocal range of the speaker P. For example, the conversion device 1 acquires a character identifier designated by the speaker P, and acquires the vocal range of the designated character identifier from the character vocal range data 12. The conversion device 1 has the speaker P read a sample sentence or the like, acquires the speech data of the speaker P, and acquires the vocal range of the speaker P from the speech data.

[0026] In step S2, the conversion device 1 acquires speech data 14 of the speaker P. The acquired speech data 14 is the target of conversion by the voice quality conversion model.

[0027] In step S3, the conversion device 1 removes non-target data 15 from the speech data 14. The non-target data 15 is data that is not a target for conversion by the voice conversion model used by the voice conversion device 3. The non-target data 15 is, for example, laughter, groaning, sneezing, etc., which are sounds made by people but are not speaking voices.

[0028] In step S4, the conversion device 1 outputs the non-target data 15 removed from the speech data 14 to the voice quality conversion device 3.

[0029] In step S5, the conversion device 1 converts the range of the speech data 16 after the non-target data 15 has been removed into the range of the character acquired in step S1, to generate speech data 17 after range conversion. The conversion device 1 converts the range by a method such as shift conversion or range compression.

[0030] In step S6, the conversion device 1 outputs to the voice quality conversion device 3 the range-converted speech data 17 generated in step S5.

[0031] In step S7, the voice conversion device 3 converts the range-converted speech data 17 input from the conversion device 1 into the character's voice quality using the voice conversion model. In step S8, the voice conversion device 3 synthesizes the non-target data 15 input in step S4 with the data converted into the character's voice quality, and plays back the synthesized data.

[0032] The process shown in FIG. 2 is an example and is not limiting.

[0033] For example, the conversion system 7 of the present disclosure describes a case in which the conversion device 1 converts the range of a block of speech data 14 input by a speaker P, and the voice quality conversion device 3 converts the voice quality of the data whose range has been converted by the conversion device 1, but this is not limited to this.

[0034] The speech of speaker P and the voice quality conversion of the speech may be performed in real time. For example, the conversion device 1 sequentially acquires speech data 14 while speaker P is speaking, and identifies unit data to be processed from the speech data 14. The conversion device 1 sequentially repeats the processes from step S3 to step S8 for each unit data. The unit data is a piece of data obtained by dividing the speech data 14 into a predetermined unit such as a phoneme, a syllable, or a mora.

[0035] This allows the conversion system 7 to perform the speech of the speaker P and convert the voice quality of the speech in real time. By performing voice quality conversion in real time, the conversion system 7 can be used for real-time processing such as voice chat with other people.

[0036] In the conversion system 7 according to the present disclosure, the conversion device 1 converts the range of the speech data 14 of the speaker P to match the range of the voice conversion model, and outputs the converted data to the voice conversion device 3. As a result, the voice conversion device 3 only needs to convert the voice quality of the data with a range that matches the range of the character of the voice conversion model, and can convert the voice quality with high accuracy and output natural speech data.

[0037] In addition, sounds that are human sounds but are not speech, such as the non-target data 15, can cause a deterioration in the accuracy of conversion by the voice conversion model. Therefore, in the conversion system 7 according to the present disclosure, the conversion device 1 removes the non-target data 15 from the speech data 14 and outputs the non-target data 15 to the voice conversion device 3, and outputs the speech data 17 after range conversion in which the range of the speech data 14 is converted to the voice conversion device 3. The voice conversion device 3 converts the voice quality of the speech data 17 after range conversion input from the conversion device 1 using the voice conversion model, and then plays it by superimposing the non-target data 15 input from the conversion device 1. In this way, the conversion system 7 can play each of the utterances of the speaker P while ensuring the accuracy of the voice quality conversion by the voice conversion model.

[0038] (Conversion device) The conversion device 1 according to the present disclosure will be described with reference to FIG.

[0039] 3, the conversion device 1 includes each data of range data 11, character range data 12, character identifier 13, speech data 14, non-target data 15, speech data after removal 16, and speech data after range conversion 17, and each function of an acquisition unit 21, a removal unit 22, and a conversion unit 23. Each data is stored in a storage device such as memory 902 or storage 903. Each function is implemented in a CPU 901.

[0040] The range data 11 is data on the range of the speaker P. For example, the speaker P is made to speak in advance by having the speaker P read a sample sentence. The conversion device 1 identifies the range of the speaker P from the speaker P's speech and sets it as the range data 11. For example, as shown in FIG. 4, the range data 11 identifies the lowest note and the highest note as the range of the speaker P. The range data 11 may be set by the speaker P. Alternatively, the range data 11 may be set to a range estimated from attribute information of the speaker P, such as the age, sex, and height.

[0041] The character range data 12 is data for specifying the range of a voice conversion model. The range of the voice conversion model is a range that can ensure the quality of voice conversion.

[0042] When there is a voice conversion model for each of a plurality of characters, the vocal range of each character is specified. The character vocal range data 12 associates the lowest note and the highest note with the character identifier, for example, as shown in FIG.

[0043] The character identifier 13 identifies the character of one of the multiple voice conversion models used by the voice conversion device 3.

[0044] The speech data 14 is data of an utterance by the original speaker P. The speech data 14 has a data format that can be processed by a computer.

[0045] The non-target data 15 is data of sounds that are not speech sounds among the speech data 14. The non-target data 15 is generated by the removal unit 22.

[0046] The post-removal utterance data 16 is data obtained after the non-target data 15 has been removed from the utterance data 14. The post-removal utterance data 16 is data of the utterance data 14 that is to be processed by the conversion unit 23. The post-removal utterance data 16 is generated by the removal unit 22.

[0047] The speech data 17 after range conversion is data obtained by converting the range of the speech data 16 after removal into the range of the voice quality conversion model. The speech data 17 after range conversion is generated by the conversion unit 23.

[0048] The acquisition unit 21 acquires the range of the voice quality conversion model, the range of the speaker P, and the speech data 14 of the speaker P.

[0049] The range of the voice conversion model is acquired, for example, by the speaker P specifying the character of the voice conversion model. In the present disclosure, the conversion device 1 holds the range of each of the characters of the multiple voice conversion models used by the voice conversion device 3 as character range data 12. The acquisition unit 21 specifies one character identifier 13 of the multiple characters defined by the character range data 12 through an operation by the speaker P, etc., and acquires the range corresponding to the specified character identifier 13 from the character range data 12.

[0050] The acquiring unit 21 acquires the vocal range of the speaker P as the vocal range data 11 in advance.

[0051] After acquiring the range of the voice conversion model and the range of speaker P, the acquisition unit 21 acquires the speech data 14 of speaker P. The acquisition unit 21 acquires the speech data 14 of speaker P by having the speaker P speak the content to be converted into the voice conversion model.

[0052] The removal unit 22 removes non-target data 15 that is not a conversion target of the voice quality conversion model from the utterance data 14 of the speaker P. The removal unit 22 identifies, from the utterance data 14, data that is a human sound but is not a speaking voice, such as laughter, groaning, sneezing, etc., as the non-target data 15. The removal unit 22 deletes the non-target data 15 from the utterance data 14 to generate post-removal utterance data 16.

[0053] The removal unit 22 further outputs the non-target data 15 to the voice conversion model of the voice conversion device 3. The non-target data is combined with data converted by the voice conversion model.

[0054] The conversion unit 23 converts the range of the speech data 14 of the speaker P into the range of the voice conversion model to generate speech data 17 after the range conversion. The conversion unit 23 outputs the speech data 17 after the range conversion to the voice conversion model of the voice conversion device 3.

[0055] The conversion unit 23 shifts the range of the speaker P in the speech data 14 to the range of the voice conversion model. For example, the conversion unit 23 calculates the amount of shift from the reference value of the range of the speaker P and the reference value of the range of the voice conversion model. The reference value is, for example, the median value of the range. The conversion unit 23 shifts the range of the speech data 14 according to the calculated amount of shift.

[0056] Consider a case where the range of the conversion target model is C4-C5, and speaker P is male and has a speaker range of C3-E4. Since speaker P's range is lower than the range of the conversion target model, a pitch correction of +12 (one octave) is applied. As a result, the range of speaker P's speech data 14 is converted from C3-E4 to C4-E5.

[0057] In this disclosure, a case will be described in which the shift amount is calculated from the vocal range of the speaker P, but the shift amount may be set by the speaker P.

[0058] The conversion unit 23 compresses the range of the speech data 14 after the shift conversion, which is outside the range of the voice conversion model, into the range of the voice conversion model.

[0059] The range after the shift conversion may deviate from the range of the voice conversion model. In order to fit the range after the shift conversion into the range of the voice conversion model, a certain range outside the range of the voice conversion model is pushed within a certain range above and below the range of the voice conversion model.

[0060] An example of compression by the conversion unit 23 is shown in Figure 6. The range of speaker P after shift conversion is wider in the high and low ranges than the range of the voice conversion model. Therefore, two semitones above and below the range of the voice conversion model are set as the compressed range. The conversion unit 23 compresses the range that falls outside the range of the voice conversion model into the compressed range. Here, the arrangement of each note within the compressed range is processed using a moving average or exponential moving average to smooth out the change in scale.

[0061] The compressed range may be determined based on the number of sounds that fall outside the range of the voice conversion model in the range of the speaker P after the shift conversion. The compressed range may be preset for each of the voice conversion models. The compressed range may be determined by adjusting the range preset for the voice conversion model using the number of sounds that fall outside the range of the voice conversion model.

[0062] When the speech data 14 includes non-target data 15, the conversion unit 23 converts the range of the data obtained by removing the non-target data 15 from the speech data 14 into the range of the voice conversion model to generate speech data 17 after range conversion. The conversion unit 23 outputs the speech data 17 after range conversion to the voice conversion model of the voice conversion device 3.

[0063] In the present disclosure, the conversion unit 23 has been described as compressing the range of the speech data after the shift conversion if the range is outside the range of the voice conversion model, but this is not limited to the above. If the range of the speaker P and the range of the voice conversion model are the same or close, the conversion unit 23 may compress the range of the speaker P in the speech data 14 that is outside the range of the voice conversion model to the range of the voice conversion model without using shift conversion. For example, if the reference value of the range of the speaker P and the reference value of the range of the voice conversion model are the same or within a predetermined value, the conversion unit 23 may only compress the range without shift conversion.

[0064] In the present disclosure, the conversion unit 23 converts the range of the voice, but this is not limited to the above. If the range of the speech data 14 of the speaker P is within the range of the voice quality conversion model, the conversion unit 23 does not need to convert the range of the speech data 14.

[0065] The conversion process performed by the conversion unit 23 will be described with reference to FIG.

[0066] In step S101, the conversion unit 23 acquires the speech data 16 after removal of the non-target data 15. In step S102, the conversion unit 23 shifts the range of the speech data 16 after removal.

[0067] In step S103, the conversion unit 23 determines whether or not there is a range that deviates from the range of the voice quality conversion model among the ranges after the shift conversion in step S102. If there is a deviating range, the process proceeds to step S104. If there is no deviating range, the process proceeds to step S105.

[0068] In step S104, the conversion unit 23 compresses the range outside the range of the voice conversion model to within the range of the voice conversion model.

[0069] The conversion unit 23 generates range-converted speech data 17 by the shift conversion in step S102, or the shift conversion in step S102 and the compression in step S104. In step S105, the conversion unit 23 outputs the range-converted speech data 17 to the voice quality conversion device 3.

[0070] The conversion system 7 according to the present disclosure converts the range of speaker P's speech data into the range of the voice conversion model, and then inputs it into the voice conversion model, so that it can convert the voice into a natural voice quality regardless of the original voice quality.

[0071] (Modification) A conversion system 7a according to a modified example will be described with reference to Fig. 8. The conversion system 7a shown in Fig. 8 differs from the conversion system 7 shown in Fig. 1 in that a practice device 5 is provided.

[0072] In general, voice conversion models can be difficult for beginners to use because they do not know whether the microphone is turned on, whether they are speaking at an appropriate volume, what pitch they should speak at to convert to a natural voice, and what speaking style will sound natural to the target character.

[0073] Therefore, the training device 5 supports the training of the speaker P so that the speaker P can master the use of the voice quality conversion model.

[0074] In the conversion system 7a, the conversion device 1 converts the vocal range of the speaker P into the vocal range of the character of the voice conversion model, and the voice conversion device 3 converts the utterance converted into the vocal range of the character into the voice quality of the character. The practice device 5 is used by the speaker P to practice the utterance of the character other than the vocal range and voice quality of the character.

[0075] The practice device 5 includes each of the data, ie, line data 51, personality data 52, speech data 53, and vocal range data 54, and each of the functions of a display unit 61, an acquisition unit 62, a playback unit 63, and an identification unit 64. Each of the data is stored in a storage device, such as a memory 902 or a storage 903. Each of the functions is implemented in a CPU 901.

[0076] The dialogue data 51 specifies characteristic lines of a character used in the conversion of the voice conversion model. The dialogue data 51 includes one or more lines. The dialogue data 51 may include symbols specifying intonations in speech, such as accents and intonations, when the character speaks the lines.

[0077] Personality data 52 identifies a character's traits or attributes. Personality data 52 identifies traits or attributes that influence a character's speech. Personality data 52 may be data that includes a character's traits, attributes, or attributes, or may be data that identifies an event that depicts a character's traits or attributes.

[0078] The speech data 53 is data of an utterance by a speaker P.

[0079] The range data 54 is data for identifying the range of the speaker P. The range data 54 is, for example, data shown in FIG.

[0080] The display unit 61 displays a message to the speaker P, who is the source of the voice quality conversion model, encouraging him / her to speak the dialogue data 51. The display unit 61 further displays the personality data 52. The display unit 61 can encourage the speaker P to speak the character's dialogue after understanding the character's personality or attributes. The conversion device 1 and the conversion unit 23 convert the range and voice quality of the speaker P into the range and voice quality of the character. The conversion system 7a makes it easier to convert the speech of the speaker P into the speech of the character.

[0081] The acquisition unit 62 acquires the speaker P's speech data 53.

[0082] The playback unit 63 plays the acquired speech data 53 of the speaker P. By playing back the speaker's speech data 14, the speaker P can check the speaker P's speech and practice to make it closer to the character's speech.

[0083] The playback unit 63 may play back the speech data 53 of the speaker P, as well as the speech data converted from the speech data 14 to the character's vocal range and voice quality by the conversion device 1 and the voice quality conversion device 3. The playback unit 63 may further play back the speech of the character of the voice quality conversion model.

[0084] The playback unit 63 may repeatedly play back the speech data 53 by the speaker P, the speech data whose range and voice quality have been converted from the speech data 14, and the character's speech. The speaker P can compare each speech and analyze it to make it closer to the character's speech.

[0085] The playback unit 63 may not only play back each piece of speech data, but may also analyze speaker P's speech data 14 to analyze characteristics such as speaking speed, voice volume, intonation, etc. The playback unit 63 may display differences from the character's speech characteristics as items to be corrected. Furthermore, when the characteristics of speaker P's speech data 14 and the character's speech characteristics are the same or within a predetermined value, the playback unit 63 may play back audio data in which the character praises the speaker's speech.

[0086] The specification unit 64 specifies the range of the speaker P from the speech data 53 and generates range data 54. The range data 54 is transmitted to the conversion device 1 and is referred to when converting the range of the speech data of the speaker P. The specification unit 64 may also specify the amount of shift when shifting the range and share the specified amount of shift with the conversion device 1. The specification unit 64 may also specify the compressed range when compressing the range and share the specified compressed range with the conversion device 1. The specification unit 64 may also specify parameters for range conversion or voice conversion and share them with the conversion device 1 or the voice conversion device 3.

[0087] The determination unit 64 may determine the range and also the microphone gain suitable for the speaker P, and set the gain to the determined gain. The determination unit 64 may analyze the speech data 14 of the speaker P, and determine the ratio of mixing the voice of the speaker P with the voice after the range and voice quality conversion, and the amount of noise cancellation. The determined ratio and the amount of noise cancellation are shared with and referred to by the conversion device 1 or the voice quality conversion device 3.

[0088] When the speaker P selects a voice conversion model to convert his / her voice, the practice device 5 displays lines that match the selected voice conversion model and plays model voice data in which the lines are read in the converted voice quality. The range and voice quality of the speaker P's utterance reading the lines are converted and recorded, and the recorded voice is played back. The practice device 5 may provide an opportunity to repeat such practice multiple times.

[0089] With reference to FIG. 10, a practice method according to the present disclosure will be described.

[0090] In step S51, the practice device 5 displays the line data 51. In step S52, the practice device 5 displays the personality data 52. At this time, the practice device 5 may play back the voice of the character speaking the lines of the line data 51.

[0091] In step S53, the practice device 5 acquires the speech data 14 of the speaker P. In step S54, the practice device 5 may play back the speech data 14. The practice device 5 may further play back the voice of the character speaking the lines of the line data 51. The practice device 5 may simultaneously play back the speech data 53 of the speaker P and the voice of the character. The practice device 5 may play back the speech data 53 of the speaker P and then the voice of the character, or may play back the speech data 53 of the speaker P and then the voice of the character.

[0092] The training device 5 also identifies the range of the speaker P from the speech data 53. The identification unit 64 may identify parameters used for converting the range or voice quality in addition to the range of the speaker P. The training device 5 transmits the range data for identifying the range and the parameters to the conversion device 1 or the voice quality conversion device.

[0093] The training device 5 according to such a modified example provides an opportunity for the speaker P to practice so that the speech data 53 of the speaker P becomes closer to the speech of the character. The speaker P can become accustomed to using the voice conversion model and practice so that the speech becomes closer to the speech of the character. The training device 5 can contribute to the spread of the voice conversion model.

[0094] (Technology A) a storage device that stores speech data that identifies characteristic speech of a character to be used in converting a voice conversion model; A display unit that displays a message encouraging a source speaker of the voice conversion model to speak the line data. A training device comprising:

[0095] (Technology B) The display unit further displays personality data that specifies the character's personality or attributes. A practice device as described in Technique A.

[0096] Each of the above-described conversion device 1, voice quality conversion device 3 and practice device 5 of the present embodiment is implemented, for example, by a general-purpose computer system including a CPU (Central Processing Unit, processor) 901, a memory 902, a storage 903 (HDD: Hard Disk Drive, SSD: Solid State Drive), a communication device 904, an input device 905 and an output device 906. In this computer system, the CPU 901 executes a program loaded on the memory 902, thereby realizing each function of each device.

[0097] Each of the devices may be implemented in one computer, or in multiple computers, or each of the devices may be a virtual machine implemented in a computer.

[0098] The programs of each device can be stored on a computer-readable recording medium such as a HDD, SSD, USB (Universal Serial Bus) memory, CD (Compact Disc), or DVD (Digital Versatile Disc), or can be distributed via a network.

[0099] Any part or all of the functional units described in this disclosure may be realized by a program. The programs mentioned in this disclosure may be non-temporarily recorded on a computer-readable recording medium and distributed, may be distributed via a communication line (including wireless communication) such as the Internet, or may be distributed in a state where they are installed on any terminal. Based on the above description, a person skilled in the art may be able to conceive additional effects and various modifications of the present invention, but the aspects of this disclosure are not limited to the individual embodiments described above. Various additions, modifications, and partial deletions are possible within the scope of the conceptual idea and intent of the present invention derived from the contents defined in the claims and their equivalents. For example, what is described in this disclosure as one device (or member, the same applies below) (including what is depicted as one device in the drawings) may be realized by multiple devices. Conversely, what is described in this disclosure as multiple devices (including what is depicted as multiple devices in the drawings) may be realized by one device. Alternatively, some or all of the means and functions included in a certain device (e.g., a server) may be included in another device (e.g., a user terminal). Furthermore, a "system" may be composed of one device, or may be composed of two or more devices (for example, a server and a user terminal, or multiple user terminals).

[0100] Furthermore, not all of the features described in this disclosure are essential requirements. In particular, features described in this disclosure but not described in the claims can be considered optional additional features.

[0101] It should be noted that the applicant is only aware of the inventions disclosed in the documents in the "Prior Art Documents" column of this disclosure, and that this disclosure does not necessarily aim to solve the problems in the disclosed inventions. The problems that this disclosure aims to solve should be identified by taking into consideration the entire disclosure. For example, if this disclosure describes that a specific configuration produces a specific effect, it can also be said that the problem that is the reverse of the specific effect is solved. However, it is not necessarily intended that such a specific configuration be a required requirement. [Explanation of symbols]

[0102] 1. Conversion device 3. Voice conversion device 5 Exercise equipment 7. Conversion System 11, 54 range data 12 Character range data 13 Character Identifiers 14, 53 Speech data 15 Excluded Data 16 Speech data after removal 17 Speech data after range conversion 21, 62 Acquisition Department 22 Removal part 23 Conversion section 51 Dialogue data 52 Personality Data 61 Display section 63 Playback Department 64 Specific part 901 CPU 902 Memory 903 Storage 904 Communication equipment 905 Input Device 906 Output Device P Speaker

Claims

1. An acquisition unit for acquiring the range of a voice conversion model and the range of a speaker; a conversion unit that converts the range of a speaker's speech data into the range of the voice conversion model and outputs the speech data after the range conversion to the voice conversion model; Equipped with The conversion unit is a conversion device that compresses a range of the speech data after shift conversion that falls outside the range of the voice conversion model into the range of the voice conversion model.

2. An acquisition unit for acquiring the range of a voice conversion model and the range of a speaker; a conversion unit that converts the range of a speaker's speech data into the range of the voice conversion model and outputs the speech data after the range conversion to the voice conversion model; Equipped with The conversion unit is a conversion device that compresses a vocal range of the speaker in the speech data that falls outside the vocal range of the voice conversion model into the vocal range of the voice conversion model.

3. The conversion unit shifts and converts the range of the speaker in the speech data to the range of the voice conversion model.

3. The conversion device according to claim 1 or 2.

4. a removal unit that removes non-target data that is not a conversion target of the voice quality conversion model from the speech data of the speaker; The conversion unit converts a range of the data obtained by removing the non-target data from the speech data into a range of the voice conversion model.

3. The conversion device according to claim 1 or 2.

5. The removal unit outputs the non-target data to the voice conversion model; The non-target data is synthesized with the data converted by the voice conversion model. The conversion device according to claim 4.

6. The computer Obtain the range of the voice conversion model and the speaker's range, converting the range of the speaker's speech data into the range of the voice conversion model, and outputting the speech data after the range conversion to the voice conversion model; The computer, Among the range of the speech data after the shift conversion, the range outside the range of the voice conversion model is compressed to the range of the voice conversion model. How to convert.

7. A computer comprising: Obtain the range of the voice conversion model and the speaker's range, converting the range of the speaker's speech data into the range of the voice conversion model, and outputting the speech data after the range conversion to the voice conversion model; The computer, Compressing a range of the speaker in the speech data that is outside the range of the voice conversion model into the range of the voice conversion model. How to convert.

8. Computer, An acquisition unit for acquiring the range of a voice conversion model and the range of a speaker; A conversion unit that converts the range of the speaker's speech data into the range of the voice conversion model and outputs the speech data after the range conversion to the voice conversion model. Function as a The conversion unit is a program that compresses a range of the speech data after shift conversion that falls outside the range of the voice conversion model into the range of the voice conversion model.

9. A computer comprising: An acquisition unit for acquiring the range of a voice conversion model and the range of a speaker; A conversion unit that converts the range of the speaker's speech data into the range of the voice conversion model and outputs the speech data after the range conversion to the voice conversion model. Function as a The conversion unit is a program that compresses a vocal range of the speaker in the speech data that falls outside the vocal range of the voice conversion model into the vocal range of the voice conversion model.

Images