JP6474518B1 - Simple operation voice quality conversion system - Google Patents

Abstract

【Task】
The present invention provides a variety of enjoyment of karaoke by simple operation using anyone's handheld information terminal device, and provides a new way of enjoying karaoke equipment that has not been available so far. The challenge is to provide technology that can deepen communication with friends.
[Solution]
A system that can freely convert voice quality with a simple operation, and is composed of a small and lightweight information terminal device that can be carried and a karaoke device, and the information terminal device selects an icon represented by a display function A voice quality conversion function for performing desired voice quality conversion by operation is provided. The karaoke apparatus is connected to a wireless LAN corresponding to a mobile data communication function provided in the information terminal apparatus, and inputs voice information whose voice quality has been converted. A means to output from the speaker through the mechanism was adopted.
[Selection] Figure 1

Description

  The present invention relates to a technique for enjoying music by karaoke, and more specifically, by using a familiar information terminal device such as a mobile phone, the voice of a singer can be converted into the voice of another person with a simple operation, The present invention relates to a technology of a system that can output an animal voice or a hand of mechanical sound from an audio device.

  Human voices vary from person to person and can only be spoken in their own voice, depending on the voice of the voice actor or actor who is comfortable to listen to, the elderly who gives a strict impression, the voice of a cute child, or the scene Many people have longing to sing with a voice other than their own.

In addition, there are many people who enjoy everyday singing using a karaoke device regardless of generation. In karaoke establishments, it is very common to sing along with accompaniment and enjoy the same space. However, even when singing the same level of voice at the same volume, the quality of the voice varies from person to person, so that the voice of a person becomes a soft “unspread voice”, whereas the voice of a person Becomes a "voice that resonates in space." Anyone who is interested in singing feels such a “single voice individual difference”. Therefore, conventionally, the karaoke apparatus is provided with an echo function (adding a “delay sound” delayed from the original sound) and is widely spread. However, this alone is not satisfactory, and some karaoke devices with a voice changer function have appeared, but they can only be used with devices with such a function and are widely released. That's not true.

  Another way to enjoy karaoke is “imitation”, where you sing as if you were someone else. People who change their voices and make “imitations” tend to become popular and entertain everyone. However, the technology of “imitation” is quite difficult, and it cannot be said that everyone can enjoy it. Therefore, it can be said that there is a demand for a technology that enables even those who do not have the “singing voice” and “imitation” talents described above to enjoy everyone with simple operations using familiar devices.

  Conventionally, various technical proposals have been made to solve such problems. For example, Patent Document 1 discloses a technique in which the name of the invention is “a method for real-time utterance natural language translation and an apparatus therefor”, which is a known technique (see Patent Document 1). Specifically, the problem is “providing a method and apparatus for performing real-time automatic translation of spoken language”, and the solution is “the spoken language input is received in a form including at least one source language on one side. Transmitted using natural language processing technology and language translation processing technology, including at least one target language on the other side, spoken language input communicated from one user to another, Includes speech, words, sentences, accents, phrases, and pronunciations.Transmitted spoken language input goes through a decomposition process performed by natural language decomposition technology on one side for target output language translation and digitized global Encoded to source decomposition language format, which is received by the other side. Is taken and decoded into an integrated digitized global target language format, which on the other side goes through a further integrated decoding process performed by natural language integration technology and is translated into spoken language output containing at least one target language It is said. This technology is excellent in that language translation can be performed in real time, and this is not used for translation, but it is a shortening used among Japanese language students who speak babies and modern high school students. It can be said that it can be applied to words and so on. However, if it is performed in accordance with karaoke, the karaoke device must have a dedicated facility that solves the problem such as the time lag that occurs during conversion and the adjustment of the number of characters after conversion in accordance with the tempo of karaoke, Existing karaoke facilities that do not have such devices are not compatible.

  Patent Document 2 discloses a technique in which the name of the invention is a “karaoke voice changer device”, which is a known technique (see Patent Document 2). Specifically, “It is a voice changer by the foot switch operation of a singer.” The solution is “providing a foot switch body in the middle of the connection between the microphone and the karaoke device, The voice changer circuit is provided. " However, in such an invention, a special device called a foot switch must be used, and the problem of using a device that anyone easily has, which is a problem of the present invention, has not been solved.

  Patent Document 3 discloses a technique in which the name of the invention is “portable audio reproduction and loudspeaker”, which is a known technique (see Patent Document 3). Specifically, in a portable karaoke apparatus, a key adjustment function capable of a simple operation suitable for karaoke use is realized. “A case main body containing an audio reproduction unit and a speaker; An audio input unit integrally connected to the case body, and an audio reproduction means for taking out an audio reproduction signal from an audio recording medium of the audio reproduction unit, and an external sound input by the audio input unit as an input audio signal Is a portable audio reproduction device including signal mixing and amplification means for mixing and amplifying signals based on the above. However, in the invention, although the operation is simplified, a dedicated device is necessary, and it cannot be said that everyone can enjoy unprecedented enjoyment using a handheld mobile phone or the like and an existing karaoke device. However, the problem of the present invention has not been solved.

Special table 2010-528389 JP-A-10-222185 Japanese Utility Model Publication No. 4-126300

  The present invention allows anyone to use their own information terminal device such as a mobile phone instead of a microphone, and even with an existing karaoke device, to give a variety of arrangements or to change the voice quality to sing Providing a simple operation voice quality conversion system that allows users who are not good at the machine to easily execute these by operating the icons on the display screen of the information terminal device they are familiar with It is to be an issue.

  The present invention is a system in which voice quality can be freely converted by a simple operation, and is composed of an information terminal device and a karaoke device that can be carried in a small size and light weight, and the information terminal device is represented by a display function. A voice quality conversion function for performing desired voice quality conversion by an icon selection operation is provided, and the karaoke apparatus is connected to a wireless LAN corresponding to a mobile data communication function provided in the information terminal device and inputs voice information subjected to voice quality conversion. Then, a means of outputting from the speaker via an acoustic mechanism was adopted.

  Further, the present invention may employ means for changing the voice quality conversion to be converted by the voice quality conversion function to a vocal cord sound source of another person different from the input singer.

  Further, the present invention employs means for changing the voice quality conversion performed by the voice quality conversion function into an animal sound, a robot tone, a musical instrument sound, a natural sound, or a mechanical sound based on the input vocal cord sound source of the singer. You can also

  Further, the present invention uses the angular velocity sensor provided in the information terminal device, the voice quality conversion function detects rotation of the information terminal device or a change in direction, and according to the movement, a percussion instrument, a rhythm instrument, Alternatively, a means for converting to the mechanical sound can be employed.

  The present invention can also employ means for adding the sound converted by the voice quality conversion function as a match.

  According to the simple operation voice quality conversion system according to the present invention, it is possible to easily change the voice quality based on the singing as described above by operating an icon displayed on the display screen using a familiar mobile phone or the like. It also has an excellent effect that even people who are not good at operating electronic devices can enjoy it easily.

  In addition, according to the simple operation voice quality conversion system according to the present invention, the voice quality of one's voice can be changed to the voice quality of another person, even in karaoke alone, a match can be made, or the voice of an instrument or the voice of an animal can be sung. However, it demonstrates an excellent effect that makes it possible to enjoy karaoke.

  Further, according to the simple operation voice quality conversion system according to the present invention, the angular velocity sensor (gyro sensor) included in the information terminal device changes the direction of the information terminal device or captures the motion of shaking the information terminal device. It produces an excellent effect of being able to obtain the effect of playing an instrument.

It is basic composition explanatory drawing which shows the basic composition of the simple operation voice quality conversion system which concerns on this invention. It is basic display screen explanatory drawing which shows the basic display screen which appears on the information terminal of the simple operation voice quality conversion system which concerns on this invention. It is detailed selection operation screen explanatory drawing which shows the detailed selection operation screen which appears on the information terminal of the simple operation voice quality conversion system which concerns on this invention. It is a basic composition explanatory view explaining the basic composition of voice quality conversion concerning the present invention. It is a structure explanatory drawing explaining the structure which the voice quality conversion which concerns on this invention changes into the voice of others, or the cry of an animal. It is a structure explanatory drawing of the structure using an angular velocity sensor for the voice quality conversion function which concerns on this invention. It is composition explanatory drawing which adds the sound converted by the voice quality conversion function which concerns on this invention as a match.

  FIG. 1 is a basic configuration explanatory view showing a basic configuration of a simple operation voice quality conversion system 1 according to the present invention. As shown in FIG. 1, the system is roughly divided into an information terminal device 10 and a karaoke device 20 in terms of articles.

  The information terminal device 10 typically means a device that has a communication function and can touch information, such as a mobile phone. Note that it includes a wide range of tablets, small personal computers, game machines, PDA (Personal Digital Assistant), etc., which are home information appliances used at home.

  The display function 11 is a function for displaying an icon 12 for operation, and is preferably a display device such as a liquid crystal screen capable of instructing a request by touching the icon 12 as a touch panel if possible.

  The icon 12 displays a symbolic pattern that is a target of voice conversion. For example, as shown in FIG. 2B, the icon 12 may be a man and a woman or a child and an elderly man. When a picture is prepared and the male selects the female icon 12, a command to convert the input male vocal cord sound source 32 into the female vocal cord sound source 32 is output to the voice quality conversion function 13. In addition, about the icon 12, the thing shown in FIG.2 (b) and FIG.3 (b) is an illustration to the last, Comprising: It does not specifically limit to these, Other character display, character display, etc. may be sufficient .

  The voice quality conversion function 13 receives a command to convert the voice quality to the selected voice quality by operating the icon 12, and includes voice synthesis techniques, corpus-based voice synthesis, voice quality conversion / morphing, prosody generation, prosody corpus, etc. Get the converted sound source. Specifically, it is also preferable to use software that currently functions as a voice changer. As a simple one, a filter that changes the pitch (voice pitch) or formant frequency up and down is used. For voice quality conversion that amplifies or attenuates any frequency band included in the original sound, such as those that change the pitch partially or totally, such as an octaver, pitch shifter, frequency shifter, or ring modulator, or an equalizer The use of a dedicated effector can also be considered.

  The mobile data communication function 14 is a function for connecting to a communication network capable of transmitting and receiving digital data for communication from the information terminal device 10, and is a Wi-Fi (registered trademark) (Wireless Fidelity) that is a kind of adapter of the wireless LAN 40. ) And Bluetooth (registered trademark), which is a short-range wireless communication standard for digital devices, can be considered. In recent karaoke apparatuses 20, the microphone M and the operation unit are often based on wireless communication, and using these communication networks is also an effective means. Note that when the information terminal device 10 is within the communication range of these communication networks, it is desirable to display an icon 12 on the display unit to prompt activation of the application.

  The text box 16 is an input unit for selecting voice feature data by inputting the name of another person to be identified, and is performed by inputting text instead of operating the icon 12. Note that it is also preferable to store characteristic data of a specific person in advance in a storage device of the information terminal device 10 or to obtain information by accessing a server on the Internet.

  The karaoke device 20 may be a general karaoke device 20 including a main unit, an operation unit, a display unit D, a communication unit, an acoustic unit, a speaker 22, and the like. In particular, it is sufficient if the vocal cord sound source 32 from the wireless microphone M can be input and amplified by an amplifier or output via the speaker 22, but it is desirable to have a communication function with a network such as the Internet. . In recent years, there is a type in which a liquid crystal touch panel serving as an operation unit can be removed, and since it communicates with the main body, it is desirable to use these communication functions in the mobile data communication function 14. In addition, it is more preferable that the content displayed on the display unit of the information terminal device 10 is displayed on the display D.

  The acoustic mechanism 21 converts a human vocal cord sound source 32 input from a microphone M included in the information terminal device 10, a sound source such as a musical instrument made by the angular velocity sensor 15, and the like into an electric signal, so that the mobile data communication function 14 This is an apparatus that amplifies an electrical signal input via the speaker 22 as a sound source for reproduction by the speaker 22 and outputs it by mixing (mixing) or adding an effect (effect).

  The speaker 22 reproduces the original reverberation by vibrating the air by the electric signal converted from the sound source. In this invention, it does not specifically limit and what is necessary is just to use the speaker 22 with which the general karaoke apparatus 20 is equipped.

  The voice quality conversion 30 obtains a sound source converted by the selected voice quality conversion 30 from techniques such as speech synthesis method, corpus-based speech synthesis, voice quality conversion / morphing, prosody generation, prosody corpus, etc., by operating the icon 12. Hereinafter, the kind of the voice quality conversion means will be described.

(Synthesis method)
In human utterance, the sound source waveform sent out from the glottis generates various sounds by resonating in a complexly changing vocal tract. This process can be modeled as an electric circuit with an input signal and a resonance circuit, and the model of the speech generation process that considers the resonance in the glottal source and the vocal tract separately is called the linearly separated equivalent circuit model (or source filter model). This gives the basics of current speech information processing. One method of synthesizing speech by a computer is a method based on a generation process model. In other words, it is a technique for simulating the speech generation process by constructing a digital filter that expresses the resonance characteristics in the vocal tract and inputting a sound source waveform approximated by an impulse sequence or white noise. be called. Another speech synthesis method that is different from this is called waveform connection synthesis method, where many time waveforms of recorded speech are stored, an appropriate waveform (or part of it) is selected and connected in the time domain. It is a method to play back.

(Analytic synthesis and rule synthesis)
General speech analysis is a process that converts a time waveform of speech into a time series of feature parameters. Usually, an analysis based on a linear separation equivalent circuit model is performed for each frame (small time interval), and the vocal tract The characteristic parameter time series can be converted into a speech waveform by obtaining a time series of characteristic parameters expressing the characteristics of the above and driving a digital filter configured based on the characteristic parameters with a sound source waveform. The process of restoring the original speech waveform from the parameter time series obtained by speech analysis for speech once uttered in this way is called analysis synthesis.

  To synthesize the speech of a sentence (such as a word) that the singer has not uttered, create a segment that expresses the features of the speech for each short unit, such as a phoneme called a synthesis unit, and synthesize it The sentence is divided into synthesis units, the segment sequence for the synthesis unit sequence is determined, and the speech is synthesized. Such a method of synthesizing an arbitrary sentence or word using a relatively short synthesis unit is generally called rule synthesis. As synthesis units, phonemes, syllables, vowel-consonant chains, and 2-phoneme chains have been tried.

(Parameter synthesis method (vocoder method))
In the speech synthesis of the parameter synthesis method, in order to accumulate phoneme features, it is necessary to convert speech waveforms into feature parameters, and speech synthesis using various feature parameters is performed.

(Voice tract analog synthesis method and formant voice method)
Vocal tract analog synthesis (articulatory synthesis) that simulates the propagation of sound waves in the vocal tract by approximating the shape of the vocal tract, and the method of simulating the characteristics of articulation in the vocal tract with an electrical circuit that connects multiple resonance circuits It is also desirable to form a synthesis filter by combining formant synthesis filters directly or in parallel and combining them. In recent years, as a speech synthesis using other parameters, a hidden Markov model (hereinafter referred to as HMM) speech synthesis using a cepstrum as a characteristic parameter has become a representative method of parameter synthesis.

(Wave connection synthesis method)
Whereas the parameter synthesis method is realized by simulating the speech generation process, concatenative synthesis divides a large number of observed speech waveforms into synthesis units and uses each time waveform as a fragment. Speech synthesis is realized by accumulating in the database, extracting the optimum segment sequence for the sentence to be synthesized from the database, and connecting and reproducing in the time domain.

(Selection of fragment)
When the sentence to be synthesized is composed of several synthesis units, it is determined as a segment sequence that minimizes the cost defined by the optimum segment sequence. Here, it is defined by linguistic features such as phonemes and accent types and physical quantities such as formant frequencies. In addition, the amount of physical (acoustic) strain at the connection between two adjacent pieces is evaluated.

(PSOLA)
In waveform connection synthesis, sound quality is degraded when there is no appropriate segment in the database. Especially, since the prosodic changes of speech are various, the fundamental frequency and time length do not sufficiently match the target synthesis unit. In order to solve this problem, there is a method PSOLA (Pitch Synchronous Overlap and Add) that changes the fundamental frequency and time length of a fragment by processing in the time domain. Also, the time length can be easily controlled by repeatedly adding or thinning out the cut 1 pitch waveform as necessary.

(Comparison between parameter synthesis and waveform connection synthesis)
Parameter synthesis requires a process that drives a filter that expresses articulation characteristics with a sound source waveform, so the sound quality is somewhat mechanical, but the connection at the segment boundary is smooth, and the fundamental frequency and time length are controlled. Can be easily performed. In recent years, voice quality and emotion can be controlled.

(Corpus-based speech synthesis)
It is preferable to use corpus-based speech synthesis for speech synthesis of dogs and cats. Corpus-based speech synthesis is a method of synthesizing speech by preparing speech corpora (database), extracting speech segments from them, and joining them together. ) Connected speech synthesis ". In the voice quality conversion in the present invention, when the diversity of phoneme arrangement, accent / intonation, etc. is small, such as dog or cat squeal, drums, maracas or motorcycle exhaust, etc., it is a practical level with high naturalness. Sound quality will be obtained. However, it is necessary to secure a storage capacity for storing audio information as a database.

(Voice conversion / morphing)
The voice quality conversion 30 converts the voice quality of a singer into the voice quality of another person, and is a device that controls non-linguistic information such as voice quality and personality without changing the language (phonological) information included in the voice. . Existing text-to-speech synthesis technology has already achieved sufficient performance from the viewpoint of the transmission of linguistic information such as phonology and clarity of synthesized speech. There are still many inadequacies from the viewpoint of adding and controlling paralinguistic information. Voice quality conversion, on the other hand, is considered as a technology that makes it easy to control non-linguistic information. Voice morphing, which generates voices of any intermediate virtual singer between different singers, is used for dialogue. Various applications such as systems, medical welfare equipment, toys, games, and entertainment are expected. The individuality included in the speech depends on the spectral outline of the speech and the features that appear in the prosody. Spectral features depend on the characteristics of the individual articulatory organs of the singer, that is, physical features such as vocal cords and vocal tract shape, and appear mainly as a difference in the voice quality of the singer, while prosodic features are intonation. And accents, voice pitch, speaking speed, phoneme duration, etc. For this reason, in order to realize voice quality conversion, conversion of both spectral features and prosodic features as described below is required.

(Spectral conversion)
Considering the voice quality conversion from one singer (former singer) to a different singer (target singer), the spectral features of the original singer's voice at each time (for example, mel cepstrum coefficient vector and line spectrum frequency vector) Voice quality conversion that focuses on the conversion of spectral features is the feature quantity of the target singer corresponding to it, by obtaining a conversion function that outputs the characteristic quantity resembling the target singer based on the spectral feature quantity of the original singer It can be formulated. In the method based on vector quantization (VQ), feature vector quantization of each singer using speech data (parallel data) for learning that the original singer and the target singer uttered the same content (phoneme sequence). When creating a codebook, finding the frequency distribution of the correspondence between feature quantities (code vectors) after vector quantization between the original singer and the target singer, and performing voice quality conversion, the input voice of the original singer An index of a code vector obtained by vector quantization of the feature quantity is obtained and used as a converted feature quantity.

(Prosody conversion)
Prosodic features are important factors that affect the individuality of the singer as well as the voice quality. However, compared to the conversion of spectral features, there are limited methods for converting prosodic features, and a method based on vector quantization similar to spectral conversion, simply learning the average value of the fundamental frequency and speech speed for the target singer. There are a method of matching to the average value of data, a linear conversion method of fundamental frequency considering dispersion, a method based on prosody generation using HMM speech synthesis, and the like.

(Spectrum / Prosody conversion)
When language information is given in advance in voice quality conversion, both spectral features and prosodic features can be converted based on HMM speech synthesis and singer adaptation. HMM speech synthesis uses a speech unit HMM in which the spectrum, fundamental frequency, and phoneme duration are modeled at the same time. By adapting the speech unit HMM based on a method similar to singer adaptation used in speech recognition, It is possible to generate a voice of any singer character including spectral features and prosodic features. By using the average voice model learned from the voice data of multiple singers as the model of the former singer, if there is a small amount of adaptation data uttered by any target singer, it has a singer character close to the target singer 6), and can be easily applied to control of paralingual and non-linguistic information including utterance styles and emotional expressions.

(Prosody generation)
The prosody that represents the pitch, speed, and strength of the voice mainly depends on the acoustic features such as the fundamental frequency of the vocal cords, the duration of the segment corresponding to the phoneme (hereinafter called the phoneme length), and the speech amplitude. Is called. In order to give an appropriate prosody to the synthesized speech, analysis of the mechanism of prosodic control and mathematical modeling are performed. As shown below, each feature value is calculated using a statistical model based on actual speech data based on language information obtained from input text or the like.

(Control of fundamental frequency of vocal cords)
Voice pitch is important for expressing accents and intonation, and is achieved by controlling the fundamental frequency of the vocal cords. In particular, Japanese accents are so-called high and low accents, which are realized as local fundamental frequency undulations, and the fundamental frequency is controlled as intonation.

(Prosodic corpus)
A speech corpus used for speech recognition or synthesis is designed mainly by focusing on segmental features of speech (features related to consonants and vowels and their combination). On the other hand, the prosodic corpus is a corpus designed by paying attention to the prosodic features of speech. Prosodic features include accent, stress, tone, prominence, intonation and the like. Linguistically, duration information is also part of the prosodic features, but this is also information provided in a normal speech corpus.

(Hierarchical structure of prosodic features)
In describing the prosodic structure of speech, a hierarchical structure is usually assumed. In Japanese, the lowest hierarchy is usually a segment of sound with a certain length of time, and from there, the utterance that is the highest hierarchy through syllables, words, accent phrases, intermediate phrases, etc. Assuming a structure that leads to That is, it is necessary to consider the interaction between words. In many cases, an accent phrase, which is a unit of a size corresponding to a phrase, or an intermediate phrase between an accent phrase and an utterance is grasped.

  The sound information 31 is a sound source obtained by any one of the methods of the voice quality conversion 30 or a combination thereof, and is amplified by an acoustic device and recognized by human hearing via the speaker 22. Is the information of the sound to be played.

  The vocal cord sound source 32 is a sound source in which a sound generated in a resonance cavity called a human pharyngeal cavity is emitted, and is a sound source that is a feature of an individual voice determined by a distribution of formant frequencies. Such a sound source is generated by a waveform coding method, a vocoder method method, a hybrid method signalization, or the like.

  The wireless LAN 40 means a wireless communication network such as a communication network (Wi-Fi (registered trademark)) based on a common standard represented by IEEE 802.11, which is currently standardized, and does not exclude wired connections. It is desirable to be able to cope with higher speeds by communication of such standards. Further, Bluetooth (registered trademark) of a communication network of IEEE802.15.1 which is the same wireless standard may be used. Bluetooth (registered trademark) can communicate if it is about 10 meters, and since a recent smartphone usually has a mobile data communication function 14 corresponding to the communication network, Connection is easy.

  FIG. 2 is an explanatory diagram illustrating a configuration in which the voice quality conversion 30 according to the present invention is changed to another person's vocal cord sound source 32, and FIG. 2 (a) is a state in which the portable information terminal 10 displays a menu screen. FIG. 2 (b) shows a state where the menu screen is shown in a state where the simplified operation voice quality conversion system 1 according to the present invention is activated, and FIG. 2 (c), FIG. 2 (d), FIG. 2 (e) shows a screen for performing further detailed setting from each icon 12 selected from the menu screen. In addition, what was represented to drawing is an illustration to the last, and is not limited to the display which concerns. The hierarchy is the same. FIG. 3 is an explanatory diagram of a detailed selection operation screen showing a detailed selection operation screen appearing on the information terminal of the simplified operation voice quality conversion system 1 according to the present invention, and FIG. 3 (a) is the same as FIG. 2 (b). Similarly, the state which represents the menu screen in the state by which the simple operation voice quality conversion system 1 which concerns on this invention was started is shown, FIG.3 (b) changes the voice quality conversion 13 to the vocal cord sound source 32 of others. FIG. 3 (c) shows a screen for changing to a vocal cord sound source 32 by an animal cry 33, and FIG. 3 (b) shows a screen for changing to a sound source generated from an instrument sound 35, a mechanical sound 37, or the like. FIG. 3 (e) shows an example in which the screen is expanded continuously to the next more detailed selection screen in a hierarchical manner.

  FIG. 4 is a basic configuration explanatory diagram illustrating a basic configuration of voice quality conversion according to the present invention. First, the application is activated by selecting from the menu screen (FIG. 2A) of the portable information terminal 10 using the icon 12 to activate the simplified operation voice quality conversion system 1 according to the present invention. (FIG. 2 (b)), the desired voice quality is selected, and at the same time, the mobile data communication function 14 of the karaoke apparatus 20 is connected so that they can communicate with each other. A basic configuration is adopted in which the voice quality conversion function 13 is executed through the microphone M, the voice quality conversion function 13 is executed, and the sound device 21 is used to output from the speaker 22.

  FIG. 5 is a configuration explanatory diagram for explaining a configuration in which the voice quality conversion 30 according to the present invention is changed to another person's voice or an animal call, and FIG. 5 (a) is a diagram illustrating the voice quality conversion 30 according to the present invention. FIG. 5B is a configuration explanatory diagram illustrating a configuration for changing to a vocal cord sound source 32 of another person, and FIG. 5B is a configuration explanatory diagram illustrating a configuration in which the voice quality conversion according to the present invention is changed to an animal call. It is changed to a vocal cord sound source 32 having the characteristics of the other person's voice to be converted by the voice synthesis or the like, an animal call or the like, and by a method such as synthesis of the various voice data or selection of phonemes. In this configuration, the obtained voice of another person is output from the sound device 21.

  The animal cry 33 performs voice quality conversion to be the cry 33 of various animals such as cats, dogs, lions and wolves. For the selection of such conversion, an icon 12 having each animal as a motif is prepared so that it can be recognized at a glance. For example, as shown in FIG. 3C, when an icon 12 having a dog and cat motif appearing in the display function 11 of the information terminal device 10 is operated, the screen is switched to the animal mode screen. Various animals such as crows, elephants, etc. can be selected, and if a dog is selected based on the input vocal cord sound source 32, it is changed via the voice quality conversion function 13 such as an effector that changes to a sound source such as “One One One”. It is the vocal cord sound source 32 after a while.

  The robot tone 34 performs a voice quality conversion 30 that is a voice uttered by the talking robot. For the selection of such conversion, an icon 12 having a robot motif as shown in FIG. 2B or FIG. 3D is prepared so that it can be seen at a glance. For example, when an icon 12 with a robotic motif is operated, two sound sources such as “Wa, Ta, Si, Ha ...” of a voice that is generated by dividing each robot sound based on the input vocal cord sound source 32 are generated. This is the sound information 31 output after using the voice quality conversion function 13 such as combining and combining the pitch of the voice with the tone generator frequency to a different pitch to make it sound like a robot.

  The musical instrument sound 35 performs a voice quality conversion 30 that is the sound of any musical instrument such as a piano, a stringed instrument, or a wind instrument. For the selection of such conversion, an icon 12 having each musical instrument as a motif is prepared so that it can be recognized at a glance. For example, as shown in FIG. 3 (c), when the icon 12 having a guitar motif is operated, the voice quality conversion is changed to a beautiful sound source such as “polon pollon” of the guitar based on the scale or pitch of the input vocal cord sound source 32. The sound information 31 is output after the function 13 is used.

  The natural sound 36 performs voice quality conversion 30 that is a sound existing in the natural world, such as a wind sound, a wave sound, and a river sound. For the selection of such conversion, for example, although not shown in the drawing, like the one shown in FIG. 3E, icons 12 representing the respective characters may be prepared so as to be recognized at a glance. . For example, when the wave character icon 12 is operated, sound information 31 output later using the voice quality conversion function 13 for changing to a sound source such as “Zazaza” of the wave based on the scale or pitch of the input vocal cord sound source 32. is there.

  The mechanical sound 37 performs a voice quality conversion 30 that is a sound emitted from a mechanical object such as a motorcycle or a whistle. For the selection of such conversion, for example, as shown in FIG. 2B and FIG. 3D, an icon 12 representing a motorcycle or a ship as a motif may be prepared so as to be recognized at a glance. For example, when the motorcycle character icon 12 is operated, sound information 31 output after using the voice quality conversion function 13 for changing to a sound source such as “Boom Buon” of a motorcycle based on the scale or pitch of the input vocal cord sound source 32. It is.

  FIG. 6 is an explanatory diagram of a configuration in which the angular velocity sensor 15 is used in the voice quality conversion function 13 according to the present invention. Prepare a route for performing the voice quality conversion 30 based on the vocal cord sound source 32 and a parameter synthesis or a voice corpus (database) for driving the vocal cord sound source 32 to drive a filter expressing the articulation characteristics registered in advance. The voice velocity of the percussion instrument 38 such as drums and cymbals prepared therein is extracted, and a route for synthesizing the speech by synthesizing the speech is used. The sound information 31 is output corresponding to the above.

  The angular velocity sensor 15 captures changes in rotation and orientation used in gyro functions of mobile phones, small game machines, and the like. For example, when the mobile phone is shaken, the moving direction of the mobile phone changes with a continuous rhythm. It is a sensor that makes it possible to reproduce a predetermined sound source such as a rhythm instrument or a percussion instrument in accordance with the timing of the change of the direction. Specifically, it uses a Coriolis force that works in the direction of vibration when an angular velocity is applied to a mass that vibrates at a predetermined velocity, and is generally called a gyro sensor. Many recent mobile phones and game machines are installed internally, and it is effective to use them.

  The percussion instrument 38 means a musical instrument that produces sound from the individual itself when it is struck, such as a drum or a cymbal. In particular, the percussion instrument 38 in the present invention refers to all instruments that do not have a clear scale. The sound source is not based on the vocal cord sound source 32, and a parameter synthesis for performing a process of driving the filter expressing the articulation characteristics registered in advance by the vocal cord sound source 32 and a voice corpus (database) are prepared. Then, sound information 31 output in accordance with the operation of the angular velocity sensor 15 using corpus-based speech synthesis in which speech units of the percussion instrument 38 such as drums and cymbals prepared therein are extracted and joined together. It is.

  This means a body instrument 39, a musical instrument that produces sounds by "swinging" and "scraping" like maracas. In particular, the body instrument 39 in the present invention has a clear scale, like the percussion instrument 38. It shall mean something that does not have. The sound source is not based on the vocal cord sound source 32, and a parameter synthesis for performing a process of driving the filter expressing the articulation characteristics registered in advance by the vocal cord sound source 32 and a voice corpus (database) are prepared. The sound output from the angular velocity sensor 15 according to the operation of the angular velocity sensor 15 is extracted by using the corpus-based speech synthesis that extracts the speech units of the body sounding instrument 39 such as maracas prepared therein and combines them. Information 31.

  FIG. 7 is an explanatory diagram showing a configuration in which the sound converted by the voice quality conversion function 13 according to the present invention is added as a “matching hand”.

  The matching hand 23 is stored as a speech corpus (database) as a speech segment obtained by the voice quality conversion function 13 for each of the animal sound 33, the instrument sound 35, the natural sound 36, and the mechanical sound 37 selected by the selection of the icon 12. A speech segment that matches the pitch at the time of delimitation for each measure is added to the vocal cord sound source 32 and output. Specifically, for example, the cat icon 12 shown in FIGS. 2 (b) and 3 (c) is selected, and then the icon 12 of the matching hand 23 is selected in FIG. 3 (e). However, by selecting the timing at which the matching hand is put into every four bars, the matching hand 23 “Nyan, Nyan” by the voice of the cat is output from the audio device 21 every four bars. Moreover, as another usage, it is also possible to enjoy solo singing using the vocal band sound source 32 of the singer as it is, single singing based on the voice quality of another person, or chorusing based on the voice quality of many others.

  The simple operation voice quality conversion system 1 according to the present invention can enjoy a wide variety of karaoke, which is an entertainment closely related to the lives of Japanese people, and provides a variety of arrangements by simple operations even for those who are not good at operating machines. Therefore, the industrial applicability in a wide range of fields such as a karaoke installation shop, a karaoke device manufacturer, and a portable information terminal manufacturer is extremely high.

1 Simple operation voice quality conversion system 1
DESCRIPTION OF SYMBOLS 10 Information terminal device 11 Display function 12 Icon 13 Voice quality conversion function 14 Mobile data communication function 15 Angular velocity sensor 16 Text box 20 Karaoke apparatus 21 Acoustic mechanism 22 Speaker 23 Matching hand 30 Voice quality conversion 31 Sound information 32 Voice band sound source 33 Animal sound 34 Robot tone 35 Instrument Sound 36 Natural Sound 37 Mechanical Sound 38 Percussion Instrument 39 Body Sound Instrument 40 Wireless LAN
M Microphone D Display

Claims (4)

It is a system that can convert voice quality freely with simple operation,
A familiar information terminal device (10) that can be carried in a small, lightweight,
Karaoke device (20) and
The information terminal device (10) includes a voice quality conversion function (13) that performs a desired voice quality conversion (30) by selecting an icon (12) represented in the display function (11).
The voice quality conversion (30) converts the feature of the input voice of the original singer into the feature of the voice of the target singer by converting either or both of the spectral feature and the prosody feature,
The karaoke device (20), the information terminal device (10) automatically said information terminal device and more linked with the icon operation in the wireless LAN (40) corresponding to the mobile data communication function (14) provided in the (10 ) is connected to,
When the voice of the former singer is input from the microphone (M), sound information (31) based on the vocal cord sound source (32) converted by the voice quality conversion is output from the speaker (22) via the acoustic mechanism (21). simple operation voice quality conversion system, characterized in that it is (1). The voice conversion function (13) the voice quality conversion (30) for converting the, animal sounds based on the glottal source of the original singer input (32) (33), the robot tone (34), the instrument sound ( 35) The simple operation voice quality conversion system (1) according to claim 1, wherein the system is changed to a natural sound (36) or a mechanical sound (37). Wherein the voice conversion for converting the voice conversion function (13) (30), using the angular velocity sensor (15) provided in the information terminal device (10), the rotation of the information terminal device (10), or the orientation of the It detects the change, according to the movement of, percussion (38), simple operation voice according to claim 1 rhythm instrument (39), or which is characterized in that converts the machine械音(37) Conversion system (1). The simple operation voice quality conversion system (1) according to any one of claims 1 to 3, wherein the sound converted by the voice quality conversion function (13) is added as a match (23).