JP2007226174A - Singing synthesizer, singing synthesizing method, and program for singing synthesis - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a singing synthesizer etc., that obtain a more natural synthesized singing voice. <P>SOLUTION: Performance data are sectioned into a transition part and a lengthened voice part and phoneme chain data from a phoneme chain template database 52 are used as they are for the transition part. For the lengthened voice part, feature parameters of transition parts disposed at both sides of the lengthened voice part are linearly interpolated and a variation component included in stationary part data from a stationary part template database is added to the interpolated feature parameter string to generate feature parameters. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a singing voice synthesizing device, a singing voice synthesis method, and a singing voice synthesis program for synthesizing human singing voice.

  In a conventional singing voice synthesizing apparatus, data acquired from an actual human singing voice is stored as a database, and data matching the contents of inputted performance data (notes, lyrics, facial expressions, etc.) is selected from the database. Then, the performance data is converted based on the selected data, thereby synthesizing a singing voice close to a real person's singing voice.

   However, in the conventional synthesizer, even if, for example, “saita” is sung, the phoneme does not move naturally between phonemes, and the synthesized singing voice is unnatural. There was a sound, and in some cases it was impossible to determine what was being sung.

  The present invention aims to solve this problem and has been made paying attention to the following points. That is, in the singing voice, for example, even when singing “saita”, individual phonemes (“sa”, “i”, “ta”) are not pronounced separately, but “[## s] sa (a), [ai], i, (i), [it], ta, (a) "(# represents silence), and the extended sound part and the transition part are inserted between each phoneme. Usually, pronunciation is made. In the example of “saita”, [#s] [ai] and [it] are transition parts, and (a), (i), and (a) are extended sound parts. Thus, the singing sound is composed of a transition portion and a stretched sound portion. For this reason, even when synthesizing a singing voice from performance data such as MIDI information, it is important how to generate the transition portion and the extended sound portion as genuine. Therefore, the present inventors have considered that it is necessary to reproduce this transition portion naturally in order to output a natural synthesized song, and have come to the present invention.

  The singing voice synthesizing device according to the first invention of the present application includes a storage unit that stores singing information for synthesizing a song, and a transition part including a phoneme chain that transfers singing data from one phoneme to another phoneme. A phonological database for storing the phoneme chain data of the transition portion and the steady portion data of the extended sound portion, distinguished from the extended sound portion including the steady portion in which one phoneme is stably pronounced, and the singing information A selection unit that selects data stored in the phoneme database, a transition part feature parameter output unit that extracts and outputs a feature parameter of the transition part from the phoneme chain data selected by the selection unit, The phoneme chain data preceding the extended sound part related to the stationary part data selected by the selection unit, and the phoneme chain data of the transition part following the extended sound part. Get the door, by interpolating the two phoneme data, characterized in that a sound part, wherein the parameter output unit stretched generates and outputs the feature parameters of the long sound part.

  The extended sound part feature parameter output unit generates and outputs a feature parameter of the extended sound part by adding a fluctuation component of the steady part data to an interpolation value obtained by storing and acquiring the two phoneme chain data. It can be constituted as follows. The phoneme chain data in the phoneme database includes feature parameters and anharmonic components related to the phoneme chain, and the transition partial feature parameter output unit can be configured to separate the anharmonic components. Similarly, the stationary part data in the phonological database includes feature parameters and anharmonic components related to the stationary part, and the extended sound part feature parameter output unit is configured to separate the anharmonic components. Can do. The characteristic parameter and the anharmonic component can be obtained by, for example, SMS analysis of speech.

  In the first invention, the singing information includes dynamics information, and further includes characteristic parameter correcting means for correcting the characteristic parameter of the transition part and the characteristic parameter of the extended sound part based on the dynamics information. it can. In this case, the singing information includes pitch information, and the feature parameter correction means includes at least first amplitude calculation means for calculating an amplitude value corresponding to the dynamics, and a feature parameter of the transition portion or a feature of the extended sound portion. A second amplitude calculating means for calculating an amplitude value corresponding to the parameter and the pitch information, and correcting the characteristic parameter based on a difference between an output of the first amplitude calculating means and an output of the second amplitude calculating means. Can be. Furthermore, the first amplitude calculation means may include a table that stores the dynamics and the amplitude value in association with each other. In addition, the table may be configured such that the correspondence between the dynamics and the amplitude value is different for each phoneme. Further, the table may be configured such that a correspondence relationship between the dynamics and the amplitude value is different for each frequency.

  In the first invention, the phoneme database stores phoneme chain data and the stationary part data in association with pitches, and the selection unit changes the characteristic parameters of the same phoneme chain for each pitch. The selection unit may select the corresponding phoneme chain data and the stationary partial data based on the input pitch information. In the first aspect of the invention, the phoneme database stores facial expression data in addition to the phoneme chain data and the stationary part data, and the selection unit is configured to input the facial expression information in the input singing information. The facial expression data may be selected based on the above.

  The singing synthesis method according to the second invention of the present application includes a transition part including a phoneme chain for transferring singing data from one phoneme to another and a stationary part in which one phoneme is stably generated. A step of storing the phoneme chain data of the transition portion and the steady portion data of the extended portion, an input step of inputting singing information for synthesizing the song, and the singing information A selection step of selecting the phoneme chain data or the stationary part data, and a transition part feature parameter output step of extracting and outputting a feature parameter of the transition part from the phoneme chain data selected in the selection step; The phoneme chain data of the transition part preceding the extended sound part related to the stationary part data selected in the selection step, and the extension An extended sound part feature parameter output step of obtaining the phoneme chain data of the transition part following the part and generating the feature parameters of the extended sound part by interpolating the two phoneme chain data And

  In the second invention, the extended sound part feature parameter output step adds the fluctuation component of the steady part data to an interpolated value obtained by interpolating the two phoneme chain data. Parameters can be generated and output. In the second invention, the singing information includes dynamics information, and further includes a characteristic parameter correction step of correcting the characteristic parameter of the transition part and the characteristic parameter of the extended sound part based on the dynamics information. be able to.

  Note that the singing synthesis method according to the second invention may be executed by a computer by a computer program.

(Principle of the present invention)
The principle of the present invention will be described using FIG. 7 and FIG. 8 by comparing with the singing synthesizer (Japanese Patent Application No. 2001-67258) previously filed by the present applicant. FIG. 7 shows the principle of the singing voice synthesizing apparatus described in Japanese Patent Application No. 2001-67258. This singing voice synthesizing apparatus includes a Timbre template database 51 that stores phonological feature parameter data (Timbre template) at one point in time as a database, and minute change (fluctuation) data of feature parameters in the extended sound ( A stationary part template database 53 in which stationary parts (stationary templates) are stored, and a phoneme chain template database 52 in which data (phonemic chain (articulation) template) indicating changes in the characteristic parameters of the transition part from phonemes to phonemes is stored. And. The feature parameters are generated by applying these templates as follows.

  That is, the synthesis of the extended sound part is performed by adding the variation included in the stationary part template to the feature parameter obtained from the Timbre template. On the other hand, the transition part is synthesized by adding the variation included in the phoneme chain template to the feature parameter, but the feature parameter to be added differs depending on the case. For example, if the phonemes before and after the transition part are both voiced sounds, the variation included in the phoneme chain template is obtained by linearly interpolating the feature parameters of the front phoneme and the feature parameters of the rear phoneme. to add. When the front phoneme is voiced and the rear phoneme is silent, the variation included in the phoneme chain template is added to the feature parameter of the front phoneme. Further, when the front phoneme is silent and the rear phoneme is voiced, the variation included in the phoneme chain template is added to the feature parameter of the rear phoneme. As described above, in the apparatus disclosed in Japanese Patent Application No. 2001-67258, the feature parameter generated from the Timbre template is used as a reference, and the feature parameter of the phoneme chain portion is changed to match the feature parameter of the Timbre portion. Singing was performed.

  In the device disclosed in Japanese Patent Application No. 2001-67258, unnaturalness may occur in the synthesized singing voice. The reason is as follows.・ Because a change is made to the phoneme chain template, it is different from the change of the characteristic parameter that the transition part originally has. The feature parameter of the extended sound part is also calculated based on the feature parameter generated from the Timbre template, and the variation of the stationary part template is added to the feature parameter of this Timbre template. The phoneme of any phoneme was the same phoneme. In short, in the device of this Japanese Patent Application 2001-67258, since the characteristic parameters of the Timbre template, which is only a part of the entire singing, are combined with the characteristic parameters of the extended sound part and the transition part, it is synthesized. There was a case where the singing was unnatural.

  On the other hand, in the present invention, as shown in FIG. 8, only the phoneme chain template database 52 and the stationary partial template database 53 are used, and the Timbre template is basically unnecessary. Then, after the performance data is divided into the transition portion and the extended sound portion, the phoneme chain template is used as it is in the transition portion. For this reason, the singing of the transition part which occupies the important part of a song can be heard naturally, and the quality of a synthetic song is increasing. In addition, for the extended sound part, the characteristic parameter of the transition part located on both sides of the extended sound part is linearly interpolated, and the characteristic parameter is added by adding the fluctuation component included in the stationary part template to the interpolated characteristic parameter string. Is generated. Since the interpolation is performed based on the data as it is without converting the template, unnatural singing does not occur.

  According to the present invention, the naturalness of the synthetic singing voice of the transition portion is kept high, and thereby the naturalness of the synthetic singing voice can be enhanced.

[First Embodiment]
FIG. 1 is a functional block diagram showing the configuration of the singing voice synthesizing apparatus according to the first embodiment. The singing voice synthesizing apparatus can be realized by, for example, a general personal computer, and the function of each block shown in FIG. 1 can be achieved by a CPU, RAM, ROM, etc. in the personal computer. It can also be configured by a DSP or a logic circuit. The phoneme database 10 holds data for synthesizing synthesized sounds based on performance data. An example of creating the phoneme database 10 will be described with reference to FIG.

  First, as shown in FIG. 2A, an audio signal such as singing data actually recorded or acquired is separated into a harmonic component (sinusoidal component) and an anharmonic component by SMS (spectral modeling synthesis) analysis means 31. Instead of the SMS analysis, another analysis method such as LPC (Linear Predictive Coding) may be used. Next, the phoneme segmenting means 32 segments the speech signal for each phoneme based on the phoneme segmentation information. The phoneme segmentation information is usually given by, for example, a human performing a predetermined switch operation while viewing the waveform of the audio signal.

  Then, feature parameters are extracted by the feature parameter extraction means 33 from the harmonic components of the audio signal cut out for each phoneme. The characteristic parameters include an excitation waveform envelope, excitation resonance, formant frequency, formant bandwidth, formant intensity, difference spectrum, and the like.

The excitation waveform envelope (ExcitationCurve) is the Egain that indicates the size (dB) of the vocal cord waveform, the EslopeDepth that indicates the slope of the spectrum envelope of the vocal cord waveform, and the depth (dB) from the maximum value to the minimum value of the spectral envelope of the vocal cord waveform. It consists of three parameters of Eslope to be expressed, and can be expressed by the following formula [Equation 1].
(Formula 1) Excitation Curve (f) = Egain + EslopeDepth * (exp (-Eslope * f) -1)
Excited resonance represents resonance by the chest. It consists of three parameters: center frequency (ERFreq), bandwidth (ERBW), and amplitude (ERAmp), and has secondary filter characteristics.

  Formants represent resonances due to the vocal tract by combining 1 to 12 resonances. It consists of three parameters: center frequency (FormantFreqi, i is an integer from 1 to 12), bandwidth (FormantBWi, i is an integer from 1 to 12), and amplitude (FormantAmpi, i is an integer from 1 to 12).

  The difference spectrum is a characteristic parameter having a spectrum of the difference from the original harmonic component that cannot be expressed by the above-described excitation waveform envelope, excitation resonance, and formant.

  This feature parameter is stored in the phoneme database 10 in association with the phoneme name. Similarly, the anharmonic component is stored in the phoneme database 10 in association with the phoneme name. In the phoneme database 10, as shown in FIG. 2B, phoneme chain data and stationary partial data are stored separately. Hereinafter, the phoneme chain data and the stationary partial data are collectively referred to as “speech segment data”.

  The phoneme chain data is a data string in which the head phoneme name, the subsequent phoneme name, the feature parameter, and the anharmonic component are associated with each other. On the other hand, the stationary partial data is a data string in which one phoneme name, a characteristic parameter string, and an anharmonic component are associated with each other.

  Returning to FIG. 1, reference numeral 11 denotes a performance data holding unit for holding performance data. The performance data is MIDI information including information such as notes, lyrics, pitch bends, dynamics, and the like. The speech element selection unit 12 accepts input of performance data held in the performance data holding unit 11 in units of frames (hereinafter, this one unit is referred to as frame data) and corresponds to lyrics data in the input frame data. A speech segment data to be selected from the phoneme database 10 and read out.

  The preceding phoneme chain data holding unit 13 and the rear phoneme chain data holding unit 14 are used for processing the stationary partial data. The preceding phoneme chain data holding unit 13 holds phoneme chain data preceding the stationary partial data to be processed, while the rear phoneme chain data holding unit 14 is a phoneme chain behind the stationary partial data to be processed. Holds data.

  The feature parameter interpolation unit 15 includes the feature parameter of the last frame of the phoneme chain data held in the preceding phoneme chain data holding unit 13 and the feature parameter of the first frame of the phoneme chain data held in the rear phoneme chain data holding unit 14. And the characteristic parameters are interpolated in time so as to correspond to the time indicated by the timer 27.

  The steady part data holding unit 16 temporarily holds the steady part data of the speech unit data read by the speech unit selection unit 12. On the other hand, the phoneme chain data holding unit 17 temporarily holds phoneme chain data.

  The feature parameter fluctuation extracting unit 18 has a function of reading out the stationary part data held in the stationary part data holding unit 16, extracting the fluctuation (fluctuation) of the characteristic parameter, and outputting it as a fluctuation component. The addition unit K1 is a part that adds the output of the feature parameter interpolation unit 15 and the output of the feature parameter fluctuation extraction unit 18 and outputs the harmonic component data of the extended sound part. The frame reading unit 19 is a part that reads the phoneme chain data held in the phoneme chain data holding unit 17 as frame data in accordance with the time indicated by the timer 27, and outputs it as feature parameters and anharmonic components.

  The pitch determining unit 20 is a part that determines the pitch of the synthesized sound to be finally synthesized based on the note data in the frame data. Further, the feature parameter correction unit 21 calculates the feature parameter of the extended sound portion output from the adder K1 and the feature parameter of the transition portion output from the frame reading unit 19 based on the dynamics information included in the performance data. This is the part to be corrected. A switch SW1 is provided in the preceding stage of the feature parameter correction unit 21, and the feature parameter of the extended sound portion and the feature parameter of the transition portion are selectively input to the feature parameter correction unit. Detailed processing contents in the feature parameter correction unit 21 will be described later. The switch SW2 switches and outputs the anharmonic component of the extended sound portion read from the steady part data holding unit 16 and the anharmonic component of the transition portion read from the frame reading unit 19.

  The harmonic sequence generator 22 is a portion that generates a harmonic sequence on the frequency axis for performing formant synthesis in accordance with the determined pitch. The spectrum envelope generation unit 23 is a part that generates a spectrum envelope in accordance with the corrected feature parameter corrected by the feature parameter correction unit 21.

  The overtone amplitude / phase calculation unit 24 is a part that calculates the amplitude and phase of each overtone generated by the overtone string generation unit 22 in accordance with the spectrum envelope generated by the spectrum envelope generation unit 23. The adder K2 adds the harmonic component as the output of the harmonic overtone amplitude / phase calculation unit 24 and the anharmonic component output from the switch SW2. The inverse FFT unit 25 performs inverse fast Fourier transform on the output value of the adder K2, and converts a signal that is a frequency expression into a signal that is a time axis expression. The superimposing unit 26 outputs synthesized singing voice by superimposing signals obtained one after another on the lyrics data processed in time series order in a form along the time series.

  Details of the characteristic parameter correction unit 21 will be described with reference to FIG. The feature parameter correction unit 21 includes amplitude determination means 41. The amplitude determination means 41 outputs a desired amplitude value A1 corresponding to the dynamics information input from the performance data holding unit 11 with reference to the dynamics-amplitude conversion table Tda. The spectrum envelope generation means 42 is a part that generates a spectrum envelope based on the feature parameter output from the switch SW1.

  The harmonic string generation unit 43 generates a harmonic string based on the pitch determined by the pitch determination unit 20. The amplitude calculation means 44 calculates the amplitude A2 corresponding to the generated spectral envelope and overtone. The calculation of the amplitude can be executed by, for example, inverse FFT. The adder K3 outputs the difference between the desired amplitude value A1 determined by the amplitude determining means 41 and the amplitude value A2 calculated by the amplitude calculating means 44. Based on this difference, the gain correction unit 45 calculates a correction amount of the amplitude value, and corrects the characteristic parameter according to the gain correction amount. Thereby, a new feature parameter matching the desired amplitude is obtained. In FIG. 3, the amplitude is determined based only on the dynamics based on the table Tda. However, in addition to this, a table may be adopted in which the amplitude is determined in consideration of the type of phoneme. That is, even if the dynamics are the same, a table that gives different amplitude values may be adopted when phonemes are different. Similarly, a table that determines the amplitude in consideration of the frequency in addition to the dynamics may be employed.

  Next, the operation of the singing voice synthesizing apparatus according to the first embodiment will be described with reference to the flowchart shown in FIG. The performance data holding unit 11 outputs frame data in chronological order. Transition portions and extended sound portions appear alternately, and the transition portion and the extended sound portion are processed differently.

  When frame data is input from the performance data holding unit 11 (S1), the speech segment selection unit 12 determines whether the frame data relates to the extended sound part or the phoneme transition part (S2). If it is the extended sound part (YES), the preceding phoneme chain data holding part 13, the rear phoneme chain data holding part 14 and the stationary part data holding part 16 are respectively preceded by the preceding phoneme chain data, the rear phoneme chain data, and the stationary part data. Is transferred (S3).

  Subsequently, the feature parameter interpolation unit 15 extracts the feature parameter of the last frame of the preceding phoneme chain data held in the preceding phoneme chain data holding unit 13 and the rear phoneme chain data held in the rear phoneme chain data holding unit 14. The feature parameters of the first frame are extracted, and the feature parameters of the extended sound portion being processed are generated by linearly interpolating these two feature parameters (S4).

  The feature parameter of the steady part data held in the steady part data holding unit 16 is supplied to the feature parameter fluctuation extracting unit 18, and the fluctuation component of the feature parameter of the steady part is extracted (S5). This fluctuation component is added to the feature parameter output from the feature parameter interpolation unit 15 in the adder K1 (S6). This added value is output to the feature parameter correction unit 21 via the switch SW1 as the feature parameter of the extended sound portion, and the feature parameter is corrected (S9). On the other hand, the anharmonic component of the steady part data held in the steady part data holding unit 16 is supplied to the adder K2 via the switch SW2. The spectrum envelope generation unit 23 generates a spectrum envelope for the corrected characteristic parameter. The harmonic amplitude / phase calculation unit 24 calculates the amplitude and phase of each harmonic generated by the harmonic string generation unit 22 in accordance with the spectrum envelope generated by the spectrum envelope generation unit 23. This calculation result is output to the adder K2 as a parameter string (harmonic component) of the extended sound part being processed.

  On the other hand, if it is determined in S2 that the acquired frame data is of the transition part (NO), the phoneme chain data of the transition part is held by the phoneme chain data holding unit 17 (S7). . Next, the frame reading unit 19 reads the phoneme chain data held in the phoneme chain data holding unit 17 as frame data in accordance with the time indicated by the timer 27, and outputs it as feature parameters and anharmonic components. The characteristic parameter is output toward the characteristic parameter correction unit 21, and the anharmonic component is output toward the adder K2. The characteristic parameter of the transition unit is subjected to the same processing as the above-described extended sound characteristic parameter by the characteristic parameter correction unit 21, the spectrum envelope generation unit 23, the overtone amplitude / phase calculation unit 24, and the like.

  Since the switches SW1 and SW2 are switched depending on the type of data being processed, the switch SW1 has the characteristic parameter correction unit 21 in the adder K1 while processing the extended sound portion. And the feature parameter correction unit 21 is connected to the frame reading unit 19 while the transition portion is processed. As for the switch SW2, the adder K2 is connected to the stationary part data holding unit 16 while the extended sound part is being processed, and the frame reading unit is being processed while the transition part is being processed. The adder K2 is connected to 19. When the characteristic parameter and the anharmonic component of the transition part and the extended sound part are calculated in this way, the added value is processed by the inverse FFT unit 25 and is superposed by the superimposing means 26, and a final synthesized waveform is output. (S10).

[Second Embodiment]
A singing voice synthesizing apparatus according to a second embodiment of the present invention will be described with reference to FIG. FIG. 5 is a functional block diagram of the singing voice synthesizing apparatus according to the second embodiment. Portions common to the first embodiment are denoted by the same reference numerals and description thereof is omitted. One of the differences from the first embodiment is that the phoneme chain data and the stationary partial data stored in the phoneme database are assigned different characteristic parameters and anharmonic components for each different pitch (pitch). It is that. The pitch determination unit 20 determines the pitch based on the note information in the performance data, and outputs the result to the speech unit selection unit.

  The operation of the second embodiment will be described. Based on the note information from the performance data holding unit 11, the pitch determining unit 20 determines the pitch of the frame data being processed, and the result is the speech unit selecting unit. 12 is output. The speech segment selection unit 12 reads the phoneme chain data and the stationary partial data closest to the phoneme information in the determined pitch and lyrics information. Subsequent processing is the same as in the first embodiment.

[Third Embodiment]
A singing voice synthesizing apparatus according to a third embodiment of the present invention will be described with reference to FIG. FIG. 6 is a functional block diagram of a singing voice synthesizing apparatus according to the third embodiment. Portions common to the first embodiment are denoted by the same reference numerals and description thereof is omitted. One of the differences from the first embodiment is that, in addition to the phoneme database 10, an expression database 30 storing vibrato information and the like, and an appropriate vibrato template from the expression database based on the expression information in the performance data. The point is that it includes an expression template selection unit 30A for selecting. The pitch determination unit 20 determines the pitch based on the note information in the performance data and the vibrato data from the expression template selection unit 30A.

  The operation of the third embodiment will be described. The phoneme chain data and the stationary partial data are read from the phoneme database 10 by the phoneme segment selector 12 based on the lyric information from the performance data holding unit 11. 1 is the same as that of the first embodiment, and the subsequent processing is also the same as that of the first embodiment. On the other hand, based on facial expression information from the performance data holding unit 11, the facial expression template selection unit 30 </ b> A reads the most suitable vibrato data from the facial expression database 30. The pitch determination unit 20 determines the pitch based on the read vibrato data and the note information in the performance data.

  Although the present invention has been described with reference to the embodiments, the present invention is not limited to these embodiments, and it is obvious to those skilled in the art that various modifications, improvements, combinations, and the like are possible.

It is a functional block diagram of the song synthesizing | combining apparatus which concerns on the 1st Embodiment of this invention. An example of creating the phoneme database 10 shown in FIG. 1 is shown. Details of the feature parameter correction unit 21 shown in FIG. It is a flowchart which shows the procedure of the data processing in the song synthesizing | combining apparatus which concerns on 1st Embodiment. It is a functional block diagram of the song synthesizing | combining apparatus which concerns on the 2nd Embodiment of this invention. It is a functional block diagram of the song synthesizing | combining apparatus which concerns on the 3rd Embodiment of this invention. The principle of the singing voice synthesizing apparatus described in Japanese Patent Application No. 2001-67258 will be described. The principle of the song synthesizing apparatus according to the present invention is shown.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Phoneme database, 11 ... Performance data holding part, 12 ... Speech segment selection part, 13 ... Precede phoneme chain data holding part, 14 ... Back phoneme chain data holding part, 15 ... Feature parameter interpolation part, 16 ... Steady partial data Holding unit, 17 ... Phoneme chain data holding unit, 18 ... Feature parameter fluctuation extracting unit, 19 ... Frame reading unit, K1, K2 ... Adder, 20 ... Pitch determining unit, 21 ... Feature parameter correcting unit, 22 ... Overtone sequence generation 23: Spectral envelope generation unit 24: Overtone amplitude / phase calculation unit 25: Inverse FFT unit 26: Superposition unit 27: Timer 31: SMS analysis unit 32: Phoneme segmentation unit 33: Feature parameter Extraction means 41... Amplitude determination means 43. Harmonic string generation means 44. Amplitude calculation means K 3 Adder 45. Gain correction section 30 Expression data Table 30A ... Expression template selection unit 51 ... Timbre database 52 ... Phoneme chain template database 53 ... Stationary partial template database

Claims (3)

A storage unit for storing song information for synthesizing a song;
The singing data is distinguished from a transition part including a phoneme chain that transitions from one phoneme to another and an extended sound part including a stationary part where one phoneme is stably generated, and the phoneme of this transition part is distinguished. A phonological database that stores chain data and stationary partial data of the extended sound part;
A selection unit that selects data stored in the phonological database based on the singing information;
A transition part feature parameter output unit that extracts and outputs a feature parameter of the transition part from the phoneme chain data selected by the selection unit;
The phoneme chain data of the transition part preceding the extended sound part related to the stationary part data selected by the selection unit and the phoneme chain data of the transition part following the extended sound part are acquired, and this 2 A singing voice synthesizing apparatus comprising: an extended sound part feature parameter output unit that interpolates two phoneme chain data to generate and output a characteristic parameter of the extended sound part. The singing data is distinguished from a transition part including a phoneme chain that transitions from one phoneme to another and an extended sound part including a stationary part where one phoneme is stably generated, and the phoneme of this transition part is distinguished. Storing the chain data and the stationary part data of the extended sound part;
An input step of inputting song information for synthesizing the song;
A selection step for selecting the phoneme chain data or the stationary partial data based on the singing information;
A transition part feature parameter output step for extracting and outputting a feature parameter of the transition part from the phoneme chain data selected in the selection step;
Obtaining the phoneme chain data of the transition part preceding the extended sound part of the stationary part data selected in the selection step and the phoneme chain data of the transition part following the extended sound part; An extended sound part feature parameter output step of interpolating two phoneme chain data to generate a feature parameter of the extended sound part. The singing data is distinguished from a transition part including a phoneme chain that transitions from one phoneme to another and an extended sound part including a stationary part where one phoneme is stably generated, and the phoneme of this transition part is distinguished. Storing the chain data and the stationary part data of the extended sound part;
An input step for inputting singing information including at least note information and lyrics information;
A selection step for selecting the phoneme chain data or the stationary partial data based on the singing information;
A transition part feature parameter generation step for extracting and outputting a feature parameter of the transition part from the phoneme chain data selected in the selection step;
Obtaining the phoneme chain data of the transition part preceding the extended sound part of the stationary part data selected in the selection step, and the phoneme chain data of the transition part following the extended sound part, A song synthesis program configured to cause a computer to execute an extended sound part feature parameter generation step of interpolating two phoneme chain data to generate a feature parameter of the extended sound part.

Images