JP2001283569A - Rust search device - Google Patents

Abstract

(57) [Summary] [Problem] To detect a rust portion from music information. SOLUTION: Music information for one tune is read from a storage medium such as a CD, converted into PCM data as required, and the converted waveform data is subjected to FFT processing every second to obtain the total power thereof. After the audibility processing, the highest audible frequency is obtained. The data detected every second is 1
Inspection is performed every 5 seconds, and the total power exceeds the threshold of 10 points or more and the highest audible frequency exceeds the threshold of 10 points or more. A section having a different peak frequency when FFT processing of 500 Hz or less is performed on the waveform data in the section is detected. That is, a section having a relatively large volume and a relatively high pitch, and having a different tempo between the section and the section before and after the section, that is, a section having a general feature of a “rust” portion is detected and is detected. Identify as part.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a rust search device for detecting a rust portion, which is the most characteristic portion, from music.

[0002]

2. Description of the Related Art Conventionally, for example, in a cassette tape player or the like, when a desired music piece is selected from a plurality of music pieces recorded on a cassette tape and played back, the tape is fast-forwarded and listened to. Searching the storage location of the desired music while searching for the storage location of the desired music from a pre-stored music title list in which the music name and its storage location are described, and reproducing this portion To reproduce a desired music. Further, there may be a case where a silent part between music pieces is detected, a head part of the next music piece is detected, and a desired music piece is searched by repeating this.

[0003] For example, a compact disk (CD)
And mini-disc (MD) and the like, the storage location is stored together with the music information, and this is used to detect the head of the music, or to reproduce only the head of each music in order. There is also known a method of searching for a target music.

[0004]

However, in the above-described music reproducing method, if the music whose title is unknown or the music at the start of the music is not known, the desired music is reproduced. Is difficult to search for, and there is a problem that it takes time and effort. In particular, when only the rust part of the music is known, the search for the beginning of the music is first performed, and then the rust part is searched from among them, and at this point it can be determined whether or not the desired music is obtained. Therefore, there is a problem that it takes time to detect a desired music.

[0005] For this reason, when a CD store or the like tries to purchase a CD that stores only the rust portion, the user cannot easily select the target CD.

Accordingly, the present invention has been made in view of the above-mentioned conventional unsolved problem, and has as its object to provide a rust search device capable of easily detecting a characteristic rust portion of a music piece. The purpose is.

[0007]

In order to achieve the above object, a rust search device according to a first aspect of the present invention provides a rust search apparatus in which a distribution condition of frequency components in a predetermined time set from music information to be searched is set in advance. High frequency range detecting means for detecting a section higher than a value, and high power range detecting means for detecting a section in which the distribution of the power level of the frequency component at the specified time is greater than a predetermined threshold from the music information. And a modulation region detecting means for detecting a section in which the distribution state of the frequency component in the predetermined frequency range set in advance from the music information is different between the distribution state at the specified time and the distribution states in sections before and after the specified time. A section corresponding to the rust detected by the area detection means and detected by the high power area detection means and detected by the modulation area detection means; It is characterized by and a rust identifying means for identifying Te.

In the invention according to the first aspect, the music information to be searched is set based on, for example, a performance time of a general rust portion based on a processing result of performing FFT processing for each unit time, and the like. The distribution state of the frequency component is detected every time, and the distribution state exceeds the threshold,
A section that is distributed in a relatively high frequency range and can be regarded as having a relatively high pitch is detected. Is high, and a section in which the volume is considered to be relatively high is detected. Furthermore, the distribution of relatively low frequency components below a preset frequency is different between the distribution at the specified time and the distribution before and after the specified period, that is, the frequency range where the pitch is relatively low and the pitch change is relatively small. Is detected, a section in which the distribution status is different, that is, a section in which the tempo changes around a specified time is detected. Then, a section in which the pitch is relatively high, the volume is relatively large, and the tempo changes before and after that, that is, a section having a characteristic of a general rust portion is detected, and this is specified as the rust portion.

Further, the rust search device according to claim 2 is
The determination frequency range is set based on a frequency corresponding to a rhythm instrument.

According to the second aspect of the present invention, the determination frequency range is set based on a frequency corresponding to a rhythm instrument such as a drum, and a change in the distribution state is detected by focusing on a frequency component corresponding to the rhythm instrument. Therefore, it is possible to accurately detect a change in tempo in the music.

[0011]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a CD / CD to which the present invention is applied.
This is an example of an MD player (CD; compact disc, MD; mini disc).

The CD / MD player 100 reads music information from a CD or MD storing a plurality of songs and the like, decodes compressed data as necessary, and generates an audio waveform, similarly to a known CD / MD player. PCM equivalent to
A reading unit 11 that converts the data into waveform data, and controls the reading unit 11 in accordance with an operation instruction from an operation unit 12,
A signal processing unit 15 temporarily stores the waveform data read by the reading unit 11 in the memory 13 and then outputs the waveform data to the audio reproducing unit 14 at a predetermined timing.

The signal processing unit 15 is a known CD / MD
The reading unit 11 is controlled in the same manner as the player, and the music information corresponding to the music designated by the operation unit 12 is read,
The waveform data converted into PCM data is input from the reading unit 11 as needed. Then, the sequentially input waveform data is temporarily stored in the memory 13, and then output to the audio reproducing unit 14 at a predetermined timing. The audio reproducing unit 14 converts the waveform data into an analog signal and performs low-pass filtering, amplification, and the like. After that, playback and output are performed.

The signal processing unit 15 controls the reading unit 11 when the operation unit 12 instructs a rust search for detecting a "rust" portion representing the feature of the music, and controls the reading unit 11 for one tune of the specified music. Is read out at a faster speed than during normal reproduction, and if necessary, waveform data converted into PCM data is input from the reading unit 11 and stored in the memory 13. This memory 13 is formed so as to be able to store music information for one music.

Then, a search is made for the waveform data of one piece of music stored in the memory 13, which is a section of about 15 seconds, which has a relatively high pitch, a relatively high volume, and a change in tempo. The section is detected as a section corresponding to “rust”.

That is, in the case of songs that are currently widely distributed and have high sales, in most cases, the “rust” portion, which most represents the characteristics of the song, is used for advertising the song itself or advertising other products. Because it is used as an advertising medium, it is made to remain relatively impressive to the listener. In addition, it is often made in consideration of use in an advertising medium, and the rust portion is made to fit within a certain fixed time of about 15 seconds in accordance with the advertising time on a television or the like. In addition, the rust part has a relatively high volume compared to other parts in order to remain in the listener's impression, and the pitch is relatively high to make it easier to hear. Is often changed.

Therefore, the signal processing unit 15 performs FFT processing on the waveform data for one music piece, and based on the frequency spectrum, has a relatively large volume and a high pitch every 15 seconds or so. A section in which the tempo changes before and after the predetermined time is searched, and this is regarded as a rust portion. FIG. 2 is a flowchart illustrating a processing procedure of the signal processing unit 15 when the rust search is designated.

In the signal processing unit 15, when the rust search is instructed by the operation unit 12, first, in step S2, the reading unit 11 is controlled, and the music information of one tune stored in the storage location of the specified tune is first controlled. Is read. Then, the waveform data read by the reading unit 11 and converted into PCM data as necessary is stored in the memory 13.

Then, the process proceeds to step S4, where the pointer n is set to n = 0.
3 is extracted sequentially from the beginning of the waveform data for one song stored in No. 3 for one second in the section specified by the pointer n, and FFT is performed on the extracted waveform data Dn.
Perform processing.

Then, the process proceeds to step S8, where the sum of the power of each frequency component is calculated based on the result of the FFT processing.
This is stored in the power data register Pn.

Subsequently, the flow shifts to step S10, where FFT is performed.
A hearing process is performed based on the processing result. In other words, if a loud sound and a loud sound enter the human ear at the same time in close proximity in frequency, there is a characteristic that the loud sound is not noticed. Processing such as removal is performed.

Next, the process proceeds to step S12, in which the highest frequency is specified from the frequency components after the audibility processing and stored in the frequency data register Fn. That is, by taking into account the audible model, the highest frequency is specified from the sounds that can be distinguished by humans, thereby specifying the sound that sounds highest in frequency.

Next, the process proceeds to step S14, where it is determined whether or not processing has been performed for all the waveform data. If not, the process proceeds to step S16, where the pointer n is updated to n = 1, and then the process returns to step S6. . on the other hand,
When the processing has been performed for all the waveform data, the process proceeds to step S20, and the rust search processing is performed.

That is, as shown in FIG. 3, first, in step S31, the pointer n is cleared to n = 0. Next, the process proceeds to step S32, in which the power data registers Pn to
The data of P (n + 14) is read out, and it is determined whether or not there are 10 or more power data exceeding the preset threshold value (step S33). The threshold value of the power data is set to a value that can be considered to be relatively large in volume, and by comparing with the threshold value, the section n to n + 14 specified by the pointer n is determined.
Determines whether the volume is relatively high.

If the power levels of ten or more points do not exceed the threshold value, the flow shifts to step S40 to update the pointer n to n = n + 1, and then returns to step S31. On the other hand, if the power data of ten or more points exceeds the threshold value, the process proceeds to step S34, and the data of F (n + 14) is read from the frequency data register Fn. Then, it is determined whether or not there are ten or more frequencies exceeding a preset threshold value (step S35). The threshold of the frequency is
The pitch is set to a value that can be considered relatively high,
By comparing with the threshold value, it is determined whether the interval n to n + 14 specified by the pointer n has a relatively high pitch.

If the frequencies of ten or more points do not exceed the threshold value, the flow returns to step S40 to perform the same processing as described above. The process moves to S36.

In this step S36, sections n to n + 14
An FFT process at a frequency of 500 Hz or less is performed on the waveform data in the sections n-2 to n-10, which is a section earlier than the above, and the peak frequency Pf1 is specified. Similarly, in step S37, FFT processing at a frequency of 500 Hz or less is performed on the waveform data in the section n to n + 14, and the peak frequency Pf2 is specified. Further, step S
At 38, the frequency 50 is added to the waveform data in the section n + 16 to n + 24, which is a section after the section n to n + 14.
FFT processing of 0 Hz or less is performed, and the peak frequency Pf
3 is specified.

The reason why the FFT processing is performed at a frequency of 500 Hz or less is to detect the peak frequency of the waveform data of the rhythm instrument.

Next, the process proceeds to step S39, in which the peak frequency P specified in steps S36 to S38 is determined.
It is determined whether f1, Pf2, and Pf3 are Pf1 ≠ Pf2 ≠ Pf3. That is, it is determined whether the frequency of the rhythm instrument in the section n to n + 14 is different from the frequency in the preceding and following sections, that is, whether the tempo has changed. If the tempo has not changed, step S
Move to 40.

On the other hand, in step S39, Pf1 {Pf2}
If it is Pf3 and it is determined that the tempo is changing, then this section n to n + 14 is determined to be a section corresponding to "rust".

Then, returning to FIG. 2, the process proceeds to step S22, in which the waveform data of the specified section n to n + 14 is extracted from the memory 13, and is output to the audio reproducing unit 14 to instruct reproduction thereof.

Here, steps S32 and S33 in FIG.
Corresponds to the high power range detecting means, and the process in step S34
And the processing of S35 correspond to the high frequency range detection means, the processing of steps S36 to S39 correspond to the modulation detection means,
In step S33, power data exceeding the threshold is 10
When it is determined that there is a point or more, the process goes through step S34 to S35
When it is determined in step S35 that there are 10 or more frequencies exceeding the threshold, steps S36 and S3
7, the process proceeds to step S39 via S38, and then proceeds to step S39.
When Pf1 ≠ Pf2 ≠ Pf3 at 39, this section n-
The process of determining n + 14 as a rust portion corresponds to the rust specifying means.

Next, the operation of the above embodiment will be described.

To perform a rust search by the CD / MD player 100, first, a CD or MD is set in the reading unit 11, and an rust search is instructed on the operation unit 12, and for example, a music name and a music name stored in advance are stored. A storage location of a desired music piece is designated from a music title list or the like in which the storage location is described.

Thus, the signal processing unit 15 controls the reading position of the reading unit 11 so as to read the music information at the designated storage position. The reading unit 11 reads the music information at the designated position, converts it into PCM data as necessary, and outputs it to the signal processing unit 15 as waveform data.
Then, the signal processing unit 15 stores the waveform data of one music in the memory 13 (step S2).

After the pointer n is set to n = 0 (step S4), first one second of waveform data Dn of the music is taken out of the memory 13 and FFT processing is performed (step S6). Then, based on the result of the FFT processing, the sum of the power of the waveform data Dn in all the frequency bands, that is, an index corresponding to the volume of the waveform data Dn is calculated and stored in the power data register Pn (step S8).
Further, after performing a predetermined audibility process on the result of the FFT process (step S10), the highest audible frequency, that is,
An index corresponding to the pitch of the waveform data Dn is detected and stored in the frequency data register Fn (step S12).

This process is repeated, and if an index corresponding to the volume and the pitch is detected every second for the waveform data of one music piece, the process shifts from step S14 to step S20 to perform a rust search process.

That is, first, the pointer n is cleared to n = 0 (step S31), and the power data register P
The data of n to P (n + 14) are read, and among them,
It is determined whether or not there are 10 or more data exceeding the preset threshold value, that is, whether or not the volume is relatively high (steps S32 and S33). If there are 10 or more data points exceeding the threshold value, then, for the frequency data register Fn in the same section, whether or not there are 10 or more points where the highest audible frequency exceeds the threshold value, that is, the pitch is relatively high It is determined whether it is high (steps S34, S35).

The power data registers Pn-P
When ten or more points of the (n + 14) data do not exceed the threshold value, or when the frequency data registers Fn to F (n
When at least 10 points out of the +14) data do not exceed the threshold value, that is, when the volume is not relatively large or the pitch is not relatively large, the pointer n is updated as not being a "rust" part (step S14). S40), a search is performed for the next section.

The power data registers Pn-P
When 10 or more of the data of (n + 14) exceed the threshold value and 10 or more of the data of the frequency data registers Fn to F (n + 14) exceed the threshold value,
Section n-2 to n-10, Section n to n + 14, Section n + 16 to n
+24 for the waveform data in each section
FFT processing of 0 Hz or less is performed, and peak frequencies in each section are calculated (steps S36 to S3).
8). Then, it is determined whether or not the peak frequencies are all different, that is, whether the tempo of the section n to n + 14 has changed from the preceding and following sections (step S3).
9).

As described above, in general, the "rust" portion has a relatively large volume, a relatively high pitch, and a change in tempo. Is in the middle of the song, from the beginning to the middle "rust", the volume is not relatively high, the pitch is not relatively high, and the tempo does not change, The processing returns to step S40 without satisfying the condition in the processing of step S33, step S35, or step S39.

Since the rust portion in the middle of the song has a relatively large volume and a relatively high pitch, when the determination is made for the sections n to n + 14 corresponding to the rust portion, steps S32 and S33 are performed. To steps S34 and S35
Then, the process proceeds to step S36, and the sections n to n +
500 for the waveform data in 14 and the section before and after
When the FFT processing at less than or equal to Hz is performed and the tempo is determined (from step S36 to step S39), the tempo of the “rust” portion changes, that is, the pitch of the rhythm instrument changes, so that the interval n to n + 14 The peak frequencies before and after the section are different. Therefore, step S3
9, the peak frequencies Pf1, Pf2, and Pf in each section
3 are different, it is specified that the sections n to n + 14 are rust portions.

Then, the waveform data corresponding to the sections n to n + 14 is read out from the memory 13 and sequentially output to the audio reproducing section 14 at a predetermined speed (step S2).
2).

The rust portion is reproduced by the audio reproducing unit 14 reproducing the input waveform data.

In this state, when performing a rust search for another piece of music, the operation unit 12 is operated in the same manner as described above to instruct the rust search and the storage location. As a result, the reading section 11 reads out the music information of the designated storage location for one music in the same manner as described above, and the signal processing section 15 stores the read music information in the memory 13 as waveform data.
Then, in the same manner as described above, the total of the power in the frequency band and the highest audible frequency are obtained every second, and based on this, the power is relatively large, the frequency is relatively high, and the sections n to n + 14 are calculated. It is determined whether the tempo has changed between the preceding and following sections.

Then, the parts satisfying these are called "rust".
The waveform data corresponding to this section is output to the audio reproduction unit 14, and the rust portion is reproduced.

On the other hand, when a new music reproduction is instructed by the operation unit 12, the signal processing unit 12 controls the reading unit 11 to sequentially read the designated music information from the CD, stores the music information in the memory 13, and stores it in the memory 13. And outputs it to the audio reproducing unit 14 at a predetermined speed. Thereby, the audio reproducing unit 14
Reproduces the desired music.

As described above, since the rust portion in the music can be detected, even when searching for a music in which only the rust portion has been heard, the rust portion can be easily detected without reproducing the entire tune. Rust can be reproduced. Therefore, for example, even in the case of searching for a song that has heard only the rust portion in an advertisement or the like, since only the rust portion can be reproduced, it is possible to reproduce only the rust portion and sequentially replace the music. A desired music can be easily detected. If the name of the music is known, it is possible to easily confirm whether or not the music is the target music by reproducing the rust portion.

In particular, when only a rust portion is known at a store such as a CD or the like, it takes a great deal of time to identify a CD containing the target music. Is quickly extracted and regenerated, which is effective.

Further, since it can be realized by adding only the processing procedure of the signal processing unit 15 in the conventional CD / MD player, there is no need to change the CD or MD side, and it can be easily realized. .

In the reading section 11, when the rust search is instructed, the CD or MD music information is read at a higher speed than in the normal reproduction, so that the speed can be increased.

In the above embodiment, the CD /
Although the case where the present invention is applied to the MD player has been described, the present invention is not limited to this, and the DCC (Digital Compact)
Cassette), DAT (Digital Aud)
io Tape) or a playback device such as an analog compact cassette. In the case of a playback device such as an analog compact cassette, the start and end of a music piece are detected by detecting a silent portion between the music pieces. What is necessary is just to detect. Moreover, you may apply to a karaoke system etc.

Also, for example, mp3 compressed music information
The present invention is applied to a music reproducing apparatus in which compressed music information is stored in a memory, and the compressed music information is read out from the memory and reproduced, such as a music reproducing apparatus in which a file is reproduced by a player. It is also possible. In this case, a rust search is performed on the music information before compressing the CD or MD music information, and for example, a section specified as a rust portion is stored in a memory in association with identification information specifying the music information. When rust playback is instructed, rust playback is instructed if a rust section corresponding to the specified music is searched, and music information corresponding to the specified section is read out from the memory and reproduced. Sometimes the rust can be quickly regenerated.

In the above-described embodiment, the case where the rust portion is detected when detecting a section having a different tempo from the preceding and following sections has been described. However, the present invention is not limited to this. A section in which the volume is high may be detected, or a section in which the tempo differs from the preceding and following sections and in which the volume is high may be detected.

In the above-described embodiment, the FFT processing is performed on the waveform data every second. However, the present invention is not limited to this.
What is necessary is just to perform T process.

In the above embodiment, the interval n
The case where the judgment of the rust portion is performed in units of .about.n + 14, that is, in units of 15 seconds has been described. However, the present invention is not limited to this, and any time corresponding to the rust playing time may be used.

Further, in the above embodiment, a case has been described where the FFT processing is performed on a frequency of 500 Hz or less in order to detect a change in tempo. However, the present invention is not limited to this. Any frequency can be used as long as a change in tempo can be detected.

Further, in the above embodiment, the case where the waveform data of one music piece is read out and stored in the memory 13 has been described.
Since the portion is used repeatedly in the song, it is not necessary to read out the entire waveform data of one song. For example, it is possible to read out about half the waveform data of the entire song and make a judgment only on this portion. .

[0059]

As described above, according to the first aspect of the present invention,
According to the rust search device according to the above, for the music information to be searched, the frequency is relatively high, the power level of the frequency is relatively large, and the frequency distribution in the determination frequency range changes before and after the specified time. Since the period is detected and is set as the rust portion, the rust portion can be easily and reliably detected.

According to the rust search device of the second aspect, the determination frequency range is set based on the frequency corresponding to the rhythm instrument, so that the distribution of the frequency components in the frequency range corresponding to the rhythm instrument is set. , The change in tempo can be accurately detected.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram showing a configuration of a CD / MD player to which the present invention is applied.

FIG. 2 is a flowchart illustrating an example of a processing procedure in a signal processing unit 15 of FIG. 1;

FIG. 3 is a flowchart illustrating an example of a processing procedure of a rust search process.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 Reading part 12 Operation part 13 Memory 14 Sound reproduction part 15 Signal processing part

Claims (2)

[Claims] 1. A high frequency range detecting means for detecting a section in which a distribution state of frequency components at a preset specified time is higher than a preset threshold value from music information to be searched, and said predetermined time based on said music information. A high power range detecting means for detecting a section in which the distribution state of the power level of the frequency component is larger than a preset threshold value, and the distribution state of the frequency component in the decision frequency range preset from the music information is defined by the above-mentioned specification. A modulation area detecting means for detecting a section different in distribution status in time and a distribution situation in sections before and after the time, and a modulation area detecting means detecting by the high frequency area detecting means and detecting by the high power area detecting means; And a rust specifying unit for specifying the section detected in the step as a portion corresponding to the rust. 2. The rust search device according to claim 1, wherein the determination frequency range is set based on a frequency corresponding to a rhythm instrument.