JP4735398B2 - Acoustic signal analysis apparatus, acoustic signal analysis method, and acoustic signal analysis program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To precisely detect the musical interval by suppressing the variation of performance of musical signal and the effect of kinds of music. <P>SOLUTION: The sound signal 2 is frequency analyzed and the frequency components c[i] [q] are generated, the element of which is component intensity of prescribed frequency for every unit time period. Among each element, an element which satisfies at least one condition is detected as a peak element, wherein the conditions are &ge;threshold &alpha;[q], &ge;calculated value calculated based on the vicinity of each element, and the value of the harmonic element in harmonic relation to each element is &ge;calculated value based on the vicinity element. The number of peak elements of the same frequency for every first interval H is counted, if the results of counting is &ge;prescribed number, the same frequency in the first interval H is detected as a musical pitch region. The number of musical pitch regions of every second interval J which is not less than the first interval H, if the result of counting is &ge;prescribed number, the second interval J is detected as the music interval. <P>COPYRIGHT: (C)2008,JPO&amp;INPIT

Description

  The present invention relates to an acoustic signal analyzing apparatus, an acoustic signal analyzing method, and an acoustic signal analyzing program for analyzing a sound signal and detecting a signal section including music as a music section.

  In recent years, it has been widely practiced to record contents such as moving images and sounds on recording media such as hard disks, DVDs, and memories. With the increase in capacity of these recording media, there is an increasing need for a function for reading and viewing only a desired recording section from content recorded for a long time and a function for viewing with a reduced playback time. In such a situation, techniques for extracting various information for dividing a content into small sections according to the contents and identifying only a signal section in which music is included in the content have been developed.

Specifically, a method is known in which sound information included in a video signal is analyzed and a music section is identified using stability in a time direction in a spectrum (see, for example, Patent Documents 1 and 2). There is also known a method for identifying a music section and a voice section using an average band energy ratio of voice subband data and a subband energy centroid (see, for example, Patent Documents 3 and 4).
Japanese Patent Laid-Open No. 10-187182 JP 2000-315094 A Japanese Patent Laid-Open No. 10-247093 JP 2000-66691 A

  In Patent Documents 1 and 2, the music section is determined based on whether or not the sum of the edge intensities of the spectrogram is larger than a preset threshold value. However, the sum of the spectrogram edge strengths varies depending on the signal level and frequency characteristics of the analysis target (variation of the characteristics of the input source) and the type of music (genre). It is difficult to cope with. For example, when there are two similar music sections, the sum of the edge strengths is large when the average sound pressure is large, whereas the sum of the edge strengths is small when the other average sound pressure is small. Therefore, there is a problem that the detection accuracy is not always sufficient when detecting a music section for various input sources and various types of music.

  Further, according to the techniques disclosed in Patent Documents 3 and 4, although there is an advantage that the music section and the voice section can be distinguished from the compression-coded voice data by a relatively simple calculation, there are still various inputs. When detecting a music section for a source or a song of various music genres, it is difficult to increase the detection accuracy. This is because in general music, the frequency component has a characteristic that a specific frequency component emitted by a musical instrument or the like lasts for a certain period of time. However, the above-described conventional technology does not fully utilize this property. Because there is no.

  Furthermore, in normal music, there are frequency components based on scales such as equal temperament and pure temperament, but in any of Patent Documents 1 to 4, since frequency components that do not directly correspond to musical scales are used, Music detection accuracy is not necessarily high.

  Therefore, the present invention has been made in view of the above problems, and its purpose is to suppress the influence of variations in the characteristics of the acoustic signal that is the input source and the type of music included in the acoustic signal, and An object of the present invention is to provide an acoustic signal analysis device, an acoustic signal analysis method, and an acoustic signal analysis program for accurately detecting a signal section including music.

In order to solve the above problems, the present invention
[1] acoustic signal and divided into a plurality of frequency bands, a frequency analysis means for generating a frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands and elements, said frequency component data The condition that the read element is equal to or greater than a predetermined threshold, or that the read element is larger than the value calculated based on the frequency component element in the vicinity of the read element. At least one of the conditions or the condition that the harmonic element that is the element corresponding to the harmonic component of the read element is larger than the value calculated based on the element of the frequency component in the vicinity of the harmonic element a peak component detecting means for detecting an element that satisfies a condition as a peak component, for each of the plurality of frequency bands, when including the unit time Counts the number of pre-Symbol peak elements for each first section having a length, the counting result is the first section is greater than or equal to a first predetermined number, detected by the pitch area detected as the pitch region at that frequency band The number of the pitch regions is counted for each of the plurality of frequency bands for each second section having a time length including the first section, and the counting result is equal to or greater than a second predetermined number . in some cases, providing a sound signal analysis apparatus characterized by comprising a music segment detection means for detecting the second interval as a music segment.
[2] Also, in split an acoustic signal into a plurality of frequency bands, a frequency analysis means for generating a frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands and elements, the frequency Read the element of the component data and the condition that the read element is equal to or greater than the predetermined threshold, or the read element is more than the value calculated based on the element of the frequency component near the read element At least of the condition that the harmonic element that is the element corresponding to the harmonic component of the read element is larger than the value calculated based on the element of the frequency component in the vicinity of the harmonic element. A peak element detecting means for detecting an element satisfying one condition as a peak element, and the unit time for each of the plurality of frequency bands. Counting the number of the peak elements and the sum of the values of the peak elements for each first section having a time length including the number of peak elements is equal to or greater than a first predetermined number, and A pitch area detecting means for detecting the first section as a pitch area in the frequency band when the sum of the values is equal to or greater than a first predetermined value, and a second length having a time length including the first section . for each interval, the number of the pitch region counted as a target to each of the plurality of frequency bands, when the count result is a second predetermined value or more, detects the second section as music section There is provided an acoustic signal analyzing apparatus comprising a music section detecting means .
[3] Further, by dividing the acoustic signals into a plurality of frequency bands, a frequency analysis means for generating a frequency component data to the component strength per unit time element in each of the plurality of frequency bands, said frequency component data The condition that the read element is equal to or greater than a predetermined threshold, or that the read element is larger than the value calculated based on the frequency component element in the vicinity of the read element. At least one of the conditions or the condition that the harmonic element that is the element corresponding to the harmonic component of the read element is larger than the value calculated based on the element of the frequency component in the vicinity of the harmonic element A peak element detecting means for detecting an element satisfying a condition as a peak element, and the unit time for each of the plurality of frequency bands. Counting the number of the peak elements and the sum of the values of the peak elements for each first section having a time length including the number of peak elements is equal to or greater than a first predetermined number, and When the sum of the values is equal to or greater than a first predetermined value, the first section is detected as a pitch region in the frequency band, and the sum of the peak element values is determined as a pitch region intensity of the pitch region. For each of the plurality of frequency bands, the sum of the pitch region intensities is counted for each of the second sections having a time length including the first section . when a predetermined value or more, to provide a sound signal analysis apparatus characterized by comprising a music segment detection means for detecting the second interval as a music segment.
[4] The first interval has a time length including a plurality of the unit times, and the second interval has a time length including a plurality of the first intervals. The acoustic signal analyzer according to any one of [3] is provided .
[5] In addition, the plurality of frequency bands, to provide an acoustic signal analyzer according to any one of [1] to [4], characterized in that a frequency band corresponding to the pitch frequency of the musical scale .
[6] The acoustic signal analyzer according to any one of [1] to [5], wherein the unit time is different depending on each of the plurality of frequency bands .
[7] In addition, the pitch region detecting means uses a value smaller than the number of unit times included in the first section as the first predetermined number [1] to [6]. An acoustic signal analysis device according to any one of the above is provided .
[8] In addition, the pitch region detection unit is not the peak element when the peak element, the element that is not the peak element, and the peak element are temporally continuous in one frequency band. An acoustic signal analyzer according to any one of [1] to [7], wherein an element is counted as the peak element .
[9] In addition, the acoustic signal is divided into a plurality of frequency bands, a frequency analysis step of generating frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands and elements, the frequency Read the element of the component data and the condition that the read element is equal to or greater than the predetermined threshold, or the read element is more than the value calculated based on the element of the frequency component near the read element At least of the condition that the harmonic element that is the element corresponding to the harmonic component of the read element is larger than the value calculated based on the element of the frequency component in the vicinity of the harmonic element. a peak component detecting step of detecting an element that satisfies one condition as the peak element, for each of the plurality of frequency bands, the single Counts the number of pre-Symbol peak elements for each first section having a length of time including the time, the counting result is the first section is greater than or equal to a first predetermined number, detected as the pitch region at that frequency band a pitch region detection step of, in the second each section having a time length that includes the first interval, the number of the pitch region counted as a target to each of the plurality of frequency bands, the count result of the second If it is more than a predetermined number, to provide a sound signal analysis method, which comprises have a music-segment detection step of detecting the second section as music section.
[10] Also, to split the acoustic signal into a plurality of frequency bands, a frequency analysis step of generating frequency component data to the component strength per unit time element before Symbol respective plurality of frequency bands, said frequency Read the element of the component data and the condition that the read element is equal to or greater than the predetermined threshold, or the read element is more than the value calculated based on the element of the frequency component near the read element At least of the condition that the harmonic element that is the element corresponding to the harmonic component of the read element is larger than the value calculated based on the element of the frequency component in the vicinity of the harmonic element. a peak component detecting step of detecting an element that satisfies one condition as the peak element, for each of the plurality of frequency bands, wherein Counting the number of the peak elements and the sum of the values of the peak elements for each first section having a time length including a rank time, the number of the peak elements being a first predetermined number or more, and A pitch area detecting step for detecting the first section as a pitch area in the frequency band when the sum of the peak element values is equal to or greater than a first predetermined value, and a time length including the first section. second for each section, the number of the pitch region counted as a target to each of the plurality of frequency bands, when the count result is a second predetermined value or more, music and the second section having to provide an acoustic signal analysis method, which comprises have a music-segment detection step of detecting as a section.
[11] In addition, by dividing the acoustic signals into a plurality of frequency bands, a frequency analysis step of generating frequency component data to the component intensity per unit time and the elements in each of the plurality of frequency bands, said frequency component data The condition that the read element is equal to or greater than a predetermined threshold, or that the read element is larger than the value calculated based on the frequency component element in the vicinity of the read element. At least one of the conditions or the condition that the harmonic element that is the element corresponding to the harmonic component of the read element is larger than the value calculated based on the element of the frequency component in the vicinity of the harmonic element a peak component detecting step of detecting an element that satisfies a condition as a peak component, for each of the plurality of frequency bands, wherein Counting the number of the peak elements and the sum of the values of the peak elements for each first section having a time length including a rank time, the number of the peak elements being a first predetermined number or more, and When the sum of the peak element values is equal to or greater than the first predetermined value, the first section is detected as a pitch region in the frequency band, and the sum of the peak element values is calculated as a pitch of the pitch region. The sum of the pitch region intensities is counted for each of the plurality of frequency bands for each pitch region detecting step to be a region intensity and for each second interval having a time length including the first interval. If it is the second predetermined value or more, to provide a sound signal analysis method, which comprises have a music-segment detection step of detecting the second section as music section.
[12] In addition, the computer, the acoustic signal is divided into a plurality of frequency bands, a frequency analysis step of generating frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands and elements The element of the frequency component data is read, and the condition that the read element is equal to or greater than a predetermined threshold, or the read element is calculated based on the element of the frequency component near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element . of a peak component detecting step of detecting an element that satisfies at least one condition as a peak component, the plurality of frequency bands that For counts the number of pre-Symbol peak elements for each first section having a time length that includes the unit time, the counting result is the first section is greater than or equal to a first predetermined number, at that frequency band The number of the pitch range is counted for each of the plurality of frequency bands for each pitch range detection step that is detected as a pitch range, and for each second interval having a time length including the first interval, and the counting result There the case where the second predetermined number or more, to provide a sound signal analysis program for executing the music-segment detection step of detecting the second section as music section.
[13] In addition, the computer, the acoustic signal is divided into a plurality of frequency bands, a frequency analysis step of generating frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands and elements The element of the frequency component data is read, and the condition that the read element is equal to or greater than a predetermined threshold, or the read element is calculated based on the element of the frequency component near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element . A peak element detecting step for detecting an element satisfying at least one condition as a peak element, and each of the plurality of frequency bands. For each of the first intervals having a time length including the unit time, the number of the peak elements and the sum of the values of the peak elements are counted, and the number of the peak elements is equal to or greater than a first predetermined number. And when the sum of the peak element values is equal to or greater than a first predetermined value, a pitch region detecting step of detecting the first interval as a pitch region in the frequency band; and a second each section having a time length that includes a number of said pitch region counted as a target to each of the plurality of frequency bands, when the count result is a second predetermined value or more, the second to provide an acoustic signal analysis program for executing the music-segment detection step of detecting a section as music section.
[14] Also, the computer divides the acoustic signal into a plurality of frequency bands, a frequency analysis step of generating frequency component data to the component strength per unit time element in each of the plurality of frequency bands, wherein Read the frequency component data element, and the condition that the read element is equal to or greater than the predetermined threshold, or the read element is a value calculated based on the frequency component element in the vicinity of the read element Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on an element of a frequency component in the vicinity of the harmonic element . A peak element detecting step of detecting, as a peak element, an element satisfying at least one condition, and each of the plurality of frequency bands. For each of the first intervals having a time length including the unit time, the number of the peak elements and the sum of the values of the peak elements are counted, and the number of the peak elements is equal to or greater than a first predetermined number. And when the sum of the peak element values is greater than or equal to a first predetermined value, the first section is detected as a pitch region in the frequency band, and the sum of the peak element values is For each pitch region detecting step for setting the pitch region intensity of the pitch region and for each second section having a time length including the first section, the sum of the pitch area intensities is counted for each of the plurality of frequency bands. , if the sum is a second predetermined value or more, to provide a sound signal analysis program for executing the music-segment detection step of detecting the second section as music section.

  According to the present invention, a sound section of a musical instrument with a high probability of being present in a signal section including music can be detected with high accuracy, so that a music section can be selected from acoustic signals from various input sources. Can be detected with high accuracy.

  Further, according to the present invention, even in the case where the intensity of a specific frequency component instantaneously increases in a voice section where there is no music, it is possible to dramatically determine that a voice section is a music section. It can be kept low.

  Furthermore, according to the present invention, since it is possible to extract a frequency component that directly corresponds to a scale used in music, when a musical instrument is pronounced, the frequency component of a pitch in the vicinity of the sounded pitch. Can be sufficiently reflected in the arithmetic processing, so that the music section can be detected more accurately.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to preferred embodiments.

  FIG. 1 shows the overall configuration of an acoustic signal analyzer according to the first embodiment of the present invention. In FIG. 1, the acoustic signal analysis device 1 includes an acoustic signal input unit 11, a frequency analysis unit 12, a peak element detection unit 13, a pitch region detection unit 14, and a music section detection unit 15. Each processing unit includes various arithmetic processes and arithmetic processing circuits 11a to 15a for controlling the processing units. Furthermore, although not shown, the acoustic signal analyzing apparatus 1 includes a control unit including a CPU for executing the acoustic signal analysis program according to the embodiment of the present invention and controlling the processing units. ing. In this acoustic signal analyzing apparatus 1, after the acoustic signal 2 is input to the acoustic signal input unit 11 and subjected to various processes according to the signal processing flow shown in FIG. It is designed to output.

  The acoustic signal 2 may be in any format among audio signal formats such as PCM data, analog audio signal, and digital compressed signal.

  The acoustic signal input unit 11 has a function of generating PCM data having a predetermined sampling frequency Fs from the input acoustic signal 2. Specifically, the acoustic signal input unit 11 performs digital conversion processing when the acoustic signal 2 is an analog audio signal, and performs decoding processing when the acoustic signal 2 is a digital compression signal. When the acoustic signal 2 is PCM data, the acoustic signal input unit 11 supplies the input acoustic signal 2 as it is to the frequency analysis unit 12 at the next stage, but the sampling frequency of the input source is the sampling frequency Fs. If they are different, a rate conversion process is executed to convert them to PCM data of the sampling frequency Fs.

  In the following description, the PCM data output from the acoustic signal input unit 11 is the acoustic data x [m] (m = 0 to L−1, L is an integer of 1 or more, and indicates the total number of acoustic data. ) Or acoustic data.

  The frequency analysis unit 12 has a function of inputting the acoustic data output from the acoustic signal input unit 11 and executing a frequency analysis process to create matrix data having a component intensity of a predetermined frequency component as an element every unit time. Have.

  The processing of the frequency analysis unit 12 will be described with reference to the flowchart of FIG. In the present embodiment, the frequency analysis unit 12 divides the acoustic data into fixed-length frames and executes processing in units of frames. As a frequency analysis method, a well-known STFT (Short-time Fourier Transform) is used, but other methods such as wavelet transform and filter bank may be used. In this embodiment, all frequency components are generated at the same time interval. However, the frequency components may be generated by changing the time interval according to the frequency band.

  In the following description, the frame length is N and the frame shift length is S. A time length corresponding to the frame shift length S is a unit time.

  When the total number of frames is M, the total number M of frames is calculated by Equation 1.

  The above floor function is a function that returns an integer with the decimal part truncated. In this embodiment, it is assumed that L ≧ N. In the flowchart of FIG. 2, the arithmetic processing circuit 12a of the frequency analysis unit 12 sets a control variable i indicating a frame number to 0 (step S110). Next, the arithmetic processing circuit 12a creates the i-th frame (step S120). That is, as shown in FIG. 3, N pieces of data are extracted from the position offset by i × S from the head of the acoustic data, and this is multiplied by the window function w as shown in Equation 2 to obtain the i th Frame data y [i] [n] (n = 0 to N−1) is generated.

  As the window function w, for example, a Hamming window shown in Equation 3 can be used. In addition, a rectangular window, Hanning window, Blackman window, or the like may be used.

  Next, the arithmetic processing circuit 12a calculates the discrete Fourier transform (DFT) of the i-th frame by Equation 4 (step S130).

  Next, the arithmetic processing circuit 12a includes the real part Re {a [i] [k]} and the imaginary part Im of the complex sequence a [i] [k] (k = 0 to N−1) calculated in step S130. {A [i] [k]} is used to calculate the spectrum sequence b [i] [k] (k = 0 to N / 2-1) of the i-th frame using Equation 5 or Equation 6 (Step 0). S140).

  When Equation 5 is applied, the spectrum sequence b [i] [k] is a power spectrum, and when Equation 6 is applied, the spectrum sequence b [i] [k] is an amplitude spectrum.

  Next, the arithmetic processing circuit 12a calculates the frequency component c [i] [q] (q = 0 to Q-1, Q is an integer equal to or greater than 1 and the number of frequency bands in the frame i and the frequency band q from the DFT result. (Step S150). Here, for example, the following two methods can be applied as the calculation method in step S150.

  The first method for calculating the frequency component c [i] [q] is a method in which part or all of the spectrum sequence b [i] [k] is made to correspond to c [i] [q] according to Equation 7.

  Here, λ is a predetermined integer of 0 or more, and is a parameter that determines the lowest frequency of the frequency band. Further, the frequency band number Q is set to a predetermined value which is equal to or less than (N / 2−λ). This first calculation method has the advantage that the calculation amount is the smallest and simple. However, since the frequency intervals of the frequency bands are equal and irrelevant to the musical scale used in music, the second method described below has higher accuracy in detecting the music section.

  The second method of calculating the frequency component c [i] [q] is a method of calculating the frequency component corresponding to the musical scale using Equation 8.

  In this method, since the frequency component corresponding to the frequency of the scale used in music is obtained, it is possible to detect the music section with higher accuracy.

  Here, z [q] [k] (q = 0 to Q-1, k = 0 to N / 2-1) is a filter group having band characteristics as shown in FIG. The center frequency of each filter corresponds to the frequency of the scale, the attenuation at the center frequency is small, the attenuation increases with distance from the center frequency, and the pass characteristic becomes zero near the center frequency of the adjacent filter. have. Figures (a) to (d) show that the C1 pitch of the scale corresponds to band 0, then corresponds to one band for each semitone, and finally the B6 pitch above 5 octaves corresponds to band Q-1. The obtained band characteristics are schematically shown. (A) is a filter that passes a frequency corresponding to C1, FIG. (B) is a filter that passes a frequency corresponding to C # 1 that is a semitone higher than (a), and (c) is the highest sound. FIG. 4D shows the band characteristics of a filter that passes a frequency corresponding to A # 6, which is lower by a semitone, and a filter that passes a frequency corresponding to B6, which is the highest tone.

  The spectrum sequence b [i] [k] has an equal frequency interval, whereas the frequency interval between adjacent pitches increases as the scale becomes higher, so that the filter group z [q] [k] Corresponding to the center frequency, the interval between the center frequencies adjacent to each other is increased as the treble part is increased. For example, z [0] [k] shown in FIG. 4 (a) and z [1] [k] shown in FIG. The difference in each center frequency between Q-2] [k] and z [Q-1] [k] shown in FIG. Similarly, the bandwidth of each filter becomes wider as the treble part increases. For example, the bandwidth of z [Q−1] [k] shown in FIG. 4D is wider than the bandwidth of z [0] [k] shown in FIG.

  The example of the band characteristics shown in FIG. 4 is a frequency band that matches each pitch (semitone) of the scale, but a frequency band that further divides each pitch of the scale may be obtained. Further, by changing the setting of the center frequency, it is possible to deal with various musical scales such as equal temperament, pure temperament, and all-medium temperament.

  Returning to the description of the flowchart of FIG. 2, the arithmetic processing circuit 12a of the frequency analysis unit 12 increases the value of the control variable i indicating the frame number by 1 after the processing of step S150 (step S160). Next, the arithmetic processing circuit 12a determines whether or not the value of the control variable i indicating the frame number is smaller than the total number M of frames (step S170). When the control variable i is smaller than the total number M of frames, the process proceeds to step S120. When the control variable i is equal to or greater than the total number M of frames, the processing for all the frames is completed and the frequency analysis process is terminated.

  When the above processing is completed, the frequency analysis unit 12 includes a frequency component c [i] [q] (frame i = 0 to M−1, band q = 0 to Q−1) as matrix data. To remember. FIG. 5 schematically shows matrix data created by the frequency analysis unit 12. In the figure, the horizontal axis represents the frame and the vertical axis represents the frequency band, and the element having a strong component intensity is represented as a black line segment. FIG. 6A shows matrix data of a voice section (for example, announcement) in which no music exists, and is characterized by having a small number of frequency components that last for a certain period of time. On the other hand, FIG. 5B shows a music section in which a musical instrument is sounding, and is characterized by a large number of frequency components that last for a certain period of time. FIG. 2B is also characterized by the presence of a harmonic frequency having a frequency that is an integral multiple of the fundamental frequency in addition to the fundamental frequency of the musical instrument. In general music, the frequency component corresponding to the pitch in the vicinity of the pitch being played is relatively small. An object of the present invention is to accurately exclude a voice section as shown in FIG. 5A and detect a music section as shown in FIG.

  Next, the processing flow of the peak element detection unit 13 will be described with reference to the flowchart of FIG. The peak element detection unit 13 has a characteristic that the frequency component corresponding to the pitch is large in the section sounded by the pitched musical instrument, and the frequency component corresponding to the pitch is relatively larger than the nearby frequency component. Is used. That is, the peak element detection unit 13 reads the frequency component c [i] [q] stored as matrix data in the frequency analysis unit 12, and when the value is equal to or greater than a predetermined threshold, A process of detecting a frequency component as a peak element when the frequency component is relatively large compared with each other, or when a harmonic component having a harmonic relationship with the frequency component is relatively large compared with a frequency component in the vicinity of the harmonic component. Execute.

  First, the arithmetic processing circuit 13a of the peak element detection unit 13 sets a control variable i representing a frame number for starting a search to 0 (step S210). Next, the arithmetic processing circuit 13a sets the control variable q indicating the frequency band to the minimum band Q1 in the target range of the peak element (step S220). Here, Q1 is a predetermined integer that is greater than or equal to an integer G2 described later, greater than or equal to an integer G4 described later, and less than or equal to an integer Q2 described later.

  Next, the arithmetic processing circuit 13a determines whether or not the frequency component c [i] [q] is a peak element (step S230). When it is determined that the frequency component c [i] [q] is a peak element, the process proceeds to step S240, and when it is determined that the frequency component is not a peak element, the process proceeds to step S250. As a specific method for determining an effective component in step S230, any of the seven methods described below or a combination of a plurality of methods can be used.

  The first method for determining the peak element is a method for determining that the peak element is a peak element when c [i] [q] is equal to or greater than the threshold value α [q] using Equation 9.

  The threshold value α [q] is set using one of the following two methods. That is, the first method of determining the threshold value α [q] is a method of determining using a preset constant, and according to this method, the calculation amount is the smallest and simple. The second method for determining the threshold value α [q] is a method using an average value of frequency components for each band of all frames (M) as shown in Expression 10. Here, β is a preset constant.

  The second method for determining the peak element is a method using Formula 11.

  The maximum value of the frequency components of a predetermined number of frequency bands (also referred to as neighboring bands) at the same time as the band q (also referred to as the center band) is obtained for each side of the center band, and the two maximum values are obtained. Is also a method of determining that it is a peak element when c [i] [q] is large. In Equation 11, G1 and G2 are integers satisfying G1 ≦ G2, and when the frequency analysis unit 12 matches each frequency band with each pitch (semitone) of the scale, G1 = G2 = 1 may be set. . Further, when a frequency band that further divides each pitch of the scale is used, G1 and G2 are set so as to correspond to frequencies of adjacent pitches or to correspond to frequencies that are not used between adjacent pitches. Set. Also, ∩ means an AND condition, and max means a function that returns the maximum value among the arguments.

  This second method for determining the peak element does not use a threshold value that needs to be set in advance, and is particularly suitable when the acoustic signal level to be analyzed varies greatly. Further, this is suitable when it is desired to keep the rate of erroneous detection of sections other than music as music sections low, or when the frequency resolution of the band used in the frequency analysis unit 12 is relatively high.

  A third method for determining the peak element is a method using Formula 12.

  This is a method in which the minimum value of the frequency component of the neighboring band is obtained for both sides of the center band, and when c [i] [q] is larger than the two minimum values, it is determined as a peak element. G1, G2, and ∩ in Expression 12 are the same as those described above. “Min” means a function that returns the minimum value among the arguments. The third method for determining the peak element is suitable for the case where the acoustic signal level to be analyzed largely fluctuates because a threshold that needs to be set in advance is not used. Further, it is suitable when it is desired to reduce the detection omission of the music section or when the frequency resolution of the band used in the frequency analysis unit 12 is relatively low.

  A fourth method for determining the peak element is a method using Formula 13.

  This is a method for determining a peak element when the frequency component c [i] [q] of the center band is larger than the value obtained by multiplying the sum of the frequency components of neighboring bands by a certain ratio γ. Here, G1 and G2 are the same as those described above, and γ is a constant.

  In a section where a musical instrument is sounding, generally there is a d-order harmonic component having a frequency d times the fundamental frequency in addition to the fundamental frequency. This characteristic is used in the fifth to seventh methods for determining the peak element described below.

  A fifth method for determining the peak element is a method using Expression 14.

  This specifies a frequency band (referred to as a harmonic band) corresponding to a d-order harmonic (2 ≦ d ≦ D, D is an integer of 2 or more) of the center band, and the maximum value of frequency components in the vicinity of the harmonic band. Is obtained for both sides of the harmonic band, and c [i] [q] is a peak element when the frequency components of the harmonic band are larger than the two maximum values for all of d = 2 to D. Here, G3 and G4 are integers satisfying G3 ≦ G4. When the frequency analysis unit 12 matches each frequency band with each pitch (semitone) of the scale, G3 = G4 = 1 may be set. In addition, when a frequency band that further divides each pitch of the scale is used, G3 and G4 are set so as to correspond to frequencies of adjacent pitches or to correspond to frequencies that are not used between adjacent pitches. Set. The function h (d, q) is a function that returns a band number corresponding to a frequency (d-order overtone) that is d times the band q.

  A sixth method for determining the peak element is a method using Formula 15.

  This specifies the frequency band (harmonic band) corresponding to the d-order harmonic (2 ≦ d ≦ D, D is an integer of 2 or more) of the center band, and the minimum value of the frequency components in the vicinity of the harmonic band is determined as the harmonic band. This is a method in which c [i] [q] is a peak element when the frequency components of the harmonic band are larger than the two minimum values for all of d = 2 to D. Here, G3, G4, and h (d, q) are the same as those described above.

  A seventh method for determining the peak element is a method using Expression 16.

  This is because the frequency component of the d-order harmonic band is larger than the value obtained by multiplying the sum of the frequency components of the bands in the vicinity of the d-order harmonic band (2 ≦ d ≦ D, D is an integer of 2 or more) by a certain ratio. In this case, c [i] [q] is a peak element. Here, G3, G4, and h (d, q) are the same as described above, and η [d] is a predetermined constant (ratio) determined by the order of harmonics.

  In addition, in the above-described formulas 11 to 16, the calculation is performed using only the frequency component of the frame i. However, the present invention is not limited to this. For example, i + 1, i−1 and the like that are close in time to the frame i It is also possible to calculate using this frame.

  Furthermore, the above-described first to seventh methods for determining a peak element may be appropriately combined with an AND condition, and may be determined as a peak element when all the conditions are satisfied. Specifically, for example, the determination may be performed by combining the first, second, and fifth methods for determining the peak element using an AND condition as shown in Expression 17.

  In particular, any one of the second to fourth methods that are conditions related to the fundamental frequency and any one of the fifth to seventh methods that are conditions related to the harmonic component are determined in combination with an AND condition. It becomes possible to detect a music section with high accuracy.

  Returning to the description of the processing of the peak element detection unit 13 in FIG. 6, after step S230, the arithmetic processing circuit 13a determines the frame number i determined as the peak element in the peak element memory included in the peak element detection unit 13. And the data (i, q, c [i] [q]) in which the band number q and the frequency component c [i] [q] are associated with each other are stored in the format shown in FIG. 12 (step S240).

  Next, the arithmetic processing circuit 13a increases the value of the control variable q representing the frequency band by 1 (step S250). Next, the arithmetic processing circuit 13a determines whether or not the control variable q is equal to or less than the maximum band Q2 of the target range of the peak element (step S260). However, Q2 is a predetermined integer of Q2 <Q−G2, Q2 <Q−G4, and Q1 or more. If the control variable q is Q2 or less, the process returns to step S230, and if not, the process proceeds to step S270.

  Next, the arithmetic processing circuit 13a increases the value of the control variable i indicating the frame number by 1 (step S270). Next, the arithmetic processing circuit 13a determines whether or not the value of the control variable i indicating the frame number is less than the total number M of frames (step S280). When the control variable i is less than the total number M of frames, the process proceeds to step S220. When the control variable i is equal to or greater than the total number M of frames, the processing of the peak element detection unit 13 is terminated.

  Next, the processing flow of the pitch area detection unit 14 will be described with reference to the flowchart of FIG. The pitch region detection unit 14 executes processing with reference to the peak element memory of the peak element detection unit 13. First, the arithmetic processing circuit 14a of the pitch area detection unit 14 sets a control variable p indicating a frame number to 0 (step S310). Next, the arithmetic processing circuit 14a sets the control variable q indicating the frequency band to the constant Q1 used in the peak element detection unit 13 (step S320).

  Next, the arithmetic processing circuit 14a refers to the peak element memory of the peak element detection unit 13, and calculates the number C of peak elements whose frame number is p or more and less than (p + H) and whose frequency band is q. Count (step S330). Here, H is a preset integer.

  Here, a process of counting the number C of peak elements while allowing a minute fluctuation in frequency may be performed. For example, when c [i-1] [q] and c [i + 1] [q] are peak elements and c [i] [q] is not a peak element, c [i] [q-1] or c If [i] [q + 1] is a peak element, it is considered that there is a minute frequency fluctuation, and c [i] [q] is regarded as a peak element and the number C is counted.

  Next, the arithmetic processing circuit 14a determines whether or not the number C of peak elements counted in step S330 is C1 or more (step S340). Here, C1 is an integer that satisfies 0 <C1 ≦ H. When the number C of peak elements is equal to or greater than C1, the region where the frame number is greater than or equal to p and less than (p + H) and the frequency band is q is determined as a pitch region, and the process proceeds to step S350. If the data number C is less than C1, the process proceeds to step S360.

  Next, the arithmetic processing circuit 14a associates the first frame number p of the pitch region, the last frame number p + H-1 of the pitch region, and the band number q with each other in the pitch region memory included in the pitch region detecting means 14. Store in the format as shown in FIG. 13 (step S350).

  Next, the arithmetic processing circuit 14a increases the value of the control variable q representing the frequency band by 1 (step S360). Next, the arithmetic processing circuit 14a determines whether or not the control variable q is equal to or less than the constant Q2 (step S370). Here, Q2 is a predetermined integer used in the peak element detection unit 13. When the control variable q is Q2 or less, the process proceeds to step S330, and otherwise, the process proceeds to step S380.

  Next, the arithmetic processing circuit 14a increases the value of the control variable p indicating the frame number by H (step S380). Next, the arithmetic processing circuit 14a determines whether or not the value of the control variable p indicating the frame number is less than (M−H) (step S390). Here, M is the total number of frames. When the control variable p is less than (M−H), the process proceeds to step S320. When the control variable p is equal to or greater than (M−H), the process of the pitch range detection unit 14 is terminated.

  Next, the processing flow of the music section detection unit 15 will be described with reference to the flowchart of FIG. The music section detection unit 15 executes processing with reference to the pitch region memory of the pitch region detection unit 14. First, the arithmetic processing circuit 15a of the music section detecting unit 15 sets a control variable i indicating a frame number to 0 (step S410).

  Next, the arithmetic processing circuit 15a refers to the pitch region memory of the pitch region detection unit 14, and calculates the number E of pitch regions whose pitch region start position is i or more and whose pitch region end position is less than (i + J). Count (step S420). Here, J is a constant that determines the minimum length of the music section to be detected, and is an integer that satisfies J ≧ H. Usually, the length may be set to several times to several tens of times H.

  Next, the arithmetic processing circuit 15a determines whether or not the number E of pitch ranges counted in step S420 is equal to or greater than E1 (step S430). Here, E1 is an integer of 1 or more. If E is equal to or greater than E1, the section having a frame number greater than or equal to i and equal to or less than (i + J−1) is determined as a music section, and the process proceeds to step S440. On the other hand, if E is less than E1, the process proceeds to step S450.

  Next, the arithmetic processing circuit 15a associates the first frame number i of the music section with the last frame number (i + J-1) of the music section, and forms the music section detecting means 15 in the format shown in FIG. (Step S440).

  Next, the arithmetic processing circuit 15a increases the value of the control variable i indicating the frame number by J (step S450). Next, the arithmetic processing circuit 15a determines whether or not the value of the control variable i indicating the frame number is less than (M−J) (step S460). Here, M is the total number of frames. If the control variable i is less than (M−J), the process proceeds to step S420. On the other hand, if the control variable i is equal to or greater than (M−J), the process proceeds to step S470.

  Next, the arithmetic processing circuit 15a outputs the music section memory data as the music section information 3, and ends the processing of the music section detection unit 15 (step S470). When the last frame number of one music section and the first frame number of another music section are consecutive (for example, the second and third lines in FIG. 14), these are integrated into one music. You may make it perform the process made into an area.

  Here, the difference from the prior art will be described with reference to FIG. In the prior art, when a sum total obtained by adding a difference value (primary difference or secondary difference) of adjacent frequency components for a predetermined time is larger than a predetermined threshold, the music section is determined. For this reason, when a specific frequency component at a certain moment is large, the difference value and the sum are increased due to the large frequency component. Therefore, even a voice segment as shown in FIG. 5A is erroneously determined as a music segment. There was a case. As described above, in the conventional technology, it is difficult to detect music sections with sufficient accuracy for various input sources and various genres of music.

  On the other hand, in this embodiment, as described above, the music section is determined without using the sum of the difference values of adjacent frequency components. The peak element detector 13 determines whether or not the peak element is present for each frame and frequency band. The pitch region detector 14 determines the pitch when there are C1 or more peak elements in the same frequency band within the predetermined time H. When the number of pitch areas within the predetermined time J is equal to or greater than E1, the music section detection unit 15 determines that it is a music section. Accordingly, even if a specific frequency component at a certain moment is large in the voice section as shown in FIG. 5A, it is difficult to determine the pitch area by the pitch area detection unit 14, and thus it is erroneously determined as a music section. Can be suppressed.

  In addition, since the frequency component used in the prior art is different from the musical scale, it is desirable that the difference between the adjacent frequency components is equivalent to the difference between the adjacent pitches on the scale, whereas In some cases, a difference between the same pitches or a difference from a distant pitch may occur. For this reason, even in the section as shown in FIG. 5B where only a musical instrument is present, the detection accuracy may not be sufficient depending on the pitch. However, in this embodiment, the frequency analysis unit 12 can calculate the frequency component corresponding to the musical scale, and the arithmetic processing using the adjacent frequency components is directly associated with the processing operation between the adjacent pitches. Therefore, the detection accuracy of the music section can be increased.

  The overall configuration of the second embodiment of the present invention is the same as FIG. Since the processing operations of the acoustic signal input unit 11, the frequency analysis unit 12, the peak element detection unit 13, and the music section detection unit 15 are the same as those in the first embodiment, description thereof is omitted.

  The processing flow of the pitch area detection unit 14 in the present embodiment will be described with reference to the flowchart of FIG. The pitch region detection unit 14 executes processing with reference to the peak element memory of the peak element detection unit 13. First, the arithmetic processing circuit 14a of the pitch area detection unit 14 sets a control variable p indicating a frame number to 0 (step S310). Next, the arithmetic processing circuit 14a sets the control variable q indicating the frequency band to the constant Q1 used in the peak element detection unit 13 (step S320).

  Next, the arithmetic processing circuit 14a refers to the peak element memory of the peak element detection unit 13, and calculates the number C of peak elements whose frame number is greater than or equal to p and less than (p + H) and whose frequency band is q. Count (step S330). Here, H is a preset integer of 1 or more.

  Here, a process of counting the number C of peak elements while allowing a minute fluctuation in frequency may be performed. For example, when c [i-1] [q] and c [i + 1] [q] are peak elements and c [i] [q] is not a peak element, c [i] [q-1] or c If [i] [q + 1] is a peak element, it is considered that there is a minute frequency fluctuation, and c [i] [q] is regarded as a peak element and the number C is counted.

Next, the arithmetic processing circuit 14a refers to the peak element memory of the peak element detector 13, and as shown in Equation 18, the frame number is greater than or equal to p and less than (p + H), and the frequency band is q. A total sum R of values of a certain peak element c [i] [q] is calculated (step S331).

  Next, the arithmetic processing circuit 14a determines whether or not the number of data C counted in step S331 is equal to or greater than C1 and R calculated in step S332 is equal to or greater than a predetermined value R1 (step S332). Here, C1 is an integer that satisfies 0 <C1 ≦ H. When the number of data C is C1 or more and R is R1 or more, it is determined that the region where the frame number is p or more and less than (p + H) and the frequency band is q is the pitch region. The process proceeds to S350. Otherwise, the process proceeds to step S360.

  Next, the arithmetic processing circuit 14a associates the frequency band q, the first frame number p of the pitch range, and the last frame number p + H-1 of the pitch range, and the pitch range included in the pitch range detection means 14 is provided. The data is stored in the memory in the format as shown in FIG. 13 (step S350).

  Next, the arithmetic processing circuit 14a increases the value of the control variable q representing the frequency band by 1 (step S360). Next, the arithmetic processing circuit 14a determines whether or not the control variable q is equal to or less than the constant Q2 (step S370). Here, Q2 is a predetermined integer used in the peak element detection unit 13. When the control variable q is Q2 or less, the process proceeds to step S330, and otherwise, the process proceeds to step S380.

  Next, the arithmetic processing circuit 14a increases the value of the control variable p indicating the frame number by H (step S380). Next, the arithmetic processing circuit 14a determines whether or not the value of the control variable p indicating the frame number is less than (M−H) (step S390). Here, M is the total number of frames. When the control variable p is less than (M−H), the process proceeds to step S320, and when the control variable p is equal to or greater than (M−H), the processing of the pitch region detection unit 14 is terminated.

  In the present embodiment, in the process of step S332 of the pitch region detection unit 14, there are C1 or more peak elements in the same frequency band within the predetermined time H, and the sum R of the peak element values is the predetermined value R1 or more. Therefore, the possibility that the voice section is erroneously determined as the music section is further reduced.

  The overall configuration of the third embodiment of the present invention is the same as that shown in FIG. And since each processing operation of the acoustic signal input unit 11, the frequency analysis unit 12, and the peak element detection unit 13 is the same as that of the first embodiment, the description thereof is omitted.

  The processing flow of the pitch area detection unit 14 in the present embodiment will be described based on the flowchart shown in FIG. However, since the processing of steps other than step S351 in the figure is the same as that of the second embodiment, the description thereof is omitted and only step S351 is described.

  In step S351 in the figure, the arithmetic processing circuit 14a of the pitch region detecting unit 14 calculates the frequency band q, the first frame number p of the pitch region, the last frame number p + H-1 of the pitch region, and the above-described step S332. The corresponding R is stored in the pitch area memory of the pitch area detecting means 14 in the format as shown in FIG.

  Next, the operation of the music section detection unit 15 will be described with reference to the flowchart of FIG. First, the arithmetic processing circuit 15a of the music section detecting unit 15 sets a control variable i indicating a frame number to 0 (step S410). Next, the arithmetic processing circuit 15a refers to the pitch region memory of the pitch region detection unit 14, and determines the intensity R of the pitch region whose pitch region start position is i or more and whose pitch region end position is less than (i + J). The sum Y is calculated (step S421). Here, J is a constant that determines the minimum length of the music section to be detected, and is an integer that satisfies J ≧ H. Usually, the length may be set to several times to several tens of times H.

Next, the arithmetic processing circuit 15a determines whether or not Y calculated in step S421 is Y1 or more (step S431). Here, Y1 is a predetermined constant. If Y is equal to or greater than Y1, it is determined that a section with a frame number greater than or equal to i and equal to or less than (i + J−1) is a music section, and the process proceeds to step S440. On the other hand, if Y is not equal to or greater than Y1 , the process proceeds to step S450.

  Next, the arithmetic processing circuit 15a associates the first frame number i of the music section with the last frame number (i + J-1) of the music section and stores it in the music section memory provided in the music section detecting means 15. (Step S440).

  Next, the arithmetic processing circuit 15a increases the value of the control variable i indicating the frame number by J (step S450). Next, the arithmetic processing circuit 15a determines whether or not the value of the control variable i indicating the frame number is less than (M−J) (step S460). Here, M is the total number of frames. When the control variable i is less than (M−J), the process proceeds to step S421, and when the control variable i is equal to or greater than (M−J), the process proceeds to step S470.

  Next, the arithmetic processing circuit 15a outputs the music section memory data as the music section information 3, and ends the processing of the music section detection unit 15 (step S470). When the last frame number of one music section and the first frame number of another music section are consecutive (for example, the second and third lines in FIG. 14), these are integrated into one music. You may make it perform the process made into an area.

  As described above, in the present embodiment, in the process of step S332 of the pitch region detection unit 14, there are C1 or more peak elements in the same frequency band within the predetermined time H, and the total sum R of the peak element values is In addition to determining the pitch region when the value is equal to or greater than the predetermined value R1, in the process of step S431 of the music section detection unit 15, the sum of Y in the predetermined time J is calculated to determine the music section. Therefore, it is suitable for the determination of the music section when the musical instrument is sounded with a relatively loud sound and the number of times of sounding is small.

  As described above, since the present invention can detect a music section from content including an audio signal, a video device and a content search apparatus having a function of dividing the content into small sections (chapter) according to the contents. Useful for.

It is a block diagram which shows the whole structure of the acoustic signal analyzer 1 in the 1st-3rd Example of this invention. It is a flowchart for demonstrating the processing operation of the frequency analysis part 12 in the 1st-3rd Example of this invention. It is a figure for demonstrating the flame | frame creation operation | movement of the frequency analysis part 12 in the 1st-3rd Example of this invention. It is the figure which showed typically the characteristic of the filter group used by the frequency component calculation operation | movement of the frequency analysis part 12 in the 1st-3rd Example of this invention. It is the figure which showed typically the characteristic of the matrix data produced in the frequency analysis part 12 in the 1st-3rd Example of this invention. It is a flowchart for demonstrating the processing operation of the peak element detection part 13 in the 1st-3rd Example of this invention. It is a flowchart for demonstrating the processing operation of the pitch area | region detection part 14 in 1st Example of this invention. It is a flowchart for demonstrating the processing operation of the music area detection part 15 in the 1st and 2nd Example of this invention. It is a flowchart for demonstrating the processing operation of the pitch area | region detection part 14 in 2nd Example of this invention. It is a flowchart for demonstrating the processing operation of the pitch area | region detection part 14 in the 3rd Example of this invention. It is a flowchart for demonstrating the processing operation of the music area detection part 15 in 3rd Example of this invention. It is the figure which showed the data storage format of the peak element detection part 13 in the 1st-3rd Example of this invention. It is the figure which showed the data storage format of the pitch area | region detection part 14 in the 1st and 2nd Example of this invention. It is the figure which showed the data storage format of the music area detection part 15 in the 1st-3rd Example of this invention. It is the figure which showed the data storage format of the pitch area | region detection part 14 in the 3rd Example of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Acoustic signal analyzer 2 Acoustic signal 3 Music area information 11 Acoustic signal input part 12 Frequency analysis part 13 Peak element detection part 14 Pitch area detection part 15 Music area detection part 11a-15a Arithmetic processing circuit

Claims (14)

The acoustic signal is divided into a plurality of frequency bands, a frequency analysis means for generating a frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands as elements,
A value calculated based on the condition that the element of the frequency component data is read and the read element is equal to or greater than a predetermined threshold, or the read element is a frequency component element near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element. Peak element detection means for detecting an element satisfying at least one condition as a peak element;
For each of the plurality of frequency bands, the first counts the number of pre-Symbol peak elements for each section, the count result is the first section is greater than or equal to a first predetermined number having a time length that includes the unit time Pitch range detection means for detecting the pitch range as a pitch range in the frequency band ;
For each second section having a time length including the first section, the number of the pitch ranges is counted for each of the plurality of frequency bands , and the count result is equal to or greater than a second predetermined number. , sound signal analysis apparatus characterized by comprising a music segment detection means for detecting the second interval as a music segment. The acoustic signal is divided into a plurality of frequency bands, a frequency analysis means for generating a frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands as elements,
A value calculated based on the condition that the element of the frequency component data is read and the read element is equal to or greater than a predetermined threshold, or the read element is a frequency component element near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element. Peak element detection means for detecting an element satisfying at least one condition as a peak element;
For each of the plurality of frequency bands, the number of peak elements and the sum of the peak element values are counted for each first section having a time length including the unit time, and the number of peak elements is the first. A pitch region detecting means for detecting the first section as a pitch region in the frequency band when the sum of the peak element values is not less than a first predetermined value .
For each second section having a length of time including the first section, the number of the pitch ranges is counted for each of the plurality of frequency bands , and the count result is equal to or greater than a second predetermined value. , sound signal analysis apparatus characterized by comprising a music segment detection means for detecting the second interval as a music segment. By dividing the acoustic signals into a plurality of frequency bands, a frequency analysis means for generating a frequency component data to the component intensity per unit time and the elements in each of the plurality of frequency bands,
A value calculated based on the condition that the element of the frequency component data is read and the read element is equal to or greater than a predetermined threshold, or the read element is a frequency component element near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element. Peak element detection means for detecting an element satisfying at least one condition as a peak element;
For each of the plurality of frequency bands, the number of peak elements and the sum of the peak element values are counted for each first section having a time length including the unit time, and the number of peak elements is the first. When the sum of the peak element values is equal to or greater than a first predetermined value, the first section is detected as a pitch region in the frequency band, and the peak element value is A pitch region detecting means for setting the sum of the pitch region intensity of the pitch region,
For each second section having a time length including the first section, the sum of the pitch range intensities is counted for each of the plurality of frequency bands, and the sum is equal to or greater than a second predetermined value. , sound signal analysis apparatus characterized by comprising a music segment detection means for detecting the second interval as a music segment.   4. The system according to claim 1, wherein the first section has a time length including a plurality of the unit times, and the second section has a time length including a plurality of the first sections. The acoustic signal analyzer according to item. The acoustic signal analyzer according to claim 1, wherein the plurality of frequency bands are frequency bands corresponding to a pitch frequency of a scale.   The acoustic signal analyzer according to any one of claims 1 to 5, wherein the unit time differs according to each of the plurality of frequency bands.   The pitch range detection means uses a value smaller than the number of unit times included in the first section as the first predetermined number. Acoustic signal analyzer.   The pitch region detecting means, when the peak element, the element that is not the peak element, and the peak element are temporally continuous in one frequency band, the element that is not the peak element is the peak element. The acoustic signal analyzer according to claim 1, wherein The acoustic signals by a plurality of divided frequency bands, and frequency analysis step of generating frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands as elements,
A value calculated based on the condition that the element of the frequency component data is read and the read element is equal to or greater than a predetermined threshold, or the read element is a frequency component element near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element. A peak element detection step of detecting, as a peak element, an element that satisfies at least one condition;
For each of the plurality of frequency bands, the first counts the number of pre-Symbol peak elements for each section, the count result is the first section is greater than or equal to a first predetermined number having a time length that includes the unit time A pitch region detecting step for detecting the pitch region as a pitch region in the frequency band ;
For each second section having a time length including the first section, the number of the pitch ranges is counted for each of the plurality of frequency bands , and the count result is equal to or greater than a second predetermined number. , sound signal analysis method, which comprises have a music-segment detection step of detecting the second section as music section. The acoustic signals by a plurality of divided frequency bands, and frequency analysis step of generating frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands as elements,
A value calculated based on the condition that the element of the frequency component data is read and the read element is equal to or greater than a predetermined threshold, or the read element is a frequency component element near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element. A peak element detection step of detecting, as a peak element, an element that satisfies at least one condition;
For each of the plurality of frequency bands, the number of peak elements and the sum of the peak element values are counted for each first section having a time length including the unit time, and the number of peak elements is the first. A pitch region detecting step of detecting the first section as a pitch region in the frequency band when the sum of the peak element values is equal to or greater than a first predetermined value .
For each second section having a length of time including the first section, the number of the pitch ranges is counted for each of the plurality of frequency bands , and the count result is equal to or greater than a second predetermined value. , sound signal analysis method, which comprises have a music-segment detection step of detecting the second section as music section. By dividing the acoustic signals into a plurality of frequency bands, a frequency analysis step of generating frequency component data to the component strength per unit time in each of the plurality of frequency band element,
A value calculated based on the condition that the element of the frequency component data is read and the read element is equal to or greater than a predetermined threshold, or the read element is a frequency component element near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element. A peak element detection step of detecting, as a peak element, an element that satisfies at least one condition;
For each of the plurality of frequency bands, the number of peak elements and the sum of the peak element values are counted for each first section having a time length including the unit time, and the number of peak elements is the first. When the sum of the peak element values is equal to or greater than a first predetermined value, the first section is detected as a pitch region in the frequency band, and the peak element value is A pitch region detection step in which the sum of the pitch region is the pitch region intensity of the pitch region,
For each second section having a time length including the first section, the sum of the pitch range intensities is counted for each of the plurality of frequency bands, and the sum is equal to or greater than a second predetermined value. , sound signal analysis method, which comprises have a music-segment detection step of detecting the second section as music section. On the computer,
The acoustic signals by a plurality of divided frequency bands, and frequency analysis step of generating frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands as elements,
A value calculated based on the condition that the element of the frequency component data is read and the read element is equal to or greater than a predetermined threshold, or the read element is a frequency component element near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element. A peak element detection step of detecting, as a peak element, an element that satisfies at least one condition;
For each of the plurality of frequency bands, the first counts the number of pre-Symbol peak elements for each section, the count result is the first section is greater than or equal to a first predetermined number having a time length that includes the unit time A pitch region detecting step for detecting the pitch region as a pitch region in the frequency band ;
For each second section having a time length including the first section, the number of the pitch ranges is counted for each of the plurality of frequency bands , and the count result is equal to or greater than a second predetermined number. , sound signal analysis program for executing the music-segment detection step of detecting the second section as music section. On the computer,
The acoustic signals by a plurality of divided frequency bands, and frequency analysis step of generating frequency component data to the component intensity per unit time at pre-Symbol respective plurality of frequency bands as elements,
A value calculated based on the condition that the element of the frequency component data is read and the read element is equal to or greater than a predetermined threshold, or the read element is a frequency component element near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element. A peak element detection step of detecting, as a peak element, an element that satisfies at least one condition;
For each of the plurality of frequency bands, the number of peak elements and the sum of the peak element values are counted for each first section having a time length including the unit time, and the number of peak elements is the first. A pitch region detecting step of detecting the first section as a pitch region in the frequency band when the sum of the peak element values is equal to or greater than a first predetermined value .
For each second section having a length of time including the first section, the number of the pitch ranges is counted for each of the plurality of frequency bands , and the count result is equal to or greater than a second predetermined value. , sound signal analysis program for executing the music-segment detection step of detecting the second section as music section. On the computer,
By dividing the acoustic signals into a plurality of frequency bands, a frequency analysis step of generating frequency component data to the component strength per unit time in each of the plurality of frequency band element,
A value calculated based on the condition that the element of the frequency component data is read and the read element is equal to or greater than a predetermined threshold, or the read element is a frequency component element near the read element. Or a harmonic element that is an element corresponding to the harmonic component of the read element is larger than a value calculated based on a frequency component element in the vicinity of the harmonic element. A peak element detection step of detecting, as a peak element, an element that satisfies at least one condition;
For each of the plurality of frequency bands, the number of peak elements and the sum of the peak element values are counted for each first section having a time length including the unit time, and the number of peak elements is the first. When the sum of the peak element values is equal to or greater than a first predetermined value, the first section is detected as a pitch region in the frequency band, and the peak element value is A pitch region detection step in which the sum of the pitch region is the pitch region intensity of the pitch region,
For each second section having a time length including the first section, the sum of the pitch range intensities is counted for each of the plurality of frequency bands, and the sum is equal to or greater than a second predetermined value. , sound signal analysis program for executing the music-segment detection step of detecting the second section as music section.

Images