JPS626299A - Electronic singing apparatus - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an electronic singing device that can sing any song using an input human voice.

(Prior Art) Generally, electronic musical instruments are known to generate electrical sounds according to musical scores.The sounds that can be generated are electronic sounds or pseudo sounds, and are not human voices. Furthermore, a speech synthesizer is known as a device that electrically generates human speech.

Conventionally, as technology in this field, Nikkei Electronics (1984-7-2) Nikkei McGraw-Hill Co., Ltd. "Regulated speech synthesis that converts arbitrary sentences into speech becomes practical JP,
There was something described in 124. below.

Its configuration will be explained using figures.

FIG. 2 is a block diagram showing an example of the configuration of a conventional speech synthesis device. This speech synthesis device includes an input interface 10 and a dictionary file 2 connected via a path line 5.
0, document-synthesis unit conversion rule file 30, program file 40 storing a synthesis unit concatenation program and accent intonation program, synthesis unit acoustic parameter file 50, speech synthesizer 80. D/A converter 70, speaker 75, control CPU 80. and a work memory 80.

Input interface 10 from a keyboard (not shown)
The character data input via the dictionary file 20 and the sentence-composition unit conversion rule file 30 are converted into a composite unit string that becomes a unit of pronunciation, and this converted composite unit string is stored in the work memory 90. The 0-synthesis weighted acoustic parameter file 50 stores acoustic parameters (for example, formant information) of synthesis units obtained by extracting human speech in advance. When synthesizing speech, the acoustic parameters of adjacent synthesis units are concatenated and input to the speech synthesizer 80 and the D/A converter fi 70. At this time, the synthesis unit concatenation program and accent intonation addition program in the program file 40 add accents for each word and intonation of the entire sentence, synthesize natural speech, and output the speaker 75.
is uttered from. Note that the control GPU 80 controls the timing of the entire device.

(Problem that the invention seeks to solve) However, with conventional electronic musical instruments, it is not possible to produce the voice of a single person, and even if it is possible to produce a pseudo-sound similar to a human voice, it is not possible to make the lyrics sing. That was impossible.

Also, if you add clauses to a sentence using a conventional speech synthesizer,
There was a problem in that data corresponding to the accent of each word and the intonation of the entire sentence had to be prepared in advance, so it was impossible to sing any song.

The present invention provides a device that solves the problem of the prior art, which is that it is not possible to sing any song.

(Means for Solving the Problems) In order to solve the above-mentioned problems, the present invention provides a singing information input means for inputting musical scores and lyrics, a lyrics decomposition means for decomposing the lyrics into composite units, and a method for extracting beats from the musical scores. Additional data generation means for generating additional data such as accent and intonation, audio means for inputting voice, acoustic parameter conversion means for analyzing voice and converting it into acoustic parameters, and acoustic parameter storage means for storing the acoustic parameters. and a speech synthesis means for synthesizing speech according to the synthesis unit and additional data using the acoustic parameters.

(Function) According to the present invention, since the electronic singing device is configured as described above, the synthesis unit and additional data such as time signature, accent, and intonation are generated from inputted singing information and analyzed from separately inputted voice. Using the acquired acoustic parameters, speech can be synthesized according to the synthesis unit and additional data. Therefore, the above problem can be eliminated.

(Embodiment) FIG. 1 is a configuration block diagram of an electronic singing device showing an embodiment of the present invention. The input interface 110 inputs a musical score and singing information of lyrics sung on the musical score. As a method for inputting score information, for example, write score information on a regular score using a barcode,
There is a method of reading this with a hand scanner, and a method of inputting musical score information in advance using a keyboard or the like and recording it on a magnetic tape or the like, and then playing it back and inputting it. The dictionary file 120 stores dictionary data for breaking down lyrics into parts of speech (nouns, verbs, particles, adverbs, etc.). Even if there is an error in the lyrics input through the input interface 110, the dictionary data in the dictionary file 120 can be used to prevent incorrect input. Conversion rules for decomposition are stored. Here, the synthesis unit is the minimum unit for synthesizing speech, and is, for example, vowel + vowel, consonant + vowel, vowel + ten consonants + vowel, consonant + vowel and ten consonants. input interface 110 to pass line 1
The lyrics input via 00 are decomposed into composite units based on the conversion rules of the composite unit conversion rule file 130.

The work memory 180 stores the decomposed composite units.
It is. In addition, in the synthesis unit concatenation program file 140, concatenation information for each synthesis unit necessary for synthesis is created,
Stored.

On the other hand, a meter accent intonation addition program file 140 is provided to process the musical score input via the input interface 110. The meter, accent, and intonation addition program file 140 stores a program for creating additional data such as meter, accent, and intonation to be added to lyrics. The created additional data is stored in this program file 140.

The voice input section 150 is for inputting human voice.

A few loud words or short sentences that are desired to be sung are defined as necessary for extracting acoustic parameters (Holtmant information) that can be used to vocalize any synthetic unit. The audio input unit 150 includes, for example, a microphone that converts audio into an electrical signal, a graphic equalizer that adjusts the sound quality, and an amplifier that amplifies the audio signal. The synthesis unit acoustic parameter conversion unit 180 analyzes the audio from the audio input unit 150 and extracts necessary acoustic parameters.

The extracted acoustic parameters are stored in the acoustic parameter file 170.

The speech synthesizer 200 uses the acoustic parameters in the acoustic parameter file 170 to synthesize speech according to the synthesis unit in the work memory 180 and additional data such as time signature, accent, and intonation.The synthesized speech is D/
It is output as actual synthesized speech through the A converter 210 and the speaker 220. As a speech synthesis method, for example, PARCOR (Partial Auto Corrl
ation) expression, LSP (Line Spectrum
There are several methods, such as Pa1rs) method, Bolt Mant method, and Mel Cepstrum method.

The control CPU 180 controls the timing of the entire system, such as receiving and passing each information and data.

Next, the operation will be explained. The operation of this electronic singing device is based on the input singing information, the lyrics synthesis unit, time signature, accent, etc.
It can be divided into a preparatory operation in which additional data such as intonation is created and acoustic parameters are extracted from input speech and stored, and a speech synthesis operation in which the speech is actually synthesized and output from a speaker.

These preparation operations and speech synthesis operations will be explained using specific examples. Lyrics rK via input interface 110
If ARASUNAZENAKUJ is input,
rKARA using the dictionary data in the dictionary file 120
SUJ is recognized as a noun (common noun), rNAZEJ is an adverb, and rNAKUJ is a verb. can do. Furthermore, using the synthesis unit conversion rule file 130, rKARASUJ is converted into rKARJ and r.
Decomposed into synthetic units RAsJ and rSUJ, rNAZE
J is decomposed into synthetic units rNAZJ and rAZEJ,
rNAKUJ is decomposed into synthetic units rNAKJ and rKUJ. These synthesis units are stored in the work memory 130 and prepared for speech synthesis operation.

In the synthesis unit concatenation program file 140, the synthesis units are rKARJ, rRASJ, rsUJ, rNAZJ.
, rAZEJ, rNAKJ, and rKUJ are generated and stored in this order.

On the other hand, a musical score indicating the lyrics rKARASUNAZENAKUJ is inputted via the input interface 110 and sent to the time signature, accent, and intonation addition program file 140. In the program file 140, time signature, accent,
Generate and store additional data such as intonation.

In addition, a predetermined word or short sentence is input from the voice input section 150 in a loud voice that the user wants to sing. The synthesized word acoustic parameter converter 160 extracts acoustic parameters from the input speech and stores them in an acoustic parameter file 170. This completes the preparation operation.

Next, the speech synthesis operation will be explained. Synthesis data in work memory 190, synthesis unit concatenation program file, and time signature, accent, and intonation program file 1
40 and additional data such as meter, accent, intonation, etc. are transferred to the speech synthesizer 200 via the path line 100 at a predetermined timing. The speech synthesizer 200 synthesizes speech based on this information, and the synthesized speech is converted into an analog electrical signal by the D/A converter 210 and output from the speaker 220. This allows the lyrics to be uttered as loud as you want the song to be sung, exactly as per the musical score.

In this way, according to this embodiment, any song can be sung with any voice.

FIG. 3 is a block diagram of an electronic singing device showing another embodiment of the present invention. The difference from the electronic singing device shown in FIG. 1 lies in the method of inputting audio to the synthesis unit acoustic parameter conversion section 160. In this embodiment, the audio input signal section 150 is not provided, and the synthesized acoustic parameter conversion section 180 is connected to the pass line 10.
Connected to 0. Audio is input interface 110
and is input to the synthetic acoustic parameter converter 160 via the pass line 100. Even with this configuration, it has the same advantages as the above embodiment.

The present invention is not limited to the above-mentioned embodiments, and various modifications are possible.For example, the present invention may be made to sing with an arbitrary person's voice in real time while inputting singing information, and among the singing information, Musical score information may be input from a keyboard and singing may be performed in real time.

(Effects of the Invention) As described in detail above, according to the present invention, there is provided a singing information input means, a lyric decomposition means for decomposing lyrics into composite units, and generating additional data such as time signature, accent, and intonation from a musical score. Since additional data generation means is provided, the amount of data to be prepared in advance for speech synthesis can be reduced compared to conventional speech synthesis devices. In addition, by providing a voice input section and a synthesis unit acoustic parameter conversion section,
You can immediately generate the acoustic parameters of the person you want to sing. Therefore, you can make the user sing any song as loudly as you want. Incidentally, since the songs uttered from the electronic singing device according to the present invention are as accurate as musical scores, they are extremely useful for singing practice.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an electronic singing device showing an embodiment of the present invention, FIG. 2 is a block diagram of a conventional speech synthesis device, and FIG. 3 is a block diagram of an electronic singing device showing another embodiment of the present invention. It is a configuration block diagram. 100... Pass line, 110... Input interface, 120... Dictionary file, 130... Composite unit conversion rule file, 14
0...Program file, 150...
- Audio input section, 160...Synthesis unit acoustic parameter conversion section, 170...Acoustic parameter file, 180...Control CPU, 190...
...Work memory, 200...Speech synthesizer, 2
10...D/A converter, 220...Speaker.

Claims (1)

[Scope of Claims] Singing information input means for inputting a musical score and lyrics to be sung on the musical score; Lyrics decomposition means for decomposing the lyrics inputted by the singing information inputting means into composite units; and the singing information input. additional data generation means for generating additional data such as time signature, accent, intonation, etc. to be added to the lyrics from the musical score input by the means; an audio input means for inputting audio; and an analysis of the audio input from the audio input means. acoustic parameter converting means for converting into acoustic parameters of the synthesis unit; acoustic parameter storage means for storing the acoustic parameters converted by the acoustic parameter converting means; and using the acoustic parameters stored in the acoustic parameter storage means. An electronic singing device comprising: a voice synthesizing means for synthesizing voice according to the synthesis unit decomposed by the lyrics decomposing means and the additional data generated by the additional data generating means.