JP4910854B2 - Fist detection device, fist detection method and program - Google Patents

Description

  The present invention relates to a fist detection device, a fist detection method, and a program.

Various karaoke apparatuses for evaluating singing have been developed. For example, in the karaoke apparatus described in Patent Document 1, parameters such as pitch (pitch), sound length, and timing are extracted from the singer's voice, and singing is evaluated based on the extracted parameters.
Japanese Patent Laid-Open No. 10-78750

  By the way, skilled singers such as singers use various expression methods such as intentionally shifting the beginning and end of singing, changing the voice quality and volume, and using techniques such as vibrato and fist. Express emotions in the song. Many singers want to sing using various ways of expression, like a singer, rather than faithfully following the content of the score.

  Now, in the conventional karaoke apparatus, singing is evaluated based on references, such as model singing and a guide melody. However, these references are often created with the same pitch and rhythm as the score, and when singing using various expression methods as described above, the evaluation tends to decrease because they differ from the reference. was there. Therefore, a technique for detecting various expression methods as described above in a karaoke apparatus and evaluating a song based on each detected expression method has been desired.

  The present invention has been made in view of the above-described circumstances, and provides a fist detection device, a fist detection method, and a program capable of detecting a section sung using the “fist” technique from a singing voice. The purpose is to do.

  A fist detection device according to a preferred aspect of the present invention includes a receiving unit that receives singing voice data representing a singing voice, a pitch detecting unit that detects a pitch from singing voice data received by the receiving unit, and the pitch detecting unit. From the detected pitch, a section where the pitch rises and then descends is specified. From the section, (1) the absolute value of the rate of change in pitch in the section where the pitch rises is larger than a predetermined threshold, (2 ) The absolute value of the rate of change in pitch in the section where the pitch is lowered is larger than a predetermined threshold, and (3) the length of the section from when the pitch starts to rise until it finishes falling is within the predetermined range. A candidate section selecting means for selecting a candidate section to be characterized, and a second section for receiving vibrato section data representing a section in which a vibrato technique is used in the singing voice If each of the candidate sections selected by the receiving means and the candidate section selecting means is not included in the vibrato section represented by the vibrato section data received by the second receiving means, the technique of fisting the candidate sections is used. And a fist section identifying means that identifies the section as being defined.

  In the fist detection device according to a preferred aspect of the present invention, in the configuration described above, the pitch starts to increase from the timing when the candidate section selection unit further starts to sound a note (musical sound) including the section. The candidate section may be selected using a condition that the time difference from the timing is smaller than a predetermined threshold.

  In the fist detection device according to a preferred aspect of the present invention, in the above-described configuration, the candidate section selection unit further adds a pitch exceeding a predetermined size in a section of a predetermined length immediately before or immediately after the section. The candidate section may be selected using a condition that there is no change.

  In the fist detection device according to a preferred aspect of the present invention, in the configuration described above, the candidate section selection unit further adds a pitch fluctuation range in a predetermined length immediately after the section or in a predetermined length section after a predetermined time. The candidate section may be selected using a condition that the value is smaller than the threshold value.

  The fist detection method according to a preferred aspect of the present invention includes a receiving step for receiving singing voice data representing a singing voice, a pitch detecting step for detecting a pitch from the singing voice data received in the receiving step, and the pitch detecting step. From the detected pitch, a section in which the pitch rises and then descends is specified. From the section, (1) the absolute value of the rate of change in pitch in the section in which the pitch rises is greater than a predetermined threshold, (2 ) The absolute value of the rate of change in pitch in the section where the pitch is lowered is larger than a predetermined threshold, and (3) the length of the section from when the pitch starts to rise until it finishes falling is within the predetermined range. A candidate section selection stage for selecting a candidate section to be characterized, and vibrato section data representing a section in which the vibrato technique is used in the singing voice. If each of the candidate sections selected in the second receiving stage and the candidate section selecting stage is not included in the vibrato section represented by the vibrato section data received in the second receiving stage, the candidate section is fisted. And a fist segment identifying stage that identifies the segment in which the technique is used.

  A program according to a preferred aspect of the present invention includes a receiving unit that receives singing voice data representing a singing voice, a pitch detecting unit that detects a pitch from the singing voice data received by the receiving unit, and the pitch detection. From the pitch detected by the means, a section where the pitch rises and then falls is identified, and from the section, (1) the absolute value of the rate of change in pitch in the section where the pitch rises is greater than a predetermined threshold value, 2) The absolute value of the rate of change in pitch in the section where the pitch is lowered is greater than a predetermined threshold, and (3) the length of the section from when the pitch starts to rise until it finishes falling is within the predetermined range. Candidate section selecting means for selecting candidate sections characterized by the above, and vibrato section data representing sections in which the vibrato technique is used in the singing voice If each of the second receiving means for receiving and the candidate section selected by the candidate section selecting means is not included in the vibrato section represented by the vibrato section data received by the second receiving means, the candidate section is fisted. It is characterized by functioning as a fist section identifying means that identifies the section in which the technique is used.

  According to the fist detection device, the fist detection method, or the program according to the present invention, it is possible to detect a section sung using the “fist” technique from the singing voice.

  Below, the karaoke apparatus which concerns on one Embodiment of this invention is demonstrated. In the following description, a person who practices singing using the karaoke apparatus is referred to as a “singer”.

(A: Configuration)
FIG. 1 is a block diagram showing a hardware configuration of the karaoke apparatus 1. The karaoke apparatus 1 has a normal karaoke function for reproducing karaoke accompaniment, and also has a fist detection function for detecting a section (hereinafter referred to as a fist section) in which the technique of “fist” is used from the singing voice of the singer. The “fist” is a technique for adding a decorative and undulating tune.

  In FIG. 1, a CPU (Central Processing Unit) 11 reads a control program stored in a ROM (Read Only Memory) 12, loads it into a RAM (Random Access Memory) 13, and executes it to execute a karaoke device. 1 part is controlled.

The display unit 15 is, for example, a liquid crystal display, and displays various screens such as a menu screen for operating the karaoke device 1 and a karaoke screen in which lyrics telop is superimposed on a background image under the control of the CPU 11.
The operation unit 16 includes various keys such as a numeric keypad, an up / down key, and a performance start key, and outputs an operation signal corresponding to the pressed key to the CPU 11.

The microphone 17 picks up the singing voice and generates a voice signal (analog data).
The audio processing unit 18 performs A / D conversion on the audio signal generated by the microphone 17, converts it into digital data (audio data), and outputs the digital data to the CPU 11. The audio processing unit 18 performs D / A conversion on the audio data received from the CPU 11, converts the audio data into an audio signal, and outputs the audio signal to the speaker 19.
The speaker 19 emits sound based on the sound signal received from the sound processing unit 18.

  The storage unit 14 is a large-capacity storage unit such as an HDD (Hard Disk Drive), and has various storage areas.

A plurality of music data is stored in the music data storage area 14a. FIG. 2 is a diagram schematically showing the contents of each piece of music data. Each piece of music data has a header, accompaniment data, lyrics data, and guide melody data.
The header includes song number data for specifying a song, song name data indicating a song title, genre data indicating a genre, performance time data indicating a performance time of a song, and the like.

In the accompaniment data, performance sounds of various musical instruments that accompany the music are recorded as the music progresses. The accompaniment data is in a data format such as MIDI (Musical Instrument Digital Interface) format.
In the lyrics data, the contents (characters) of the lyrics are described in association with the timing to be displayed and the position to be displayed on the screen of the display unit 15.
In the guide melody data, a guide melody indicating the melody that the singer should sing is written.

In the singing voice data storage area 14b, voice data (hereinafter referred to as singing voice data) generated by A / D conversion of a voice signal representing the singer's song output from the microphone 17 by the voice processing unit 18 is provided. , Stored for each song. The singing voice data is voice data such as WAVE format or MP3 (MPEG-1 Audio Layer-3) format.
The vibrato section data storage area 14c stores data indicating a section (hereinafter referred to as a vibrato section) in which the “vibrato” technique is used in the singing voice data of each music piece.

The fist section data storage area 14d stores data indicating a fist section in which the “fist” technique is used in the singing voice data of each music piece.
The parameter storage area 14e stores a pitch extracted from the singing voice data of each musical piece and various parameters extracted from the pitch.
Each unit of the karaoke apparatus 1 described above exchanges data with each other via the bus 20.

(B: Operation)
Next, a process in which the karaoke apparatus 1 specifies a “fist section” from singing voice data will be described.

(B-1: Karaoke accompaniment processing)
When the singer operates the operation unit 16 to select a song to be sung, an operation signal for specifying the song such as song number data of the song is output from the operation unit 16 to the CPU 11. The CPU 11 reads the music data indicated by the operation signal supplied from the operation unit 16 from the music data storage area 14a, and performs a karaoke accompaniment process on the read music data.

FIG. 3 is a flowchart showing the flow of karaoke accompaniment processing.
In step SA100, the CPU 11 causes the speaker 19 to emit the karaoke accompaniment and causes the display unit 15 to display the lyrics telop. That is, the CPU 11 reads the accompaniment data included in the music data from the music data storage area 14 a and outputs it to the audio processing unit 18. The audio processing unit 18 converts the accompaniment data into an analog audio signal and outputs the analog audio signal to the speaker 19. Further, the CPU 11 reads out the lyrics data included in the song data from the song data storage area 14a, and causes the display unit 15 to display the lyrics telop according to the lyrics data.

  The singer sings along with the karaoke accompaniment emitted from the speaker 19 while watching the lyrics telop displayed on the display unit 15. The voice signal generated by the microphone 17 is A / D converted to generate singing voice data (step SA110). The singing voice data is written in the singing voice data storage area 14b.

In step SA120, the CPU 11 analyzes the pitch of the singing voice data as the music progresses, and generates singing pitch data representing the analysis result. That is, the singing voice data generated in step SA110 is read from the singing voice data storage area 14b, the pitch is detected from the read singing voice data in units of frames of a predetermined time length (for example, 3 msec), and the detected pitch is indicated. Singing pitch data is generated. In the singing pitch data, the pitch is shown as a relative value of the detected pitch with reference to the pitch of the guide melody included in the music data. The generated singing pitch data is written in the parameter storage area 14e.
FIG. 4 shows a part of the song (time 12s000ms to 18s000ms) of the singing pitch data generated in step SA120 as a graph A1. In FIG. 4, the horizontal axis represents time (elapsed time since the music was started). The vertical axis indicates the relative value of the singing pitch data at each time with respect to the guide melody.

  In step SA130, the CPU 11 determines whether or not the music performance has been completed for one song. If the determination result in step SA130 is “Yes”, the karaoke accompaniment process is terminated. If the determination result in step SA130 is “No”, the process from step SA100 to step SA120 is performed for the remaining portion of the music.

(B-2: Vibrato section specifying process)
In specifying the “fist section” from the singing voice data, the CPU 11 performs a vibrato section specifying process for specifying a section in which “vibrato” indicating characteristics similar to “fist” is used in advance. Vibrato is a singing technique that gives the sound a rich reverberation by creating a trembling tone by slightly raising and lowering the pitch while stretching the sound.

  FIG. 5 is a flowchart showing the flow of the vibrato section specifying process. In step SB100, the CPU 11 performs a filtering process for extracting a specific frequency component on the singing pitch data written in the parameter storage area 14e. In the present embodiment, the CPU 11 processes the singing pitch data with a low-pass filter that extracts components having a frequency lower than 6 Hz, and generates new pitch data (hereinafter referred to as filter singing pitch data). A graph A2 in FIG. 4 shows filter singing pitch data generated by processing the singing pitch data of the graph A1 by the low-pass filter.

  As shown in FIG. 4, the singing pitch data (A1) before applying the filter has a fine disturbance in the waveform. Such waveform disturbance is caused by, for example, reverb. When a pitch is detected from audio data subjected to reverberation, the detection result is not a sine wave but a waveform disturbance. For this reason, it has been difficult to detect a vibrato section from reverberated speech. Furthermore, it is difficult to determine from the audio data whether or not the audio is reverberated. However, in the singing pitch data processed by the low-pass filter, the influence of the reverb applied to the voice is removed, and it becomes possible to appropriately specify the vibrato section in the processing described later.

  In step SB110, CPU11 specifies the area (henceforth a vibrato candidate area) which shows the characteristic of a vibrato area in song voice data on the following conditions. That is, the CPU 11 identifies a point where the pitch represented by the filter singing pitch data generated in step SB100 changes from negative to positive or from positive to negative (zero crossing) as a zero crossing point. Specifically, for example, in the example shown in FIG. 4, the time at which the graph A2 representing the filter singing pitch data zero-crosses (for example, time P1, P2, P3, P4, etc.) is specified as the zero-cross location.

  Next, the CPU 11 measures the time interval at which the zero-cross point appears in the filter singing pitch data, the measured time interval is within a predetermined range, and the time interval is continuously detected a predetermined number of times or more. The section is specified as a vibrato candidate section. By this process, in the example shown in FIG. 4, the section A3 in which the zero cross points appear at almost equal intervals is specified as the vibrato candidate section. In FIG. 4, one section is specified as the vibrato candidate section. However, the vibrato candidate section is also specified in the music portion not included in FIG. 4.

  Then, in step SB110, the CPU 11 extracts a plurality of parameters for each of the sections as follows in order to more strictly analyze that the vibrato technique is actually used in the specified vibrato candidate section. (Step SB120). In the following description, for example, when the value of the filter singing pitch data fluctuates periodically as in section A3 in FIG. 4, the number of vibrations per unit time is referred to as “vibrato frequency”.

(1) Average frequency of vibrato (Af; Average of frequency)
The parameter Af is the average value of the vibrato frequency in each vibrato candidate section, and is calculated as the average value of the reciprocal of the time interval at which the filter singing pitch data zero crosses the horizontal axis.
(2) Standard deviation (Df: Deviation of frequency) of vibrato
The parameter Df is calculated as the standard deviation of the reciprocal distribution of the time interval at which the filter singing pitch data crosses zero with the horizontal axis. From this parameter, it is possible to estimate the magnitude of “variation” of the vibrato frequency. That is, the closer the value of this parameter is to 0, the better the vibrato has a uniform frequency.

  Here, the “pitch vibration width” used in the description of the following parameters will be described. FIG. 6 is a graph showing the filter singing pitch data (A2) extracted from FIG. In FIG. 6, the CPU 11 calculates the “pitch vibration width” in the vibrato candidate section as follows. First, the CPU 11 differentiates the filter singing pitch data with respect to time, thereby specifying a maximum value and a minimum value from the graph of the data.

  For example, in FIG. 6, Q2, Q4, Q6, Q8, and Q10 indicate maximum values, and Q1, Q3, Q5, Q7, and Q9 indicate minimum values. The CPU 11 uses the difference between the specified minimum value and the maximum value immediately adjacent in time as the pitch vibration width, and uses the pitch vibration width as the minimum value and the maximum value used to calculate the value. Position at an intermediate time. For example, the pitch vibration width W1 is generated from the minimum value Q1 and the maximum value Q2. In FIG. 6, pitch vibration widths W1 to W5 generated in this way are written.

Now, the description returns to the parameters extracted in step SB120.
(3) Average value of pitch vibration width (Ap: Average of pitch)
The parameter Ap indicates an average value of the pitch vibration width found in each vibrato candidate section.
(4) Standard deviation of pitch vibration width (Dp; Deviation of pitch)
The parameter Dp indicates the standard deviation of the pitch vibration width found in each vibrato candidate section. From this parameter, the “variation” of the vibration width of the pitch in the vibrato section can be estimated. That is, the closer the value of this parameter is to 0, the better the vibrato that the pitch vibrates with a uniform vibration width.

(5) Slope of pitch (Sp; Slope of pitch)
The parameter Sp indicates the slope of the linear approximation line in the pitch vibration width graph. FIG. 7 shows a graph of the pitch vibration width calculated in FIG. The CPU 11 determines a linear approximation line for the pitch vibration width point in the vibrato candidate section (A3 in the figure). For example, in the section A3 shown in FIG. 7, the linear approximate straight line graph is determined as a straight line L1 and expressed as (Equation 1).
(Formula 1) P = 15t + 150
Thus, by calculating the linear approximate straight line, the slope Sp of the straight line is determined. In the above example, the slope Sp of the linear approximation line of the pitch vibration width is 15.
From this parameter, the stability of the vibration width of the pitch during vibrato can be estimated. That is, the smaller the absolute value of Sp is, the better the vibrato is, in which the pitch fluctuation range is kept uniform during the vibrato.

In step SB130, the CPU 11 determines whether to finally determine each of the vibrato candidate sections specified in step SB110 as a vibrato section based on the following criteria. That is,
(1) Df is smaller than a predetermined threshold (2) Ap is within a predetermined range (3) Dp is smaller than a predetermined threshold (4) The absolute value of Sp is smaller than a predetermined threshold (1) to ( A vibrato candidate section that satisfies all the conditions in 4) is finally determined as a vibrato section.

  In the vibrato section specified by the above conditions, it is expected that the possibility that vibrato is used is very high. This is because, generally, in vibrato, variations in the vibration frequency and pitch vibration width of vibrato are small, and the vibration width is within a predetermined range (for example, within 500 cents), and further, the fluctuation range of the pitch. Is substantially constant throughout the vibrato section. Note that “cent” is a unit indicating a relative pitch difference of pitches. For example, a pitch indicated by +100 cents indicates a pitch that is a semitone above a reference pitch. The CPU 11 stores vibrato section data representing the specified section in the vibrato section data storage area 14c.

FIG. 8 shows vibrato section data generated for the singing pitch data shown in FIG. As shown in FIG. 8, in the vibrato section data, the start time and the end time are written for the vibrato section detected in the singing voice data for each song.
This completes the vibrato section specifying process.

  As described above, in step SB110, once the time interval of pitch vibration in the filter singing pitch data is within a predetermined range and the time interval has been continuously detected a predetermined number of times or more, the vibrato is temporarily performed. We narrowed down the candidates for the section. Based on the parameters extracted in step SB120, it was strictly determined whether the vibrato candidate section is appropriate as a vibrato section. As described above, it is possible to finally specify an accurate vibrato section by determining whether or not a change in pitch peculiar to vibrato is exhibited based on a plurality of conditions.

(B-3: Fist section specifying process)
When the CPU 11 finishes the above-described vibrato section specifying process, the CPU 11 performs the fist section specifying process. The fist section specifying process is a process for specifying the fist section from the singing voice data. FIG. 9 is a flowchart showing the flow of the fist section specifying process.

In step SC100, the CPU 11 reads out the filter singing pitch data from the parameter storage area 14e.
Next, in step SC110, CPU11 specifies the area (henceforth a fist candidate area) which may contain a fist area in singing voice data as follows. In addition, in the following description, it demonstrates with reference to FIG. 10 which showed a part of filter song pitch data typically. In FIG. 10, P _A (> 0) and P _B (<0) indicate the pitch fluctuation ranges in the section A where the pitch is increasing and the section B where the pitch is decreasing, respectively. Section C indicates a section from the start of section A to the end of section B. t _A , t _B , and t _C indicate the time widths of the sections A, B, and C, respectively.

The CPU 11 specifies a section that simultaneously satisfies the following conditions (1) to (3) from the filter singing pitch data as a “fist candidate section”. That is, in the section where the pitch increases and decreases again (section C),
(1) The absolute value (| P _A / t _A |) of the rate of change in pitch in the section where the pitch increases (section A) is greater than a predetermined value.
(2) The absolute value (| P _B / t _B |) of the rate of change in pitch in the section where the pitch is lowered (section B) is greater than a predetermined value.
(3) The length (t _C ) of the section from when the pitch starts to rise until it finishes falling is within a predetermined range. That is, the time required for a temporary increase in pitch is not too short and not too long.

When the fist candidate sections are identified in FIG. 11 showing the same filter singing pitch data as in FIG. 4 by the above conditions (1) to (3), the sections 1, 2, 3, 4, 5, and 6 are identified. . The CPU 11 generates data indicating the fist candidate section as fist candidate section data and writes it in the fist section data storage area 14d.
FIG. 12 shows an example of fist candidate section data. In the fist candidate section data, the pitch fluctuation start time and end time are written for each fist candidate section specified in the singing voice data for each piece of music. For example, in FIG. 12, the data of 00m14s500 to 00m15s400ms corresponds to the pitch fluctuation in section 3 in FIG.

In step SC120, the CPU 11 reads the vibrato section data from the vibrato section data storage area 14c and the fist candidate section data from the fist section data storage area 14d, and is included in the vibrato section from the section indicated by the read fist candidate section data. The final fist section is specified by excluding the fist candidate section.
For example, the fist candidate section data shown in FIG. 12 and the vibrato section data shown in FIG. 8 are generated from the filter singing pitch data shown in FIG. 11, but a plurality of fist candidate sections shown in FIG. Of these, only the section of 00m12s200ms to 00m12s800ms is not included in the vibrato section, but other sections other than the section are excluded from the fist section because they are included in the vibrato section. In addition, when the start time or end time of the fist candidate section and the vibrato section are shifted, if a part of the fist candidate section is included in the vibrato section, the fist candidate section is included in the vibrato section. It may be determined that

  As described above, the identification of the fist candidate section is based on the condition that the pitch rises temporarily within a predetermined time, so if the singing voice includes vibrato, the vibrato is configured. Vibrations of individual pitches are included. Therefore, the fist candidate section is specified by specifying both the fist and the vibrato, and the fist section is specified by excluding the separately specified vibrato section from the specified fist candidate section.

(C: Modification)
As mentioned above, although one Embodiment of this invention was described, this invention is not limited to the above-mentioned embodiment, It can implement with another various form. An example is shown below.

(1) In the above-described embodiment, the case has been described in which the section in which the pitch variation mode satisfies the conditions (1) to (3) is specified as the “fist candidate section” in the filter singing pitch data. However, in addition to the above conditions, the fist candidate section may be specified by the following conditions (a), (b), (c), and (d), for example.
(A) A fist candidate section having a pitch rise within a predetermined period immediately after the sound generation timing of each note (musical sound) is determined.
As shown in FIG. 13A, which shows a part of the filter singing pitch data, a predetermined time or more has elapsed from the start of sounding (time t1) in a note that is sounded from time t1 to time t3. If a pitch rise (time t2) is observed after that, the section is not specified as a fist candidate section. This is because it is common to turn the fist immediately after the start of pronunciation of each note.
(B) There is no pitch drop exceeding a predetermined level immediately before and after the fist.
As shown in FIG. 13B, when a large drop in the pitch showing the minimum value at time t4 is seen immediately before the transient increase in the pitch showing the maximum value at time t5, The temporarily increasing section of the pitch showing the maximum value at the time t5 is not specified as the fist candidate section.
(C) There is no pitch increase exceeding a predetermined level immediately after the fist.
As shown in FIG. 13C, when a large increase in the pitch showing the maximum value at time t7 is observed immediately after the transient increase in the pitch showing the maximum value at time t6, A section where the pitch is temporarily increased at time t6 is not specified as a fist candidate section.
(D) After a temporary increase in pitch, there is a flat portion (a portion where the difference between the maximum value and the minimum value of the pitch in the section is within a certain value) for a certain period. That is, there is a flat portion in a predetermined length section immediately after a section in which the pitch increases temporarily, or in a predetermined length section after a predetermined time in a section in which the pitch increases temporarily.
As shown in FIG. 13 (d), after the transient rise in the pitch showing the maximum value at time t8 is finished, the pitch showing the minimum value is lowered at time t9, or the maximum value is shown at time t10. Since a pitch increase is observed and the pitch fluctuation range in the section is larger than a predetermined threshold value, the transient increase section of the pitch showing the maximum value at time t8 is not specified as the fist candidate section.
It should be noted that a plurality of conditions are selected and used from the conditions (1) to (3) and the above conditions (a), (b), (c), and (d) shown in the above embodiment. Thus, the fist candidate section may be specified, and the combination method of the conditions may be set appropriately.

(2) In the above-described embodiment, the case where it is based on the variation mode of the pitch has been described as an example of the method for specifying the vibrato section. However, the method for specifying the vibrato section is not limited to the above-described method. For example, any of a plurality of conditions used in the above embodiment for specifying a vibrato section may not be used, or other conditions may be used in combination with the above conditions.

(3) In the embodiment described above, the singing voice data is data in the WAVE format or MP3 format, but the format of the data is not limited to this, and any format as long as it represents singing voice data. It may be the data.

(4) In embodiment mentioned above, the case where the karaoke apparatus 1 implement | achieves all the functions which concern on the said embodiment was demonstrated. However, two or more devices connected via a network may share the above function, and a system including the plurality of devices may realize the function of the karaoke device 1 of the embodiment.

(5) In the above-described embodiment, the case where the vibrato section is specified from the singing voice data, the vibrato section is excluded from the fist candidate sections, and the fist section is specified has been described. However, when data indicating the section to be sung using vibrato techniques, such as attached to music data, is obtained, the vibrato section may be specified using the data, and the above-described vibrato section specifying process is performed. There is no need to do it. In that case, the vibrato section indicated by the data may be written in the vibrato section data storage area 14c, and the written vibrato section data may be read and used in the fist section specifying process.

(6) In embodiment mentioned above, the case where a pitch was detected from singing voice data with the progress of karaoke accompaniment, and the singing pitch data was produced | generated was demonstrated. However, this process does not necessarily have to be performed with karaoke accompaniment. For example, the pitch may be detected at the stage where the karaoke accompaniment is completed from the singing voice data once generated by accumulating the singing voice from the beginning to the end of the music.

(7) In the said embodiment, the case where a fist area was specified from song audio | voice data was demonstrated. However, for example, the fist section may be specified in advance by the method described in the above embodiment from data indicating an exemplary song such as a singer's song, and karaoke music data including data indicating the specified section may be created. In that case, if the karaoke apparatus 1 displays on the display part 15 which section should be sung using a fist in a music, the singer can sing using a fist at an appropriate timing. .

(8) In embodiment mentioned above, the case where the fist area in a music was specified from a singer's song audio | voice data was demonstrated. However, for example, the audio data stored in some storage means may be read and the fist detected from the audio data.

(9) In the above-described embodiment, the CPU 11 performs low-pass filter processing for extracting frequency components equal to or lower than a specific frequency with respect to the singing voice data. However, the filter processing performed by the CPU 11 is not limited to this. A filter that extracts a frequency component having a frequency width of may be used. In short, any filtering process that extracts components in a specific frequency band may be used.

(10) In embodiment mentioned above, the case where the fist area was specified from the singing voice data showing a singer's song was demonstrated. However, the audio data to be processed is not limited to data representing the singing voice, and may be audio data representing the performance sound of a musical instrument such as a violin or a flute. By doing so, it is possible to detect a performance technique having characteristics similar to those of a fist, for example, in an instrument.

(11) Programs executed by the CPU 11 of the karaoke apparatus 1 in the above-described embodiment are a magnetic tape, a magnetic disk, a flexible disk, an optical recording medium, a magneto-optical recording medium, a CD (Compact Disk) -ROM, a DVD (Digital Versatile). Disk), RAM, ROM, and the like. It is also possible to download to the karaoke apparatus 1 via a network such as the Internet.

(12) In the above-described embodiment, when specifying a fist section, if a part of the fist candidate section is included in the vibrato section, it is determined that the fist candidate section is included in the vibrato section Explained. However, the method for determining whether or not each fist candidate section is included in the vibrato section is not limited to the above method. For example, when the vibrato section includes from the start time to the end time of each fist candidate section, it may be determined that the fist candidate section is included in the vibrato section. Further, in each fist candidate section, when the time at which the pitch shows a maximum value is included in the vibrato section, it may be determined that the fist candidate section is included in the vibrato section.

1 is a block diagram illustrating a configuration of a karaoke apparatus 1. FIG. It is a figure which shows the content of music data. It is a flowchart which shows the flow of a karaoke accompaniment process. It is a figure which shows singing pitch data and filter singing pitch data. It is a flowchart which shows the flow of a vibrato area specific process. It is a figure for demonstrating the calculation method of a pitch vibration width. It is a figure which shows the calculation method of the linear approximate line of pitch vibration width. It is a figure which shows an example of vibrato area data. It is a flowchart which shows the flow of a fist area specific process. It is a figure for demonstrating the identification method of a fist candidate area. It is a figure which shows filter song pitch data. It is a figure which shows an example of fist candidate area data. It is a figure for demonstrating the identification method of a fist area.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Karaoke apparatus, 11 ... CPU, 12 ... ROM, 13 ... RAM, 14 ... Memory | storage part, 15 ... Display part, 16 ... Operation part, 17 ... Microphone, 18 ... Audio | voice processing part, 19 ... Speaker, 20 ... Bus.

Claims (6)

Receiving means for receiving singing voice data representing the singing voice;
Pitch detection means for detecting the pitch from the singing voice data received by the receiving means;
From the pitch detected by the pitch detection means, identify a section that falls after the pitch has risen, from the section,
(1) The absolute value of the rate of change in pitch in a section in which the pitch increases is greater than a predetermined threshold;
(2) The absolute value of the rate of change in pitch in the section where the pitch is lowered is greater than a predetermined threshold value,
(3) Candidate section selection means for selecting a candidate section characterized in that the length of the section from the start of the pitch rise to the end of the fall is within a predetermined range;
A second receiving means for receiving vibrato section data representing a section in which the vibrato technique is used in the singing voice;
If each of the candidate sections selected by the candidate section selecting means is not included in the vibrato section represented by the vibrato section data received by the second receiving means, a section in which the technique of fisting the candidate section is used. A fist detection device characterized by comprising: a fist section specifying means.   The candidate section selecting means selects the candidate section using a condition that a time difference between a timing at which a note (musical tone) including the section is started and a timing at which the pitch starts to rise is smaller than a predetermined threshold. The fist detection device according to claim 1.   The candidate section selecting means selects the candidate section using a condition that there is no variation in pitch exceeding a predetermined size in a section of a predetermined length immediately before or after the section. The fist detection device according to 1.   The candidate section selecting means selects the candidate section using a condition that a pitch fluctuation range is smaller than a predetermined threshold immediately after the section or in a section having a predetermined length after a predetermined time. Item 3. The fist detection device according to Item 1. A receiving stage for receiving singing voice data representing the singing voice;
A pitch detection step of detecting a pitch from the singing voice data received in the reception step;
From the pitch detected in the pitch detection step, identify a section that descends after the pitch has increased, from the section,
(1) The absolute value of the rate of change in pitch in a section in which the pitch increases is greater than a predetermined threshold;
(2) The absolute value of the rate of change in pitch in the section where the pitch is lowered is greater than a predetermined threshold value,
(3) a candidate section selection stage for selecting a candidate section characterized in that the length of the section from the start of the pitch rise to the end of the fall is within a predetermined range;
Receiving a vibrato section data representing a section in which the vibrato technique is used in the singing voice;
If each of the candidate sections selected in the candidate section selection stage is not included in the vibrato section represented by the vibrato section data received in the second receiving stage, a section in which the technique of fisting the candidate section is used A fist detection method characterized by comprising: a fist segment identification step identified as Computer
Receiving means for receiving singing voice data representing the singing voice;
Pitch detection means for detecting the pitch from the singing voice data received by the receiving means;
From the pitch detected by the pitch detection means, identify a section that falls after the pitch has risen, from the section,
(1) The absolute value of the rate of change in pitch in a section in which the pitch increases is greater than a predetermined threshold;
(2) The absolute value of the rate of change in pitch in the section where the pitch is lowered is greater than a predetermined threshold value,
(3) Candidate section selection means for selecting a candidate section characterized in that the length of the section from the start of the pitch rise to the end of the fall is within a predetermined range;
A second receiving means for receiving vibrato section data representing a section in which the vibrato technique is used in the singing voice;
If each of the candidate sections selected by the candidate section selecting means is not included in the vibrato section represented by the vibrato section data received by the second receiving means, a section in which the technique of fisting the candidate section is used. A program for functioning as a fist section identification means.

Images