JP2019101071A - Singing evaluation device, singing evaluation program and karaoke device - Google Patents

Abstract

To perform singing evaluation with high accuracy while suppressing a processing load even for music which has no singing evaluation information (main melody information) which becomes a reference for singing evaluation.SOLUTION: A singing evaluation device includes: singing pitch vicinity range calculation means for performing frequency analysis of a singing voice signal inputted from a microphone and for calculating a singing pitch vicinity range which is a frequency range of the pitch vicinity of the singing voice signal; model singing voice characteristic calculation means for calculating a model singing characteristic in the singing pitch vicinity range from an accompaniment sound containing the model singing sound; and evaluation means for performing evaluation of the singing voice signal by comparing the singing voice characteristic of the singing voice signal and the model singing characteristic.SELECTED DRAWING: Figure 6

Description

  The present invention relates to a song evaluation apparatus, song evaluation program, song evaluation method, and karaoke apparatus for performing song evaluation when performing singing along with accompaniment sounds to be reproduced such as karaoke.

  Conventionally, some karaoke apparatuses that can enjoy singing along with accompaniment sounds are provided with a scoring function as a singing evaluation. In general, the scoring function in the karaoke apparatus is performed by comparing the main melody information in the accompaniment information with the singing voice signal input from the microphone. Therefore, with a sound source that does not have main melody information, such as a CD, the scoring function can not be used.

  Patent Document 1 discloses a voice evaluation device that can perform scoring (song evaluation) even for a sound source such as a CD that does not have main melody information. In this voice evaluation apparatus, an evaluation target pitch extracted along a time axis from voice data input from a microphone, and a plurality of sounds extracted along a time axis from audio data including a plurality of sounds, The voice data is evaluated by comparing with the reference pitch of.

JP, 2016-122164, A

  According to the voice evaluation device disclosed in Patent Document 1, it is possible to perform song evaluation even with a sound source that is not provided with a main melody (correct data), such as a commercially available music CD. However, in patent document 1, since it is necessary to extract several reference | standard pitch, it is assumed that the processing load required for song evaluation will become large. In addition, it is conceivable that the processing load required for comparison becomes large in order to compare the evaluation target pitch with the reference pitch of all the sounds included in the audio data, such as the main vocal, chorus, and various instruments. Moreover, if it does not compare with a suitable reference pitch among several reference pitch, song evaluation may become unsuitable.

  The present invention takes such circumstances into consideration, and even if the sound source does not have main melody information such as CD, etc., it is possible to perform song evaluation with high accuracy and reduced processing load. It aims at providing a song evaluation device, a song evaluation program and a karaoke device that can be performed.

Therefore, the song evaluation device according to the present invention is characterized in that the following configuration is adopted. Sing voice pitch proximity range calculation means for performing frequency analysis of a singing voice signal input from the microphone and calculating a singing voice pitch proximity range which is a frequency range near the pitch of the singing voice signal;
Model singing voice characteristic calculation means for calculating model singing characteristics within the vicinity of the singing pitch from the accompaniment sound signal including the model singing sound;
An evaluation means for evaluating a singing voice signal by comparing a singing voice characteristic of the singing voice signal with a model singing characteristic is provided.

Furthermore, in the song evaluation device according to the present invention,
The song pitch vicinity range is a predetermined distribution based on the pitch of the singing voice signal in at least one of the frequency axis direction and the time axis direction.

The song evaluation device according to the present invention is
Singing sound pitch vicinity range calculation means for calculating a singing sound pitch vicinity range from an accompaniment sound signal accompanied by information that can specify the singing sound pitch vicinity range or the singing sound pitch vicinity range;
Model singing voice characteristic calculation means for calculating model singing characteristics within the vicinity of the singing pitch from the accompaniment sound signal including the model singing sound;
It is characterized by comprising: evaluation means for evaluating a singing voice signal by comparing singing voice characteristics of singing voice signals inputted from a microphone with model singing characteristics.

The karaoke apparatus according to the present invention is
Any one song evaluation device mentioned above,
And reproducing means for reproducing an accompaniment sound signal including a model singing sound.

The song evaluation program according to the present invention is
Singing voice height vicinity range calculation processing of performing frequency analysis of the singing voice signal input from the microphone and calculating a singing voice pitch vicinity range which is a frequency range near the pitch of the singing voice signal;
Model singing voice characteristic calculation processing for calculating model singing characteristics within the vicinity of the singing pitch from the accompaniment sound signal including the model singing sound;
It is characterized in that it is possible to execute an evaluation process of evaluating a singing voice signal by comparing a singing voice characteristic of the singing voice signal and a model singing characteristic.

The song evaluation program according to the present invention is
Singing sound height vicinity range calculation processing for calculating a singing sound height vicinity range from an accompaniment sound signal to which information capable of specifying a singing sound height vicinity range or a singing sound height vicinity range is attached;
Model singing voice characteristic calculation processing for calculating model singing characteristics within the vicinity of the singing pitch from the accompaniment sound signal including the model singing sound;
It is characterized in that it is possible to execute an evaluation process of evaluating a singing voice signal by comparing a singing voice characteristic of a singing voice signal input from a microphone with a model singing characteristic.

  According to the song evaluation apparatus, the song evaluation program, and the karaoke apparatus according to the present invention, the accompaniment sound signal is used even when singing is performed using an accompaniment sound signal such as a CD sound source with no main melody information attached. By calculating model singing voice characteristics from the above, it becomes possible to perform singing evaluation such as scoring. Then, when calculating the model singing voice characteristic from the accompaniment sound signal, it is not necessary to cover the entire frequency range of the accompaniment sound signal because the calculation is performed using the singing sound pitch vicinity range to be sung by the user. While suppressing, it becomes possible to perform song evaluation with high accuracy.

  Further, the singing voice pitch vicinity range used in the singing voice evaluation device, the singing voice evaluation program and the karaoke device according to the present invention has a predetermined distribution based on the singing voice height in the frequency axis direction. By giving a predetermined distribution to the singing voice height range in this way, it is possible to perform a more accurate singing evaluation by weighting the model singing voice characteristics.

  Furthermore, the song sound height vicinity range used by the song evaluation apparatus, the song evaluation program, and the karaoke apparatus according to the present invention has a predetermined distribution based on the song sound height in the time axis direction. By considering the time axis direction of the singing voice height range in this manner, it is possible to perform singing evaluation with higher accuracy.

  Further, according to the song evaluation device, the song evaluation program, and the karaoke device according to the present invention, by adding the information that can specify the singing sound pitch vicinity range or the singing sound pitch vicinity range to the accompaniment sound signal, Even if singing is performed using an accompaniment sound signal such as a CD sound source or the like without accurate main melody information, song evaluation such as scoring can be performed by calculating model singing voice characteristics from the accompaniment sound signal. It is possible to do Then, when calculating the model singing voice characteristic from the accompaniment sound signal, it is not necessary to cover the entire frequency range of the accompaniment sound signal because the calculation is performed using the singing sound pitch vicinity range attached in advance. It is possible to perform accurate song evaluation while suppressing

The figure which shows the structure of the karaoke system of this embodiment A diagram showing various data configurations used in the karaoke system of the present embodiment Flow chart showing the performance sound probability density calculation process of the present embodiment Flow chart showing music reproduction processing of the present embodiment Flow chart showing the evaluation process of the present embodiment The figure for demonstrating the song sound vicinity vicinity of this embodiment Diagram for explaining a two-dimensional normal distribution region of pitch probability density according to the present embodiment The figure for demonstrating the model song characteristic of this embodiment A figure for explaining comparison processing of this embodiment

  FIG. 1 is a diagram showing the configuration of the karaoke system of the present embodiment. The karaoke system in the present embodiment is configured to include a karaoke device 2 (sometimes called a commander) and a remote control device 1. The karaoke apparatus 2 and the remote control apparatus 1 are communicably connected to form a network using the LAN 100 and the access point 130.

  The karaoke apparatus 2 installed in a store such as a karaoke box includes a sound control unit 25 as a playing unit for playing music. Moreover, the karaoke apparatus 2 is provided with the operation part 21 which receives the various input from a user. The karaoke apparatus 2 includes an operation processing unit 22 that interprets an input from the operation unit 21 and transmits the interpretation to the CPU 30. Moreover, the karaoke apparatus 2 is equipped with the hard disk 32 as a memory | storage part which memorize | stores various information. The karaoke apparatus 2 is provided with a LAN communication unit 24a as a communication means connected to the LAN 100 to join the network. In addition, the karaoke apparatus 2 according to the present embodiment also includes the wireless LAN communication unit 24b, and instead of wired network connection using the LAN communication unit 24a, wireless network connection using the wireless LAN communication unit 24b is performed. Is also possible.

  In addition, the karaoke apparatus 2 includes video reproduction means for displaying a lyric image and a background image on the monitor 41. The video playback means is a video playback unit 29 for playing back a video based on video information, a video RAM 28 for temporarily storing a video to be played back, a video for superposing a lyric telop on the played video, and applying a video effect. The control unit 31 is configured.

  Furthermore, in the karaoke apparatus 2, in addition to the monitor 41 connected to the outside, various information can be displayed on the touch panel monitor 33. The touch panel monitor 33 is configured by superimposing a display unit 35 for displaying the video information input from the video control unit 31 and a touch panel 34 for outputting the position of the touch input to the operation processing unit 22. The touch panel monitor 33 is disposed on the front surface of the casing of the karaoke apparatus 2 or the like, and functions as an input unit, similarly to the operation unit 21 of the karaoke apparatus 2 or the touch panel monitor 11 of the remote control device 1. The user can perform various operations on the karaoke apparatus 2 such as making the karaoke apparatus 2 directly make a reservation by selecting music by the touch panel monitor 33.

  Furthermore, the karaoke apparatus 2 is configured to include a CPU 30 for integrally controlling each component and a control unit including a memory 27 for temporarily storing information necessary for executing various programs.

  With such a configuration, the karaoke apparatus 2 executes various processes, but as main functions of the karaoke apparatus 2, it is possible to execute music reservation processing, music reproduction processing, and the like. The song reservation process is a process for designating and reserving a song based on a designation from a user, and is executed in cooperation with the remote control device 1. The reservation information formed in the music selection process of the remote control device 1 is transmitted to the karaoke device 2. The karaoke apparatus 2 registers the received reservation information in the reservation table in the memory 27. The music reproduction process is a process for reproducing a reserved music, and is a process in which a music performance process and a lyric display process are performed in synchronization.

  The music performance process is a process for causing the sound control unit 25 to perform a performance based on the performance information included in the music information. The music played by the sound control unit 25 is emitted from the speaker 42 together with the singing voice input from the microphones 43a and 43b. The lyric display process is a process of performing song assistance by displaying the lyric information included in the music information on the monitor 41. It is also possible to execute background video display processing in which background video is superimposed and displayed on the lyrics displayed in the lyrics display processing.

  On the other hand, the remote control device 1 can execute music selection processing for searching for music based on an instruction from the user and transmitting reservation information to the karaoke apparatus 2 for the music instructed to be reproduced. Further, the remote control device 1 can receive various information from the karaoke device 2 or the host device 5 connected to the Internet, and can execute various processes. In the present embodiment, the operation unit 17 and the touch panel monitor 11 are provided as user interfaces for receiving various instructions from the user. The touch panel monitor 11 is configured to have a display unit 11a and a touch panel 11b, displays various interfaces on the display unit 11a, and can receive touch input from a user.

  Further, the remote control device 1 controls the memory 14 as a storage unit for storing a database required for music selection processing, various programs, and various information generated along with the execution of the program, and controls these configurations collectively. The remote control side control unit of The remote control control unit interprets the input from the CPU 15, the video control unit 13 for forming a video to be displayed on the touch panel monitor 11, the video RAM 12 for temporarily storing the video information to be displayed, the touch panel monitor 11 or the operation unit 17 And an operation processing unit 18 for transmitting the information to the CPU 15.

  The remote control device 1 is connected to the network configured by the LAN 100 by being wirelessly connected to the access point 130 by the wireless LAN communication unit 16. Each remote control device 1 is associated in advance with a specific karaoke device 2. Various commands output from the remote control device 1 are received by the associated karaoke device 2.

  According to the configuration of the remote control device 1 as described above, the karaoke device 2 is provided by receiving various inputs from the user from the touch panel monitor 11 or the operation unit 17 and providing various information by displaying the video information on the touch panel monitor 11. It is possible to perform various processing such as music selection processing of transmitting reservation information to be output.

  The karaoke apparatus 2 of the present embodiment can reproduce two types of music information. One type (A type) of music information is music information including song evaluation information for performing song evaluation. As conventionally known, in the song evaluation at the time of reproducing such music information, the song evaluation information as the main melody information is compared with the singing voice signal input from the microphone, and the degree of coincidence is determined. It is possible to calculate a scoring value based on that.

  FIG. 2A is a diagram showing a data configuration of music information (A type) according to the present embodiment. The music information includes meta information including various information related to the music information, and actual information for executing various processes such as performance and display of lyrics. The meta information includes music related information such as a music ID for identifying music information, a music title, a singer name, and a genre. The music related information can be used as a search target item when the user searches for music. The actual information of the music information includes performance information, lyric information, background video information, and the like. The performance information is information for playing the accompaniment sound of karaoke, which includes control information for an electronic musical instrument based on the MIDI standard, or compressed audio information obtained by recording an actual performance. The lyric information is information that is displayed in synchronization with the performance information for song assistance, and may be configured to change the color of the displayed lyrics in synchronization with the performance. The song evaluation information is information for evaluating a user's singing voice at the time of music reproduction, and is configured to include a melody to be sung and the like. When performing song evaluation at the time of music reproduction, it is possible to calculate a score or the like by comparing the singing voice input to the microphones 43a and 43b with the singing evaluation information.

  The other type (B type) of music information is music information that does not include song evaluation information, for example, music information that does not have song evaluation information such as music recorded on a CD. Conventionally, at the time of reproduction of such music information, it is difficult to perform song evaluation because the song evaluation information is not included. In the karaoke apparatus 2 of this embodiment, it is possible to perform song evaluation also about such music information.

  FIG. 2 (B) is a diagram showing a data configuration of music information (B type) of the present embodiment. Similar to the music information of FIG. 2A, the music information includes meta information including various information related to the music information, and actual information for executing various processes such as performance and lyrics display. The meta information is the same as the music information of FIG. 2A, and thus the description thereof is omitted here. The actual information includes performance information including an accompaniment sound (corresponding to the "accompaniment sound signal" in the present invention) and video information. The performance information includes an accompaniment sound obtained by recording an actual performance like a commercially available CD. Further, the performance information includes not only the accompaniment sound but also the singing voice by the singer (corresponding to the “model singing voice” in the present invention). The performance information in the music information (type B) is in a state where the accompaniment sound and the singing voice are mixed. The video information includes background video and lyrics displayed in synchronization with the accompaniment sound. The karaoke apparatus 2 according to the present embodiment can perform song evaluation using performance information even for music information (B type) that does not include song evaluation information.

  In the present embodiment, the playing sound probability density obtained in advance based on the playing information is included in the music information. The performance sound probability density is information obtained by frequency analyzing the performance sound information including the singing sound in addition to the accompaniment sound, and includes the pitches of various musical instruments used for accompaniment in addition to the pitch of the singing sound. Information.

  In addition, as music information (B type) not including song evaluation information to be directly compared, such as main melody information, not only the form including the playing sound probability density as in this embodiment, but also the playing sound probability density It may be music information not included. In that case, the playing sound probability density is calculated at the time of the music playing process.

  FIG. 3 is a flowchart showing the playing sound probability density calculation processing of the present embodiment. The performance sound probability density used in the present embodiment is information obtained by analyzing performance information (corresponding to "accompaniment sound signal" of the present invention) including model song sound and accompaniment sound, and is disclosed in Japanese Patent No. 3413634. It is calculated according to the pitch estimation method described in the publication. In this embodiment, the performance information included in the music information (B type) is processed in advance and included in the music information (B type). The song sound probability density may be calculated and song evaluation may be performed. In that case, the music information (B type) does not include the playing sound probability density.

  The performance sound probability density calculation process is started by acquiring performance information from music information to be analyzed (S100). In the pitch estimation method described in Japanese Patent No. 3413634, the locus of the fundamental frequency is obtained for each of the melody line (song sound) and the baseline (bass sound) by passing the BPF, but in the present embodiment, Performance information of all bands is targeted without passing through the BPF. In order to improve analysis accuracy, zero padding (zero padding) is performed as pre-processing before performing frequency analysis (FFT) (S101). Then, frequency analysis (S102) is performed on the zero-padding performance information. In this embodiment, although FFT is used as a frequency analysis method, it is also possible to use other frequency analysis methods.

  The performance sound probability density included in the performance information is calculated using the EM algorithm (Expectation-Maximization) for the frequency-analyzed performance sound information. The EM algorithm is executed by alternately and repeatedly applying E step (Expectation Step, S103) and M step (Maximization Step, S104). At that time, the performance sound probability density with high accuracy is calculated for the singing sound by applying the sound model of the harmonic structure of the singing sound. The calculated performance sound probability density is stored in the music information to be analyzed (S105).

  In the present embodiment, the musical performance probability density calculated in this manner is used to include song evaluation information that directly indicates the pitch of the singing voice such as the main melody information, like the musical composition information (A type). It is possible to perform song evaluation properly even if it is not music information.

  The music reproduction process including the evaluation process of performing the song evaluation on the karaoke apparatus 2 of the present embodiment will be described. FIG. 4 is a flowchart showing the music reproduction process of the present embodiment. The karaoke apparatus 2 reserves a music piece based on an operation on an input unit such as the remote control device 1 or the touch panel monitor 33. FIG. 2C shows a data configuration of reservation information stored in the memory 27 of the karaoke apparatus 2 based on the reservation operation. The reservation information includes, in addition to the music ID for identifying the music information, a user ID indicating a user who has made a reservation, a pitch setting value based on a pitch setting at the time of reservation, and the like.

  The karaoke apparatus 2 checks the reservation table stored and managed in the memory 27 and confirms the music to be reproduced (S201). Next, when there is a music to be reproduced (S202: Yes), the music information corresponding to the music ID in the reservation information is read out and reproduction of the music is started (S203). During the reproduction period of the music information, a standard evaluation process (S205) for evaluating the user's singing or an evaluation process (S300) is executed. When it is determined that the music information being reproduced is music information (A type) including singing evaluation information (S204: Yes), a standard evaluation process (S205) is performed. This standard evaluation process (S205) is a song evaluation that has been conventionally performed, and is performed by comparing song evaluation information (main melody information) included in the music information with the singing voice signal input from the microphones 43a and 43b. It is a process which calculates evaluation results, such as a graded value. In addition, since the standard evaluation process (S205) is a process which is well known conventionally, the detailed description here is omitted.

  On the other hand, when it is determined that the music information being reproduced is music information (B type) not including singing evaluation information (S204: No), an evaluation process (S300) is performed. This evaluation process (S300) is a process that is a feature of the present embodiment, and is a process of performing song evaluation based on the playing sound probability density included in the music information (B type). Details of the evaluation process (S300) will be described later. When the reproduction of the music ends for any of A type and B type (S206, S207: Yes), the evaluation result determined in the standard evaluation process (S205) or the evaluation process (S300) is displayed on the monitor 41 etc. Will notify you of your singing ability. Then, the process returns to the beginning of the music reproduction process, and confirmation of the music to be reproduced next is performed.

  FIG. 5 is a flowchart showing the evaluation process (S300) of the present embodiment. In the evaluation process (S300), evaluation is performed based on the singing voice signals input from the microphones 43a and 43b and the playing sound probability density acquired from the music information. For the singing voice signal (S303) input from the microphones 43a and 43b, the probability density is calculated for the fundamental frequency of the singing voice signal (S304). At that time, it is determined whether or not the current time is a singing section (S305). The determination of the singing section may be made using the singing voice signal itself, or may be made with reference to the performance sound information or the performance sound probability density. Alternatively, the determination may be made using the progress status of the lyric information included in the music information.

  If it is determined to be a singing section (S305: Yes), a two-dimensional normal distribution area is calculated for the probability density of the fundamental frequency calculated in S304 (S306). FIG. 6 is a diagram for explaining a two-dimensional normal distribution region of pitch probability density according to the present embodiment. FIG. 6A is a graph showing the probability density of the fundamental frequency, in which the probability density (certainty) is given to the fundamental frequency fc at the time of analysis. In the present embodiment, the evaluation accuracy is improved by applying a predetermined distribution with the fundamental frequency fc as a center (reference). In the present embodiment, normal distribution is applied to each of the frequency direction and the time axis direction to create a two-dimensional normal distribution region shown in FIG. 6 (B). The distribution based on the fundamental frequency fc is not limited to the two-dimensional normal distribution, and may be only in the frequency axis direction or only in the time axis direction. Also, the distribution form is not a normal distribution, and various other distribution forms may be adopted. The created two-dimensional normal distribution region is used to extract model song characteristics in the performance sound probability density and to evaluate a singing voice signal.

  As described with reference to FIG. 3, the performance sound probability density includes information such as accompaniment sound and back chorus in addition to the singing sound required for song evaluation. Therefore, if this performance sound probability density is used as it is for evaluation, it is possible that the evaluation accuracy is lowered. In this embodiment, the model sound characteristics are extracted by filtering the performance sound probability density with the singing voice signal, and the model singing characteristics are used for comparison with the singing voice signal to improve the evaluation accuracy. And This means that although there are differences in the skill of the singing voice depending on the singers, if singing properly, the characteristics will be similar to the model singing characteristics, and conversely, if not properly singing, performance The reason is that the model singing characteristics can not be properly extracted from the sound probability density.

  In the evaluation process, the performance sound probability density of the section to be evaluated is acquired (S301), and a singing sound pitch vicinity range based on the fundamental frequency of the singing voice signal is extracted from the acquired singing sound height probability density, Model singing characteristics are calculated (S302). In this embodiment, the performance sound probability density is multiplied by the two-dimensional normal distribution area of the singing voice signal to simultaneously perform the extraction of the singing sound pitch vicinity range and the calculation of the model singing characteristics. The user who sings may sing with different octaves with respect to the singing voice to be originally singing. Therefore, the singing sound pitch vicinity range may be configured to extract the singing sound pitch vicinity range for each of a plurality of octaves. Alternatively, it is also possible to extract for each of a plurality of octaves, and extract an octave having a high proportion of the playing sound probability density as the singing sound pitch vicinity range.

  FIG. 7 is a diagram for explaining the singing sound pitch vicinity range of the present embodiment. FIG. 7A is a graph showing the temporal change of the singing voice pitch probability density. FIG. 7 (B) is a graph showing a singing tail pitch vicinity range for the graph of FIG. 7 (A). The song pitch vicinity range shown in FIG. 7B is a range based on the fundamental frequency of the user who sings, and the singing voice pitch probability density outside this range is not the target of the song evaluation. Therefore, it is possible to reduce the evaluation range of the frequency and to reduce the load required for the evaluation process.

  FIG. 8 is a diagram for explaining the calculation of model singing characteristics of the present embodiment. FIG. 8A shows a performance sound probability density at a certain time and a two-dimensional normal distribution area calculated from a singing voice signal. The model singing characteristics shown in FIG. 8B are calculated by multiplying the two-dimensional distribution area by the playing sound probability density.

  The song evaluation is performed by comparing the calculated model song characteristics with the two-dimensional normal distribution region (calculating the degree of coincidence) (S307). FIG. 9 is a view for explaining comparison processing (S307) of the present embodiment. FIG. 9 (A) shows model singing characteristics calculated based on the performance sound information, and FIG. 9 (B) shows a two-dimensional normal distribution area calculated based on the singing voice signal. The comparison process is executed by calculating the degree of coincidence between the two. The higher the degree of agreement between the two, the higher the degree of coincidence, ie, the song evaluation. In the present embodiment, the two-dimensional normal distribution region is provided in both the frequency direction and the time axis direction. Therefore, it is possible to perform the evaluation in consideration of the time axis direction. When considering the time-axis direction, the model singing characteristics are also converted in two dimensions, such as using a temporal transition for the model singing characteristics, or based on the temporal transition of the two-dimensional normal distribution region It is preferable to compare and compare in a one-dimensional distribution region with respect to the frequency axis.

  A score value is calculated by integrating the comparative score value calculated in the comparison process (S307) (S308). By repeatedly executing the evaluation process (S300) during the reproduction period of the music information, the score value of the entire music is calculated. When it is determined that the reproduction of the music is finished (S207: Yes), the score value calculated as the evaluation result of the evaluation process (S300) is displayed on the monitor 41 or the like to score the singing user as a singing ability The value is notified.

  As described above, according to the karaoke system of the present embodiment, even if the song information does not include song evaluation information such as main melody information, the song pitch which is a frequency range near the pitch of the singing voice signal in the performance information. By setting the vicinity range as an evaluation target, it is possible to suppress the processing load and to perform song evaluation with high accuracy.

  In the present embodiment, the probability density is calculated for both the singing voice signal and the performance sound information and is used as a comparison target. However, not only the probability density but also various indexes can be used for each.

  Further, in the present embodiment, the musical performance probability density obtained by analyzing the musical performance information is attached to the musical composition information (B type). However, the musical performance probability density is not incidental in advance, and the musical composition reproduction processing is performed. It may be calculated during the period. In addition, the information that can specify the song pitch vicinity range or the song pitch vicinity range is attached to the music information in advance instead of calculating using the singing voice signal that the user sings during the music reproduction process. It is good to keep it. In that case, the singer prepared by the music information providing side (manufacturer's side) sings, and the song sound vicinity range obtained from the singing voice signal is attached to the music information, or the singing voice It is conceivable to attach information capable of specifying a singing sound pitch vicinity range obtained from a signal to music information.

  As described above, by attaching information capable of specifying the song sound vicinity range or the song sound vicinity range to the music information, a professional singer can be performed well on the music information provider side (manufacturer side). It is possible to use a singer who can sing appropriately without having to provide a suitable singing pitch neighborhood range.

  As mentioned above, although this invention was demonstrated using the karaoke system, this invention is not limited to a karaoke system. A song evaluation device that performs song evaluation inside or outside the karaoke device also belongs to the category of the present invention. In addition, at present, a form for performing karaoke with a smartphone by installing a karaoke application (program) on a smartphone, or a program for karaoke to be executed in a game machine is well known. The song evaluation program implemented in the program for karaoke which realizes the function of the present invention by being installed in such various information processing apparatuses also belongs to the category of the present invention. Furthermore, it goes without saying that the singing evaluation method also belongs to the category of the present invention.

1: Remote control device 24b: Wireless LAN communication unit 2: Karaoke device 25: Sound control unit 5: Host device 27: Memory 11: Touch panel monitor 28: Video RAM
11a: Display unit 29: Video reproduction unit 11b: Touch panel 30: CPU
12: Video RAM 31: Video control unit 13: Video control unit 32: Hard disk 14: Memory 33: Touch panel monitor 15: CPU 34: Touch panel 16: Wireless LAN communication unit 35: Display unit 17: Operation unit 41: Monitor 18: Operation Processing unit 42: Speaker 21: Operation unit 43a, 43b: Microphone 22: Operation processing unit 130: Access point 24a: LAN communication unit

Claims (6)

Sing voice pitch proximity range calculation means for performing frequency analysis of a singing voice signal input from the microphone and calculating a singing voice pitch proximity range which is a frequency range near the pitch of the singing voice signal;
Model singing voice characteristic calculation means for calculating model singing characteristics within the vicinity of the singing pitch from the accompaniment sound signal including the model singing sound;
What is claimed is: 1. A song evaluation apparatus comprising: evaluation means for evaluating a singing voice signal by comparing a singing voice characteristic of a singing voice signal with a model singing characteristic. The song evaluation device according to claim 1, wherein the singing sound pitch vicinity range is a predetermined distribution based on the pitch of the singing voice signal in at least one of the frequency axis direction and the time axis direction. . Singing sound pitch vicinity range calculation means for calculating a singing sound pitch vicinity range from an accompaniment sound signal accompanied by information that can specify the singing sound pitch vicinity range or the singing sound pitch vicinity range;
Model singing voice characteristic calculation means for calculating model singing characteristics within the vicinity of the singing pitch from the accompaniment sound signal including the model singing sound;
What is claimed is: 1. A song evaluation apparatus comprising: an evaluation unit that evaluates a singing voice signal by comparing singing voice characteristics of a singing voice signal input from a microphone with a model singing characteristic. The song evaluation apparatus according to any one of claims 1 to 3;
A karaoke apparatus comprising: reproduction means for reproducing an accompaniment sound signal. Singing voice height vicinity range calculation processing of performing frequency analysis of the singing voice signal input from the microphone and calculating a singing voice pitch vicinity range which is a frequency range near the pitch of the singing voice signal;
Model singing voice characteristic calculation processing for calculating model singing characteristics within the vicinity of the singing pitch from the accompaniment sound signal including the model singing sound;
A song evaluation program characterized in that an evaluation process for evaluating a singing voice signal by comparing a singing voice characteristic of a singing voice signal with a model singing characteristic is executable. Singing sound height vicinity range calculation processing for calculating a singing sound height vicinity range from an accompaniment sound signal to which information capable of specifying a singing sound height vicinity range or a singing sound height vicinity range is attached;
Model singing voice characteristic calculation processing for calculating model singing characteristics within the vicinity of the singing pitch from the accompaniment sound signal including model or silence;
A song evaluation program characterized in that it is possible to execute an evaluation process of evaluating a singing voice signal by comparing singing voice characteristics of a singing voice signal input from a microphone with a model singing characteristic.