JP2001013976A - Karaoke device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To output phonated sound of a right vocalism from a speaker 24 even if a singer makes a mistake in the vocalism of the KARAOKE strains. SOLUTION: A voice recognition section 16 recognizes the vocalism of the singer by input signals from a microphone 11. Meanwhile a data extract section 17 extracts right vocalism out of reproduced signals of the KARAOKE strains from an optical disk player 12. A collation section 18 collates inputs from the voice recognition section 16 and the data extract section 17 and sends collated result to a waveform editorial section 19. When collated results coincide, the waveform editorial section 19 outputs phonated sound of the singer as it is from the microphone 11 to a mixing section 23. When collated results do not coincide, the phonated sound of the singer from the microphone 11 is corrected based on the vocalism from the data extract section 17 and then is outputted to the mixing section 23. The mixing section 23 mixes reproduced sound of the KARAOKE strains from the optical disk player 12 with the output from the waveform editorial section 19 and outputs the mixed sound to the speaker 24.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a karaoke apparatus, and more particularly to a karaoke apparatus for improving singing of a singer.

[0002]

2. Description of the Related Art Singers of karaoke songs often make mistakes in the lyrics or incorrectly use singing techniques during singing.

[0003]

The singer learns the lyrics while practicing the karaoke tune, but it is convenient if the singer informs the lyric uttered at the time of the practice. Also, even if the singer mistakes the lyrics and singing technique,
The audience sometimes wants to show off with the right lyrics and singing techniques. There is no conventional karaoke apparatus that satisfies such a demand.

[0004] It is an object of the present invention to provide a karaoke apparatus that can address the above-mentioned problems.

[0005]

A karaoke apparatus according to the present invention.
According to (10), during the playback of the karaoke song, the singer's utterance in each melody part is recognized, and the recognized singer's vocal sound is compared with the correct lyrics of each melody part. To the singer.

[0006] For reproducing karaoke songs, for example, predetermined karaoke songs recorded on a predetermined recording medium (13) such as an optical disk such as a laser disk, a CD, and a DVD, or a music tape are reproduced, or distributed by communication. This is performed by outputting a reproduction signal generated to the music data of the incoming karaoke song. Karaoke song data usually includes not only the melody but also the video displayed as a background screen on the display,
Similarly, the display contains information on the lyrics to be displayed as subtitles, and the correct lyrics for each melody part of the karaoke song may be extracted from that information, or data dedicated to the correct lyrics has already been recorded. It can be obtained using software.

[0007] The coincidence or mismatch between the recognized singing voice and the correct lyric in each melody portion is determined by, for example, converting the recognized singing voice into text using well-known software and comparing it with the text of the correct lyric. be able to.

In each melody part, the case where the recognized singer's uttered sound does not match the correct lyrics means not only when the singer utters the wrong lyrics, but also when the singer forgets the lyrics and Nevertheless, this includes the case where no lyrics are uttered.

The notification of the lyric error to the singer may be performed immediately upon detection of the lyric error, or the singer may be able to know which melody part of the karaoke singer made the lyric. The recording may be made while the singer is singing, and the singer may check the recording after the singing. The notification of the lyric error to the singer further includes a case of notifying by a sound from the audio output means (24) such as a speaker, a case of notifying visually by a display, and a case of notifying by a printout. It is particularly convenient for the user to practice a karaoke song while driving a car while notifying the singer's erroneous lyric error while singing.

[0010] The singer utters the lyrics while watching the subtitles displayed as the lyrics on his or her memory or display in accordance with the reproduced sound of the karaoke song. The singer's utterance is recognized by a well-known voice recognition device, and it is checked whether the recognized singer's utterance matches the correct lyrics of each melody part,
If there is a mismatch, the singer is notified that the singer has mistaken the lyrics. Thereby, the singer can know the mistake of his / her lyrics accurately, and can practice the lyrics efficiently.

According to the karaoke apparatus (10) of the present invention, during the reproduction of the karaoke song, the singer's vocal sound in each melody part is recognized, and the recognized singer's vocal sound is compared with the correct lyrics of each melody part. When the two do not match, the correct lyrics are output from the voice output means (24) in a voice resembling the voice of the singer.

The generation of a correct utterance sound of a lyric that resembles a singer's voice is performed by, for example, “Singe of singing voice generation software”.
r (Trademark of Nippon Telegraph and Telephone Corporation) ", etc., using a text-to-speech engine equipped with a melody from the correct lyrics and voice-changing it to approach the singer's voice. In a situation where it is difficult to automatically process the voice after the voice change so as to resemble the voice of the singer, it is necessary for the singer or the like to adjust the state of the voice change before singing.

According to the karaoke apparatus (10) of the present invention, during the reproduction of the karaoke song, the singing technique of the singer in each melody part is recognized, and the recognized singer singing technique is compared with the model singing technique of each melody part. If they do not match, the singer's voice is given an example singing technique, and the voice output means (2
Output from 4).

The singing technique includes, for example, a vocal sound, a voice volume, a pitch, a lengthening, a vibrato, and a fixed pitch. The audio output means (24) is a speaker, a headphone, or the like. When the recognized singing singing technique does not match the singing technique of the example of each melody part, not only when the singer sang with the wrong singing technique, but also the singer forgets the singing technique and This includes singing without any singing technique, despite singing techniques.

The mismatch between the model singing technique and the singer's singing technique in each melody portion is detected, for example, as follows. A singer's vocalization is converted to a waveform editor (eg, GOLD
WAVE (trademark) or the like and a musical score generation device (Song Laita (trademark)). The musical score generating device has a function of generating a musical score (a scale and a singing technique are included in the musical score) from a vocal sound, and a musical score and a singing technique generated from a singer's vocal sound are registered in advance. The correct score is compared with the correct score, and when there is a difference, the data relating to the correct score is sent to the waveform editing device, and the vocal sound of the singer is edited in accordance with the correct singing technique. As a result, even if the singer makes a mistake in the singing technique, the utterance sound that is corrected to the correct singing technique is output from the score generating apparatus, and the voice output means (2
From 4), the utterance when the singer sang with the correct singing technique is output.

Instead of using such a waveform editing device and a musical score generating device, it is also possible to use a recording medium (13). In other words, not only the data of the karaoke song but also the data of the model singing technique are recorded in the recording medium (13) that records the karaoke song in association with each melody part of the karaoke song. For example, when the data of the karaoke song and the data of the model singing technique are recorded on different channels, the playback of the karaoke song is performed based on one channel of the data of the karaoke song, and synchronized with the playback. A model singing technique can be detected from the other channel. In the above-mentioned singing voice generation software Singer, singing techniques such as power adjustment and vibrato addition can be added for each phoneme. In Singer, the singing voice before applying the singing technique is a computer-synthesized sound.
The correct singing technique can be given to the singer's voice.

Thus, even if the singer does not know the singing technique of each melody part, or performs a singing technique different from the example, the singer's uttered sound arranged to the correct singing technique is output by the voice output means ( Since it is output from 24), even poor singers can perform songs with confidence in public.

[0018] According to the karaoke apparatus (10) of the present invention, a tune into which the model singing voice is blown is reproduced, the lyrics and the singing technique of each melody portion are extracted from the model singing voice, and the extracted melody portion of each melody portion is extracted. An extracted data recording medium (55) in which lyrics and singing techniques are recorded in association with each melody part is produced, and the correct lyric or model singing technique of each melody part is read from the extracted data recording medium (55).

It is desirable that the lyric data and singing technique in each melody portion be recorded in a karaoke song from the beginning, but such sources may be insufficient.
It reproduces the source in which the singing voice of a real singer or the like is blown in, extracts the lyrics and singing technique of each melody part, and produces an extracted data recording medium (55) in which it is recorded. In the collation processing between the singing voice of the singer during karaoke and the correct lyrics or singing technique, the data read from the extracted data recording medium (55) is used as the correct lyric or singing technique data, so that the correct lyric or singing technique can be used. The time required for the extraction process can be saved.

[0020]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of the karaoke apparatus 10. The microphone 11 is grasped by the singer of the karaoke song, converts the singer's voice into an electric signal, and the electric signal is sent to the sound recognition unit 16 and the waveform editing unit 19 of the sound processing device 15. The optical disk player 12 is a CD or D
An optical disk 13 such as a VD is set, a specified karaoke song is reproduced from a plurality of karaoke songs recorded on the optical disk 13, and the reproduced signal is sent to the data extraction unit 17 and the mixing unit 23 of the audio processing device 15. Output. The voice recognition unit 16 detects a singing voice and a singing technique from an input signal from the microphone 11. In recognizing a singer's utterance, a function such as voice-text conversion software, which is already known in personal computer software, is used. In recognizing a singer's singing technique, a function of well-known music score generation software (eg, Song Ryota) is used. Data extractor 17
Extracts the lyrics of each melody portion and the singing technique of the model from the reproduction signal from the optical disc 13. On the optical disc 13,
Text data is recorded for the caption of the lyrics, and the text data can be used as the correct lyrics for each melody part. Further, in order to extract the model singing technique from the reproduction signal from the optical disk 13, the voice of the model singer is detected from the channel of the model singer in the reproduction signal from the optical disk 13, and is detected by the voice recognition unit 16. In the same manner as in the case described above, processing is performed using well-known music score generation software (eg, Song Yota) to extract a model singing technique. The collation unit 18 collates the recognition result of the speech recognition unit 16 with the extraction result of the data extraction unit 17, and outputs information on whether they match or not to the waveform editing unit 19. If the recognition result of the voice recognition unit 16 and the extraction result of the data extraction unit 17 match, the waveform editing unit 19 outputs the singer's voice signal from the microphone 11 as it is to the mixing unit.
The data is output to the mixing unit 23, and if they do not match, the audio signal from the microphone 11 is edited based on the extracted data from the data extraction unit 17, and is output to the mixing unit 23. The waveform editing unit 19 uses, for example, the function of singing voice generation software “Singer (trademark of Nippon Telegraph and Telephone Corporation)”, and can add a singing technique to each vocal sound to the singer's voice. I have. The waveform editing section 19 also has a voice changer and a text-to-speech conversion function. Then, it is converted into a voice sound that resembles the singer's voice by a voice changer. In this way, when the singer makes a mistake in the lyrics or singing technique, the sound of the correct lyric and singing technique and a voice that resembles the singer's voice is mixed by the mixing section 23.
Output to In the mixing unit 23, the melody sound of the karaoke song from the optical disc player 12 and the singing voice of the singer from the waveform editing unit 19, the singing utterance corrected by a correct singing technique, or a voice similar to the singing voice are generated. The correct lyric voice is mixed and output to the speaker 24, and the mixed sound is output from the speaker 24. In this way, the output sound from the speaker 24 becomes the sound of the correct lyrics, together with the reproduced sound of the karaoke song, even though the singer mistakes or forgets the lyrics.

FIG. 2 shows a karaoke source creating apparatus with data.
It is a block diagram of 50. In the karaoke apparatus 10 shown in FIG. 1, while the karaoke song is reproduced by the optical disc player 12, the data extraction unit 17 extracts the correct lyrics and singing technique in the reproduced karaoke song. Processing requires considerable speed. On the other hand, if there is software that records the correct lyrics and singing technique data of each melody portion of the karaoke song to be reproduced, the recorded data of the software is read to read out the correct lyrics and singing technique of each melody portion. Data can be obtained, and complicated extraction processing can be omitted. In the data-added karaoke source creating apparatus 50, the music source 51 with a song (vocal) to be reproduced is, for example, an optical disk such as a laser disk or a DVD into which a real singer blows a song. The melody is played. The player 52 outputs the playback sound of the song-with-song source 51 output from both the left and right channels to the voice and singing technique extractor 53 and the vocal removal circuit 56. Although the vocal sound and the melody sound are mixed in the playback sound of the song-equipped music source 51 output from the player 52, as is well known, the frequency band of the vocal sound is limited. By passing the 51 reproduced sound through a predetermined band-pass filter, only the vocal sound and the melody sound can be easily extracted. Further, since the singer usually utters from the center of the left and right, as is well known, the difference between the left and right reproduced sounds of the player 52 before passing the reproduced sound of the player 52 through the band pass filter is known. , The quality of the extracted sound can be improved. Thus,
The voice and singing technique extractor 53 detects a correct lyric and singing technique from the extracted vocal sound of the singer and outputs it to the L channel terminal of the recording / reproducing apparatus 54. Also, the vocal removal circuit 56
Outputs only the remaining melody sound obtained by removing the vocal sound from the reproduction sound of the player 52 to the R channel terminal of the recording / reproducing device 54. The recording / reproducing machine 54 operates a music tape 55 (the music tape 55 may be another recording medium such as an MD or a CVD-RW, which can be freely recorded and played back). The recording is performed on the R channel and the L channel of the music tape 55 in synchronization with the technique. Thus, when the music tape 55 is reproduced next from the recording start position,
The melody sound of the karaoke tune in each melody portion and the data of the correct lyric and singing technique are output simultaneously (= in synchronization) from the R channel terminal and the L channel terminal of the recorder / reproducer 54, respectively. Recording / playback machine 54 with the produced music tape 55
Means that when performing karaoke, its R channel and L
The channels are connected to the mixing unit 23 and the matching unit 18, respectively. When using the recording / reproducing machine 54, the data extracting unit 17
Can be omitted.

FIG. 3 illustrates the relationship between the utterance state of the model of each melody part and the singing technique. As an example, the song adopts the song of Hibari Misora. "-" Is stretched, "-" is vibrato, and "from the river" to "... you" is understood as a singing technique with a constant pitch.

[Brief description of the drawings]

FIG. 1 is a block diagram of a karaoke apparatus.

FIG. 2 is a block diagram of an apparatus for creating a karaoke source with data.

FIG. 3 is a diagram exemplifying a relationship between a vocal state of a model of each melody part and a singing technique;

[Explanation of symbols]

 Reference Signs List 10 Karaoke apparatus 13 Optical disk (recording medium) 24 Speaker (audio output means) 55 Music tape (extracted data recording medium)

Claims (4)

[Claims] 1. During playback of a karaoke song, a singer's utterance in each melody part is recognized, and the recognized singer's vocal sound is compared with the correct lyrics in each melody part. A karaoke apparatus for notifying a singer. 2. During playback of a karaoke song, a singer's utterance in each melody part is recognized, and the recognized singer's vocal sound is compared with correct lyrics in each melody part. A karaoke apparatus characterized in that the karaoke apparatus outputs a voice resembling the voice of a singer from the voice output means (24). 3. During playback of a karaoke song, a singer's singing technique in each melody part is recognized, and the recognized singer's singing technique is compared with a model singing technique in each melody part. A karaoke apparatus characterized in that an example singing technique is applied to voice and output from voice output means (24). 4. Playing a song in which a model singing voice is blown, extracting lyrics and singing techniques of each melody portion from the sample singing voice, and corresponding the extracted lyrics and singing techniques of each melody portion to each melody portion. Extracted data recording medium (5
The karaoke apparatus according to any one of claims 1 to 3, wherein 5) is produced, and a correct lyric or a model singing technique of each melody portion is read from the extracted data recording medium (55).