Singing evaluation program, singing evaluation device, karaoke program, and singing evaluation method

Abstract

This invention provides a singing evaluation program, singing evaluation device, karaoke program, and singing evaluation method that more accurately determine the similarity to the original singing performance. [Solution] The singing evaluation program causes an information processing device to execute a first acquisition process to acquire singing data, a second acquisition process to acquire evaluation data for the song to be played, a singing noise component extraction process to extract singing noise components from the singing data acquired in the first acquisition process, and an evaluation process to evaluate the singing data based on the extracted singing noise components and the evaluation data.

Description

Technical Field

[0001] The present invention relates to a singing evaluation program for objectively evaluating singing, as well as a singing evaluation device, a karaoke program, and a singing evaluation method.

Background Art

[0002] Conventionally, karaoke, where people enjoy singing along with accompaniment, has been practiced in social settings and the like. In karaoke, alongside singing well, the skill of singing as close as possible to the original singer's performance may be required.

[0003] Patent Document 1 discloses a singing scoring display karaoke device that scores and displays the similarity between a model singing performance of a singer or the like and the singing performance of an amateur singer or the like. This singing scoring display karaoke device extracts audio feature points from the model singing signal and the singing signal, and performs a specific speaker word speech recognition process. The sound quality and the length of the uttered words obtained by this specific speaker word speech recognition process are compared, and a scoring video signal indicating the similarity between the model singing signal of a singer or the like and the singer's singing signal is output from the distance value of the deviation state in this comparison, and displayed on the screen of a monitor device.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] The karaoke device that displays singing scores, as described in Patent Document 1, compares the sound quality and length of spoken words obtained through specific speaker word speech recognition processing, and scores and displays the degree of similarity (degree of imitation) between the model singing (such as a singer) and the singer's singing based on the distance value of the separation state obtained from this comparison. However, when using such specific speaker word speech recognition processing, the degree of similarity (degree of imitation) between the model singing and the singer's singing is not always calculated. One of the objectives of the present invention is to more accurately determine the similarity between the model singer (original singer) and the singer. [Means for solving the problem]

[0006] Therefore, the singing evaluation program according to the present invention is The first acquisition process for obtaining singing data, A second acquisition process to obtain evaluation data for the song being played, The first acquisition process extracts singing noise components from the singing data, and The information processing device is instructed to perform an evaluation process that assesses the singing data based on the extracted singing noise components and evaluation data.

[0007] Furthermore, in the singing evaluation program according to the present invention, Singing noise components are components of singing data other than the pitch component.

[0008] Furthermore, in the singing evaluation program according to the present invention, The extraction process is, The first acquisition process involves obtaining the frequency spectrum of the singing data, A second acquisition process is performed, in which the envelope formed by connecting the peaks of the troughs in the acquired frequency spectrum is acquired as the singing noise component.

[0009] Furthermore, in the singing evaluation program according to the present invention, The evaluation data includes model singing noise components extracted from model singing data. The evaluation process involves performing a first comparison process that compares the extracted singing noise components with the model singing noise components to evaluate the singing data.

[0010] Furthermore, in the singing evaluation program according to the present invention, The singing noise component and the exemplary singing noise component are frequency spectra that show the level for each frequency.

[0011] Furthermore, in the singing evaluation program according to the present invention, The evaluation process assesses the singing data by comparing the shape of the extracted singing noise component with that of the model singing noise component.

[0012] Furthermore, in the singing evaluation program according to the present invention, The noise components of the model singing are divided into syllable units. The first comparison process compares the singing noise component with the model singing noise component corresponding to the syllables of the singing noise component.

[0013] Furthermore, the singing evaluation program according to the present invention is A second extraction process is performed to extract the singing noise component content rate from the singing data and singing noise components obtained in the first acquisition process. The evaluation process involves a second comparison process that compares the extracted singing noise component content with the model singing noise component content included in the evaluation data to evaluate the singing data.

[0014] Furthermore, in the singing evaluation program according to the present invention, The noise component content of the model singing is divided into syllable units. The second comparison process compares the percentage of singing noise components with the model singing noise components corresponding to the syllables with that percentage.

[0015] Furthermore, the singing evaluation device according to the present invention is The first acquisition process for obtaining singing data, A second acquisition process to obtain evaluation data for the song being played, an extraction process of extracting singing noise components from the singing data acquired in the first acquisition process, and an evaluation process of evaluating the singing data based on the extracted singing noise components and the evaluation data, are executed.

[0016] Further, the karaoke program according to the present invention a reproduction process of reproducing music data, a first acquisition process of acquiring singing data, a second acquisition process of acquiring evaluation data corresponding to the music data to be reproduced, an extraction process of extracting singing noise components from the singing data acquired in the first acquisition process, and an evaluation process of evaluating the singing data based on the extracted singing noise components and the evaluation data, are caused to be executed by an information processing apparatus.

[0017] Further, the singing evaluation method according to the present invention a first acquisition process of acquiring singing data, a second acquisition process of acquiring model data for the music to be reproduced, an extraction process of extracting singing noise components from the singing data acquired in the first acquisition process, and an evaluation process of evaluating the singing data based on the extracted singing noise components and the evaluation data, are performed. [Effects of the Invention]

[0018] According to the singing evaluation program, the singing evaluation device, the karaoke program, and the singing evaluation method of the present invention, it is possible to more accurately determine the similarity to the original singing. [Brief Description of the Drawings]

[0019] [Figure 1] A diagram showing the configuration of the karaoke system of the present embodiment [Figure 2] A flowchart showing music data generation processing of the present embodiment [Figure 3] A diagram showing the data configuration of music data of the present embodiment [Figure 4]This figure shows the model singing data, volume, and fundamental frequency of this embodiment. [Figure 5] This figure illustrates the extraction of the exemplary singing spectral components and singing noise components in this embodiment. [Figure 6] Flowchart showing the music playback process of this embodiment [Figure 7] Flowchart showing the singing evaluation process of this embodiment [Figure 8] Flowchart showing noise component processing in this embodiment [Figure 9] A diagram illustrating the syllable-by-syllable comparison of this embodiment. [Figure 10] This figure illustrates a comparison of the singing spectral components and singing noise components in this embodiment. [Modes for carrying out the invention]

[0020] Figure 1 shows the configuration of the karaoke system of this embodiment. The karaoke system in this embodiment consists of a karaoke device 2 (sometimes called a commander) and a remote control device 1. The karaoke device 2 and the remote control device 1 are communicated together to form a network using a LAN 100 and an access point 130.

[0021] The karaoke device 2 includes an audio control unit 25 as a sound playback unit for playing songs. The karaoke device 2 also includes an operation unit 21 that receives various inputs from the user. The karaoke device 2 includes an operation processing unit 22 that interprets the inputs from the operation unit 21 and transmits them to the CPU 30. The karaoke device 2 also includes a hard disk 32 as a storage unit for storing various information. The karaoke device 2 includes a LAN communication unit 24a as a communication means for network communication by connecting to the LAN 100. Furthermore, the karaoke device 2 of this embodiment also includes a wireless LAN communication unit 24b, and it is possible to perform a wireless network connection using the wireless LAN communication unit 24b instead of a wired network connection using the LAN communication unit 24a.

[0022] Furthermore, the karaoke device 2 includes a video playback unit that displays lyrics and background images on the monitor 41. This video playback unit comprises a video forming unit 29 that plays back images based on video data, a video RAM 28 that temporarily stores the images to be played back, and a video control unit 31 that superimposes lyrics onto the played back images and applies video effects. In this embodiment, the background image is a video, but the other images can be either videos or still images.

[0023] Furthermore, in addition to the externally connected monitor 41, the karaoke machine 2 is also capable of displaying various information on a touch panel monitor 33. The touch panel monitor 33 is composed of a display unit 35 that displays video input from the video control unit 31 and a touch panel 34 that outputs the touch input position to the operation processing unit 22. This touch panel monitor 33 is located on the front of the karaoke machine 2's casing and functions as an input unit, similar to the operation unit 21 of the karaoke machine 2 or the touch panel monitor 11 of the remote control device 1. By selecting a song on the touch panel monitor 33, the user can directly reserve a song on the karaoke machine 2 and perform various operations on the karaoke machine 2.

[0024] Furthermore, the karaoke device 2 is configured to include a control unit that includes a CPU 30 for overall control of each component and a memory 27 for temporarily storing information necessary for executing various programs.

[0025] In this configuration, the karaoke device 2 performs various processes, but its main functions include song reservation processing and song playback processing. Song reservation processing is the process of selecting and reserving a song based on the user's specifications, and is performed in cooperation with the remote control device 1. The reservation information formed by the song selection process of the remote control device 1 is sent to the karaoke device 2. The karaoke device 2 registers the received reservation information in the reservation table in memory 27. Song playback processing is the process of playing the reserved song, and is a process in which the performance processing, lyrics display processing, and video playback processing are executed synchronously.

[0026] The performance process involves instructing the sound control unit 25 to perform a performance based on the performance data included in the music data. The music performed by the sound control unit 25 is emitted from the speaker 42 along with the singing voice input from microphones 43a and 43b. The lyrics display process assists singing by displaying a lyric video based on the lyric data included in the music data on the monitor 41. Alternatively, a background video display process may be performed to superimpose the lyric video displayed in this lyrics display process onto a background video.

[0027] Meanwhile, the remote control device 1 is capable of performing song selection processing, which involves searching for songs based on user instructions and sending reservation information for the requested songs to the karaoke device 2. The remote control device 1 is also capable of receiving various information from the karaoke device 2 or a server device 5 connected to the internet and performing various processes. In this embodiment, the user interface for receiving various instructions from the user includes an operation unit 17 and a touch panel monitor 11. The touch panel monitor 11 is configured to have a display unit 11a and a touch panel 11b, and displays various information on the display unit 11a and is capable of receiving touch input from the user.

[0028] Furthermore, the remote control device 1 is configured to include a memory 14 as a storage unit for storing a database, various programs, and various information generated in conjunction with program execution, which are necessary for the music selection process, and a remote control unit for comprehensively controlling these components. The remote control unit includes a CPU 15, a video control unit 13 that forms the image to be displayed on the touch panel monitor 11, a video RAM 12 that temporarily stores the video data of the image to be displayed, and an operation processing unit 18 that interprets input from the touch panel monitor 11 or the operation unit 17 and transmits it to the CPU 15.

[0029] The remote control device 1 is connected to the network configured by the LAN 100 by wirelessly connecting to the access point 130 via the wireless LAN communication unit 16. Each remote control device 1 is pre-associated with a specific karaoke machine 2. Various commands output from the remote control device 1 are received by the associated karaoke machine 2.

[0030] With this configuration of the remote control device 1, various inputs from the user are received from the touch panel monitor 11 or the operation unit 17, and various information is provided through the display on the touch panel monitor 11, enabling various processes such as song selection, which sends reservation information to the karaoke machine 2.

[0031] The karaoke device 2 of this embodiment performs a singing evaluation process during music playback. In particular, the singing evaluation process evaluates the similarity to the original singer who sings the master recording. First, the generation of music data including evaluation data necessary for performing the singing evaluation process will be explained.

[0032] Figure 2 is a flowchart showing the music data generation process of this embodiment. First, vocal data is extracted from the master recording (original data recorded on a CD (Compact Disc), etc.) (S101). Vocal data may include harmonies in addition to the main vocals. Therefore, harmonies are removed from the vocal data extracted in S101 to generate model singing data (S102). Then, evaluation data is obtained from the model singing data (S103).

[0033] In this embodiment, five data points are acquired as evaluation data: fundamental frequency, volume, information related to singing noise components, vocal tract characteristics, and formant shape. Figure 3 shows the data structure of the song data in this embodiment. The fundamental frequency is used to determine the pitch of the singing. The volume is used to determine the volume of the singing. The remaining three data points (information related to singing noise components, vocal tract characteristics, and formant shape) are used to determine similarity with the original singer. These three data points are extracted for each syllable of the singing voice.

[0034] The singing noise component-related information is information that indicates the husky (hoarse) characteristics of the singer's singing voice. The husky (hoarse) characteristics of a singing voice can also be described as undesirable sounds or noises contained in the voice, or a voice with a low, rough texture. The husky (hoarse) characteristics of a singing voice are elements that greatly influence the texture and impression of the voice, and are particularly important for determining similarity to the original singer. As shown in Figure 3, the singing noise component-related information in this embodiment consists of a model singing noise spectrum and the model singing noise component content rate. Details of this singing noise component-related information will be explained later. Vocal tract characteristics are information about the shape of the vocal tract obtained by cepstrum analysis, etc. Formant shape is information that indicates the shape of formants, which are frequency bands with greater intensity than the surroundings on the frequency spectrum of the sound, and can be obtained by LPC analysis, etc. In this embodiment, these three data (singing noise component-related information, vocal tract characteristics, and formant shape) are used to determine similarity to the original singer.

[0035] In the music data generation process, after acquiring evaluation data, performance data (MIDI data, etc.) synchronized with the tempo of the master recording is generated (S104). Then, by mapping the evaluation data and lyric data to synchronize with the performance data (S105), the music data shown in Figure 3 is generated.

[0036] Next, we will describe the details of the evaluation data used in the singing evaluation process of this embodiment. Figure 4 is a diagram showing the model singing data, volume, and fundamental frequency of this embodiment. The model singing data is an audio waveform, which is shown here separated by syllables of the lyrics for illustrative purposes. Below the model singing data, data showing the fundamental frequency and volume extracted from the model singing data is shown. The vertical axis of the fundamental frequency represents the magnitude of the frequency, and the vertical axis of the volume represents the loudness of the sound. In the singing evaluation process, it is possible to obtain a high score by singing in accordance with these fundamental frequencies and volumes.

[0037] Figure 5 illustrates the extraction of exemplary singing spectral components and singing noise components in this embodiment. Figure 5(A) is the exemplary singing spectrum obtained by spectrally analyzing a certain syllable, for example, the first lyric "o" in the exemplary singing data of Figure 4. The horizontal axis of the exemplary singing spectrum is frequency, and the vertical axis is level, representing a frequency spectrum. Using this exemplary singing spectrum, exemplary singing spectral components and exemplary singing noise components are extracted.

[0038] Figure 5(B) shows the extraction of the spectral components and noise components of the model singing. In Figure 5(B), the solid line (black line) represents the spectral components of the model singing, and the dashed line (white line) represents the noise components of the model singing. The spectral components of the model singing are obtained by taking the envelope connecting the peaks of the model singing spectrum. On the other hand, the noise components of the model singing are obtained by taking the envelope connecting the peaks of the valleys of the model singing spectrum. In other words, the noise components of the model singing can be said to be components other than the pitch component of the model singing data. Figure 5(C) shows the obtained spectral components and noise components of the model singing.

[0039] In this embodiment, the percentage of the model singing noise component is calculated from the obtained model singing spectral component and model singing noise component, and stored together with the model singing noise component as model singing noise component-related information. Here, the calculation of the model singing noise component percentage will be explained. Figure 10(A) is a graph schematically showing model singing spectral component A and model singing noise component B. In reality, model singing spectral component A and model singing noise component B are curves as shown in Figure 5(C). Area S2 is the area below model singing noise component B, and area S1 corresponds to the area below model singing spectral component A (S1+S2) minus area S2. Here, area S2 is an index indicating the degree of huskiness (hoarseness of voice) in the model singing data. On the other hand, area S1 is an index indicating the degree of voice clarity in the model singing data. In this embodiment, the noise component content of the model singing is calculated as the value obtained by dividing area S2 by area S1 (area S2 / area S1). In other words, the noise component content of the model singing serves as an indicator of the degree of huskiness in the model singing data.

[0040] Furthermore, the model singing noise component related information of this embodiment also includes the model singing noise component. This model singing noise component serves as an indicator of the frequency characteristics of huskyness (hoarseness of the voice) in the model singing data. Thus, in this embodiment, the model singing noise component content rate is calculated as part of the singing noise component related information, and the degree of huskyness of the voice and the frequency characteristics of the huskyness are compared by using the model singing noise component.

[0041] The remaining evaluation data (vocal tract characteristics, formant shape) will also be acquired syllable by syllable, similar to the information related to the noise components of the model singing. The above explains the information related to the singing noise components of the model singing data. The singing voice data used for comparison with the evaluation data will also obtain various types of information on the same items as the evaluation data.

[0042] Next, we will explain the music playback process, including the singing evaluation process. Figure 6 is a flowchart showing the music playback process of this embodiment. In the music playback process performed by the karaoke device 2, first the reservation table is checked, and then it is confirmed whether there is a song to be played (S201). If there is a song to be played (S202: Yes), the corresponding music data is read out and music playback begins (S203). During music playback, the performance data in the music data is played back by the sound control unit 25, and the lyrics data is played back by the video playback unit (video forming unit 29, video RAM 28, video control unit 31).

[0043] In this embodiment, singing data input from microphones 43a and 43b is recorded during the playback of the song. The recorded singing data is used for the singing evaluation process (S300) after the song playback ends (S205: Yes). In this embodiment, the singing evaluation process (S300) is performed using the singing data recorded after the song playback ends, but the singing evaluation process may also be performed in parallel during song playback. Once the singing evaluation process (S300) is completed, the karaoke device 2 displays the evaluation result of the singing evaluation process (S300) on the monitor 41 (S206).

[0044] Now, let's describe the details of the singing evaluation process (S300) according to this embodiment. In the singing evaluation process (S300), first, the acquisition of evaluation data (S301) and then the acquisition of recorded singing data (S302) are performed. The singing evaluation process (S300) in this embodiment is mainly divided into two processes. One is evaluation on a syllable basis, and the other is evaluation of a series of flows (not on a syllable basis). The evaluation of a series of flows (not on a syllable basis) is performed on the fundamental frequency and volume. For the fundamental frequency, the fundamental frequency is acquired from the singing data (S315), the acquired fundamental frequency is compared with the fundamental frequency in the evaluation data (S316), and the item score for the fundamental frequency is calculated (S317). On the other hand, for volume, the volume is acquired from the singing data (S318), the acquired fundamental frequency is compared with the fundamental frequency in the evaluation data (S319), and the item score for the fundamental frequency is calculated (S320).

[0045] On the other hand, evaluation at the syllable level is performed on singing noise component-related information, exemplary vocal tract characteristics, and exemplary formant shape. First, the acquired singing data is divided into syllables (S303), and one syllable is picked up in order (S304). In the evaluation data, data for the syllable corresponding to the picked-up syllable (exemplary singing noise component-related information, exemplary vocal tract characteristics, exemplary formant shape) is searched (S305).

[0046] Figure 9 is a diagram illustrating the syllable-by-syllable comparison in this embodiment. The horizontal axis represents the passage of time. The upper part shows the syllable-by-syllable intervals A1 to A7 for the model singing data, and the lower part shows the syllable-by-syllable intervals B1 to B7 for the singing data. When singing using the karaoke device 2, as shown in Figure 9, the lyrics (syllables) sung do not necessarily correspond to the same time as the model singing data. Furthermore, it is conceivable that the accuracy of the similarity determination with the model singing data used in the master recording can be improved by performing the determination syllable by syllable. Therefore, in this embodiment, as shown in Figure 9, corresponding syllables indicated by dashed lines (for example, intervals A1 and B1, etc.) are searched, and evaluation data (model singing noise component related information, model vocal tract characteristics, model formant shape) is compared with data extracted from the singing data.

[0047] Regarding vocal tract characteristics, first, vocal tract characteristics are obtained from singing data divided into syllables (S307). Then, the obtained vocal tract characteristics are compared with the model vocal tract characteristics in the evaluation data (S308), and the item score for vocal tract characteristics is calculated (S309). Regarding formant shape, first, formant shape is obtained from singing data divided into syllables (S310). Then, the obtained formant shape is compared with the model formant shape in the evaluation data (S311), and the item score for formant shape is calculated (S312).

[0048] Next, the noise component processing (S400) according to this embodiment will be described. Figure 8 is a flowchart showing the noise component processing (S400) of this embodiment, and Figure 10 is a diagram illustrating a comparison of the singing spectral components and singing noise components of this embodiment. In the noise component processing (S400), first, the singing spectrum is calculated from the singing data of the syllables picked up in S304 (S401). The singing spectrum shows the level for each frequency (frequency spectrum), similar to the model singing spectrum in Figure 5(A). However, here, singing data acquired from microphones 43a and 43b is used. Next, the singing spectral components are extracted from the singing spectrum calculated in S401 (S402). The singing spectral components are obtained as an envelope connecting the peaks of the singing spectrum. Then, the singing noise components are extracted from the singing spectrum calculated in S401 (S403). The singing noise component is obtained as an envelope connecting the peaks of the valleys in the singing spectrum. It should also be said that the singing noise component is the component of the singing data other than the pitch component.

[0049] Figure 10(B) schematically shows the singing spectral component C and singing noise component D of the singing data. As mentioned above, Figure 10(A) schematically shows the exemplary singing spectral component A and exemplary singing noise component B, but it can be seen that the magnitude (vertical axis) is not uniform compared to Figure 10(B). Therefore, in this embodiment, normalization (S404) is performed on the singing spectral component C and singing noise component D of the singing data.

[0050] Figure 10(B') shows the singing spectral component C and singing noise component D of the normalized singing data. In this embodiment, the noise component content is calculated from the singing spectral component C and singing noise component D of the normalized singing data (S405). In this embodiment, the singing noise component content is calculated as the value obtained by dividing area T2 by area T1 (area T2 / area T1). Here, area T2 is the area below the singing noise component D', and area T1 corresponds to the area obtained by subtracting area T2 from the area below the singing spectral component C' (T1+T2).

[0051] Next, the shape of the singing noise component calculated in S404 is compared with the shape of the model singing noise component included in the evaluation data (S406). Since the singing noise component and the model singing noise component are indices that show the frequency characteristics of the degree of huskiness (hoarseness of voice), by comparing the shapes of the singing noise component and the model singing noise component, it is possible to determine how similar the voice quality of the singing data is to the voice quality of the model singing data. By evaluating the similarity of the shapes of the singing noise component and the model singing noise component, it is possible to compare the voice quality of the syllables being evaluated. As a comparative example, this allows for a more accurate evaluation compared to comparing the amount of singing noise component and the amount of model singing noise component of the syllables being evaluated. For the shape comparison of the singing noise component and the model singing noise component, indices related to the similarity of the shapes of the two, such as cosine similarity or Pearson correlation coefficient, can be used. In this embodiment, the higher the similarity between the singing noise component and the model singing noise component, the higher the score for the singing noise component item.

[0052] Next, the percentage of singing noise components obtained from the singing data is compared with the percentage of model singing noise components included in the evaluation data (S407). Since the percentage of singing noise components and the percentage of model singing noise components are indicators of the degree of huskiness (hoarseness of the voice), the closer the two are, the more similar the voice quality of the singing data and the model singing data are. By comparing the singing noise components and the model singing noise components, it is possible to determine how similar the voice quality of the singing data is to the voice quality of the model singing data. In this embodiment, the closer the percentage of singing noise components is to the percentage of model singing noise components, the higher the score for the singing noise component item.

[0053] Then, by adding together, for example, the item scores obtained from the shape comparison of the noise component spectral components obtained in S406 and the item scores obtained from the comparison of the noise component content rate obtained in S407, the item score for the singing noise component for the currently targeted syllable is calculated (S408).

[0054] The calculation of item scores for vocal tract characteristics (S307-S309), formant shape (S310-S312), and singing noise component (S400), as described above, is repeated for all syllables. Once processing for all syllables is complete (S313: Yes), for example, the similarity score, an index indicating how similar the singing data is to the model singing data, is calculated by summing the item scores for vocal tract characteristics, formant shape, and singing noise component (S314). Then, a total score is calculated based on the item score for fundamental frequency obtained in S317, the item score for volume obtained in S320, and the similarity score obtained in S314 (S321).

[0055] The calculated total score and the scores for each item are displayed on the monitor 41 as the evaluation result display of the music playback process shown in Figure 6. The evaluation result display may also be configured to show the scores for vocal tract characteristics, formant shape, and singing noise components on the monitor 41 in real time as the music is played. Such a configuration allows the user to see evaluation information at the syllable level, potentially contributing to the improvement of singing technique. Furthermore, the evaluation result display may also be configured to show the scores for each item at any arbitrary division of the music (e.g., phrase, line, chapter, etc.). Such a configuration allows the user to see evaluation information at the division level, potentially contributing to the improvement of singing technique. The karaoke device 2 that performs the singing evaluation process (S300) of this embodiment has been described above. In particular, by performing noise component processing (S400) in the singing evaluation process (S300) of this embodiment, it is possible to accurately determine to what extent the voice of the singing data (singer) is similar to the voice of the model singing data (singer on the original recording). It should be noted that the embodiment is not limited to those described above, and various modifications such as those described below can be adopted.

[0056] [First variation] In the embodiment described above, as explained in the noise component processing (S400) in Figure 8, a comparison of the shape of the noise spectral components (S406) and a comparison of the noise component content (S407) are performed. However, it is also possible to compare only one of these. In that case, the evaluation data will only include the information necessary for comparison (model singing noise components or model singing noise component content).

[0057] [Second variation] In the embodiment described above, as explained in Figure 10, the percentage of model singing noise components was calculated as (area S2 / area S1), and the percentage of singing noise components was calculated as (area T2 / area T1). The exemplary singing noise component content and the singing noise component content can be calculated using various methods, not just the method described above. For example, the exemplary singing noise component content can be calculated as (Area S2 / (Area S1+Area S2)), and the singing noise component content as (Area T2 / (Area T1+Area T2)).

[0058] The above description focuses on the application of the singing evaluation program of this embodiment to a karaoke device 2. However, the present invention extends beyond the application to a karaoke device 2 to include singing evaluation programs provided as such, or to singing evaluation devices or singing evaluation methods to which the singing evaluation program is applied in various information processing devices. [Explanation of Symbols]

[0059] 1: Remote control device 24b: Wireless LAN communication unit 2: Karaoke machine 25: Sound control unit 5: Server device 27: Memory 11: Touch panel monitor 28: Video RAM 11a: Display unit 29: Image forming unit 11b: Touch panel 30: CPU 12: Video RAM 31: Video control unit 13: Video control unit 32: Hard disk 14: Memory 33: Touchscreen monitor 15: CPU 34: Touch panel 16:Wireless LAN communication section 35:Display section 17: Control panel 41: Monitor 18: Operation processing unit 42: Speaker 21: Control panel 43a, 43b: Microphone 22: Operation Processing Unit 130: Access Point 24a:LAN communication department

Claims

1. The first acquisition process for obtaining singing data, A second acquisition process to obtain evaluation data for the song being played, The first acquisition process extracts singing noise components from the singing data, and The information processing device is instructed to perform an evaluation process that evaluates the singing data based on the extracted singing noise components and evaluation data. A program for evaluating singing performance.

2. Singing noise components are components of singing data other than the pitch component. The singing evaluation program according to claim 1.

3. The extraction process is, A first acquisition process to obtain the frequency spectrum of singing data, A second acquisition process is performed to obtain the envelope formed by connecting the peaks of the troughs in the acquired frequency spectrum as the singing noise component. The singing evaluation program according to claim 1.

4. The evaluation data includes model singing noise components extracted from model singing data. The evaluation process involves a first comparison process that compares the extracted singing noise components with the model singing noise components to evaluate the singing data. The singing evaluation program according to claim 1.

5. The singing noise component and the model singing noise component are frequency spectra that show the level for each frequency. The singing evaluation program according to claim 4.

6. The evaluation process involves comparing the shape of the extracted singing noise component with that of the model singing noise component to assess the singing data. The singing evaluation program according to claim 5.

7. The noise components of the model singing are divided into syllable units. The first comparison process compares the singing noise component with the model singing noise component corresponding to the syllable of the singing noise component. The singing evaluation program according to claim 4.

8. A second extraction process is performed to extract the singing noise component content rate from the singing data and singing noise components obtained in the first acquisition process. The evaluation process involves a second comparison process that compares the extracted singing noise component content with the model singing noise component content included in the evaluation data to evaluate the singing data. The singing evaluation program according to claim 1.

9. The noise component content of the model singing is divided into syllable units. The second comparison process compares the percentage of singing noise components with the model singing noise components corresponding to the syllables with that percentage of singing noise components. The singing evaluation program according to claim 8.

10. The first acquisition process for obtaining singing data, A second acquisition process to obtain evaluation data for the song being played, The first acquisition process involves extracting singing noise components from the singing data acquired in the first acquisition process, An evaluation process is performed to evaluate the singing data based on the extracted singing noise components and evaluation data. Singing evaluation device.

11. The playback process for playing music data, The first acquisition process for obtaining singing data, A second acquisition process to obtain evaluation data corresponding to the music data to be played, The first acquisition process involves extracting singing noise components from the singing data acquired in the first acquisition process, The information processing device is instructed to perform an evaluation process that evaluates the singing data based on the extracted singing noise components and evaluation data. A program for karaoke.

12. The first acquisition process for obtaining singing data, The second acquisition process involves obtaining sample data for the song to be played, The first acquisition process involves extracting singing noise components from the singing data acquired in the first acquisition process, Based on the extracted singing noise components and evaluation data, an evaluation process is performed to assess the singing data. Method for evaluating singing performance.

Images