JPH09179572A - Voice converting circuit and karaoke singing equipment - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically sing just like a famous singer without imitating a voice devotedly while singing with his or her own voice. SOLUTION: Music data for KARAOKE consists of a musical performance signal and a reference interval signal 9 which are synchronized. The music data is separated by a performance-time read and separating circuit 4 into the musical performance signal and reference interval signal 9. The voice of the singer has featured parts such as a format extracted by a voice analyzing circuit 31 and a phoneme recognizing circuit 39 recognizes a single tone to obtain a phoneme; and a phoneme converting circuit 35 converts the phoneme into the phoneme of another singer by referring to a phoneme table 38 for the singer and a synthesizing circuit 34 modulates the interval with a timing signal based upon song data and the reference interval signal, so that a singing voice and an instrument performance are outputted through a sound source mixer.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice conversion circuit and a karaoke device for converting a self voice into a voice of another person, for example, a professional or a famous singer.

[0002]

2. Description of the Related Art In a conventional karaoke apparatus, a user's singer merely sings the lyrics in accordance with the tune played by the singer, and does not have the function of imitating the voice of a professional singer in a mimicry manner.

[0003]

In the conventional karaoke apparatus, the singing voice of a professional singer is sounded so as to be heard quietly,
At the same time, some singers tried to sing well, but lacked the fun because they didn't have the ability to sing like a professional singer.

An object of the present invention is to be able to sing with one's own voice in a voice that mimics a professional singer.

[0005]

A voice conversion circuit according to the invention of claim 1 analyzes the voice data of a user's voice input from a microphone 19 and replaces the user's voice data with the data of another person, or Add someone else's song data to the user's song data. In the karaoke apparatus according to the second aspect of the present invention, the song data 6 for karaoke is composed of the synchronized performance signal 8 and the reference pitch signal 9. This song data 6 is stored in the memory 20
The music data 6 is read from the memory 20, and the separation circuit 4 separates the performance signal 8 and the reference pitch signal 9.
The microphone and 19 convert the voice of the singer to song data, the voice analysis circuit 31 extracts a characteristic part such as a formant from the song data, and the speech recognition circuit 39 recognizes a single phoneme from the characteristic part such as the formant and sings the song. Output phonemes of data. The phoneme conversion circuit 35 converts the phoneme of the song data into the phoneme of another person's singer by referring to the phoneme table 38 of another person, and the timing extraction circuit 37 simultaneously generates a timing signal for each single note from the song data. Synthesis circuit 34
Then, in accordance with this timing signal, the pitch of another singer's phoneme is modulated according to the reference pitch signal. The tone generator mixer 17 mixes the performance signal with the modulated phoneme signal,
With this mixed signal, the speaker 18 is sounded so that the singing voice of another singer and the performance of the musical instrument can be heard. Even if you are singing in your own voice, you can automatically sing like the voice of another famous singer without having to concentrate on imitating.

[0006]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1 FIG. The present invention will be described below with reference to the drawings.
First, referring to FIG. 2, the karaoke apparatus of the present invention will be described. The karaoke device includes a karaoke reproducing unit 1, a separation circuit 4, a voice conversion circuit 5, a communication control circuit (I / F) 11, a bus 12, a CPU 13, a work RAM 14, a MIDI sound source 15, a sound source mixer 17, a speaker 18, a microphone 19, and H.
A storage device 20 such as a DD (hard disk drive), a lyrics character generation circuit (VDP) 22 for image reproduction, a video mixer 23, an external laser death (LD) changer 24, a display device CRT 25, an input / output device ( I / O)
26.

[0007] The karaoke reproducing section 1 via the communication control circuit 11
Is connected to the communication line 3 and can receive music data from the host 2 at a remote place as needed. The input / output device 26 receives a request from a remote controller (not shown), and the CPU 13 reads out karaoke data, for example MIDI data, from the storage device 20 if the requested song exists. The MIDI data is decompressed in the work RAM 14 and sent to the voice conversion circuit 5 and the MIDI sound source 15 via the separation circuit 4.

As shown in FIG. 4, the music data 6 for karaoke comprises a genre code 7, MIDI performance data (signal) 8 and a reference pitch signal 9.
The separation circuit 4 includes a CPU 16 and a RAM 21,
The CPU 16 receives the music piece data 6 from the bus 12 and separates it into performance data 8 and reference pitch signal 9.

The performance data 8 and the reference pitch signal 9 are temporarily stored in the RAM 21, and the MIDI performance data 8 is sent from the RAM 21 to the MIDI sound source 15. MIDI
The sound source 15 reproduces the accompaniment sound based on the song data, and these are mixed with the singing voice from the microphone 19 in the sound source mixer 17,
Sound is output from the speaker 18. The lyrics data of the karaoke data generates a character pattern in the character generation circuit (VDP) 22, the CPU 13 selects a background image according to the genre code 7, the background image from the LD changer 24 is mixed with the image mixer 23, and the CRT 25 is displayed. Is displayed.

Next, referring to FIG. 1, the configuration of the voice conversion circuit 5 will be described in detail. The voice conversion circuit 5 includes a voice analysis circuit 31, an A / D converter 30, and a phoneme recognition circuit 3.
9, standard phoneme table 33, phoneme table 38 of others,
Phoneme conversion circuit 35, synthesis circuit 34, D / A converter 3
2. The timing extraction circuit 37 and the CPU 36. The song data of the singing voice signal from the microphone 19 is input to the voice analysis circuit 31, and a characteristic part such as a formant is extracted from the song data.
It is input to the / D converter 30. The analog signal analyzed into a formant or the like is converted into a digital signal by the A / D converter 30 and becomes a single-tone digital signal for easy phoneme recognition and is input to the phoneme recognition circuit 39. Phoneme recognition circuit 3
9: For each single note in the lyrics, "If it's Sumida River in the spring of Urara (Hururu no Ura ~ Ra no Sumida ...)", "ha" is "ru"
C. A single sound is recognized according to a formant, such as "la" of "la".

The phoneme waveform of FIG. 3 is stored in the standard phoneme table 33, and the phoneme recognition circuit 39 refers to the standard phoneme table 33 to determine which standard phoneme the phoneme of each phoneme digital signal corresponds to. Recognize. In the standard phoneme table 33, standard digital signals for uttering 50 sounds, numbers, alphabets, etc. are pre-stored in single-tone segments. The standard digital signal recognized by the phoneme recognition circuit 39 is sent to the phoneme conversion circuit 35, and the phoneme conversion circuit 35 detects a phoneme of another person's standard digital signal corresponding to the standard digital signal and replaces it with the recognized standard digital signal. The phoneme of the standard digital signal of the other person of the lever is output to the synthesis circuit 34.

In another person's phoneme table 38, for example, a phoneme waveform of a professional singer of another person shown in FIG. 4 is stored, and in the other person's phoneme conversion circuit 35, each single-tone digital signal corresponds to one of the other person's phoneme waveforms. Or other person's phoneme table 38 for reference. 50 on other person's phoneme table 38
A standard digital signal of a professional singer, that is, another person, such as a sound, a number, and an alphabet, is stored in advance as a single note segment.

Since the phoneme waveform of the other person's phoneme table 38 is different for each other person, a memory structure of a plurality of pages can be used to imitate a plurality of other person's. A reference pitch signal 9 is given from the RAM 21 of the separation circuit 4 to the synthesizing circuit 34, and the pitch of the standard digital signal of another person is modulated so as to be corrected according to the reference pitch signal 9, and D /
Output to the A converter 32. The singing voice signal (song data) from the microphone 19 may be directly compared with the phonemes in the phoneme table 38 of another person, but in order to convert accurately, it is preferable to convert the phoneme of all people once.

The singing voice signal (song data) from the microphone 19 is also input to the timing extraction circuit 37, and for each single note of the lyrics, "If it is Sumida River of Urara in spring (it is the corner of Urara's Urara ...)". But "ha" of "u" la "la" of ...
・ The timing is extracted as. Thus, the timing signal is given to the synthesis circuit 34 from the timing extraction circuit 37, and the synthesis circuit 34 modulates the pitch of each phoneme of the standard digital signal of another person according to the reference pitch signal 9 in synchronization with the timing signal, and To the D / A converter 32. The CPU 36 controls the synthesizing circuit 34 to change the volume of the singing voice from the microphone 19 sent from the timing extracting circuit 37 at the same time. If the volume from the microphone 19 is minimum, only the modulated other person's song data is produced. If the volume from the microphone 19 is medium, the song data obtained by adding another person's song data to the self-voiced song data is synthesized circuit 34.
Output from

That is, the singing voice signal from the microphone 19 is converted into the voice of another person such as a professional, and it sounds as if a singer of another person is singing. It sounds like a chorus. The CPU 36 sets the voice analysis circuit 31 to the signal line no. 1
Then, the phoneme recognition circuit 39 is connected to the signal line no. 2, the phoneme conversion circuit 35 is connected to the signal line no. 3, the timing extraction circuit 37 is connected to the signal line no. 4, the combining circuit 34 is connected to the signal line no. Each is controlled by 5. In the sound source mixer 17, the analog signals of these singing voices and the performance signal from the MIDI sound source 15 are mixed and output to the speaker 18. Although the standard phoneme table 33 and the phoneme table 38 of other people in FIGS. 3 and 4 show analog waveforms for convenience, the phoneme data is stored in the form of digital signals.

Next, the operation of the present invention will be described in detail with reference to the flowchart of FIG. In step S1, the power of the karaoke reproducing section 1 is turned on, and the CPU 13 performs a power supply test. Then, in step S2, the system including the voice conversion circuit 5 is initialized. In step S3, the synthesis circuit 34 of the voice conversion circuit 5 receives the MIDI reference pitch signal 9 from the RAM 21 of the separation circuit 4. Step S4
Then, an analog signal of a singing voice is input from the microphone 19 to the voice analysis circuit 31. In step S5, the voice analysis circuit 3
Reference numeral 1 extracts a characteristic part such as a formant from an analog signal (song data) of a singing voice.

In step S6, the phoneme recognition circuit 39 recognizes the phonemes of the individual phonemes with reference to the standard phoneme table 33 for the characteristic parts such as formants extracted from the analog signal of the singing voice. In step S7, the phoneme conversion circuit 35 compares the recognized phonemes with the phoneme table 38 of the other person to recognize the phonemes of the professional singer and the other person corresponding to the phoneme of each single note. In step S8, the synthesizing circuit 34 synchronizes the other person's phoneme with the timing signal from the timing extracting circuit 37 and modulates the pitch according to the reference pitch signal 9 to perform D / A.
Output to the converter 32. In step S9, it is determined whether or not the reference pitch signal 9 is finished. If it is not finished, the process returns to step S3, and steps S3 to S3 are performed.
Repeat 9 to continue the modulation of the pitch of the song until the end of the song.

When the user's singer stops singing, the timing signal of step S8 is not generated, so the synthesis circuit 34
No longer modulates the pitch of others' voices, and no mimicry is generated. Therefore, even if an amateur karaoke singer actually sings, the song is heard as if "Kato Tokiko" was singing, and the "Kato Tokiko" song also progresses in synchronization with the progress of vocalization of the karaoke singer. become. The phoneme uttered by "Toriko Kato" is stored in the phoneme table 38 of another person. Since the phoneme of a professional singer is output instead of the voice of a karaoke user, the enjoyment of singing a song of a famous singer can be enjoyed.

[Brief description of the drawings]

FIG. 1 is a block diagram of a voice conversion circuit according to a first embodiment of the present invention.

FIG. 2 is an overall block diagram of the karaoke apparatus according to Embodiment 1 of the present invention.

FIG. 3 is a diagram of standard phoneme waveforms in the phoneme table of the speech conversion circuit of the present invention.

FIG. 4 is a diagram of a phoneme waveform of another person in the phoneme table of the speech conversion circuit of the present invention.

FIG. 5 is a flowchart illustrating an operation of the karaoke device according to the first embodiment.

FIG. 6 is a configuration diagram of song data processed by the karaoke device according to the first embodiment.

[Explanation of symbols]

1 karaoke playback unit, 2 host, 3 communication line,
4 separation circuit, 5 voice conversion circuit, 6 song data,
7 genre code, 8 performance data, 9 reference pitch signal, 11 communication control circuit, 12 bus, 13, 1
6, 36 CPU, 14, 21 Work RAM, 17
Sound source mixer, 18 speakers, 19 microphones,
20 storage device, 22 character generation circuit (VDP), 2
3 image mixer, 24 laser death (LD) changer, 25 display device, 26 input / output device, 30
A / D converter, 31 voice analysis circuit, 32 D /
A converter, 33 standard phoneme table, 34 synthesis circuit, 35 phoneme conversion circuit, 37 timing extraction circuit, 38 star phoneme table, 39 phoneme recognition circuit.

Claims (2)

[Claims] 1. A voice conversion circuit that analyzes song data of a user's voice input from a microphone and replaces the user's song data with the song data of another person, or adds the song data of another person to the song data of the user. 2. A karaoke song data is composed of a performance signal and a reference pitch signal, a separation circuit for separating the song data into a performance signal and a reference pitch signal, and a singer's voice is converted into song data. A microphone, a phoneme table of another person that stores phonemes uttered by another person, a voice analysis circuit that extracts characteristic parts such as a formant by analyzing the song data, and a phoneme recognition circuit that recognizes a single phoneme from this formant and the like.
A phoneme conversion circuit that converts the phonemes of this song data into phonemes of another singer by referring to the phoneme table of the other person, a timing extraction circuit that generates a timing signal for each single note from the song data, and a timing signal matching this timing signal. And a synthesizing circuit for modulating the pitch of the singer's phoneme by the reference pitch signal, a sound source mixer for mixing the performance signal and the modulated phoneme signal, and an output signal of the sound source mixer for the singer's phoneme of another person. Karaoke device composed of a singing voice and a speaker for playing a musical instrument.