CN102568454A - Method and device for analyzing music BPM (Beat Per Minutes) - Google Patents
Method and device for analyzing music BPM (Beat Per Minutes) Download PDFInfo
- Publication number
- CN102568454A CN102568454A CN201110415218XA CN201110415218A CN102568454A CN 102568454 A CN102568454 A CN 102568454A CN 201110415218X A CN201110415218X A CN 201110415218XA CN 201110415218 A CN201110415218 A CN 201110415218A CN 102568454 A CN102568454 A CN 102568454A
- Authority
- CN
- China
- Prior art keywords
- music
- bpm
- snatch
- frame
- refrain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Auxiliary Devices For Music (AREA)
- Telephone Function (AREA)
Abstract
The invention provides a method and a device for analyzing music BPM (Beat Per Minutes). The method comprises the following steps: S1. identifying musical clips of N preset clip types from music to be analyzed, wherein N is a preset positive integer, and the preset clip type is at least one of refrain, interlude and verse; S2. respectively calculating BPM for each musical clip; and S3. combining BPM calculated for all musical clips to obtain the BPM of music to be identified. According to the method and the device, the accuracy of the BPM of the analyzed music can be improved.
Description
[technical field]
The present invention relates to the Computer Processing technology, the method and apparatus of particularly a kind of analysis music per minute beat number (BPM, Beat Per Minute).
[background technology]
BPM is one of important attribute of music, and the beat number of sign music per minute plays a significant role in application such as audio analysis, individualized music recommendation, music assorting.For example in the application of musical instrument digital interface (MIDI), can speed and an existing track of MIDI SEQ sequencer is synchronous through BPM; In the musical database system, can classify to music and retrieve according to BPM.In the application of dj (DJ), can the BPM value of two kinds of music source be compared, so that seamlessly carry out the transition to another music source from a music source, or the like.
Existing BPM analytical technology is normally selected some snatch of musics at random, utilizes the correlativity extracting cycle information of signal, and the beat number that is converted into per minute is as output.Existing this analysis mode is obviously relevant with selected snatch of music position, the accuracy of the BPM that the very difficult assurance of the snatch of music of selecting is at random analyzed.
[summary of the invention]
The invention provides a kind of method and apparatus of analyzing music BPM, so that improve the accuracy of the music BPM that analyzes.
Concrete technical scheme is following:
A kind of method of analyzing music BPM, this method comprises:
S1, from music to be analyzed, identify the snatch of music of N preset clip types, N is preset positive integer, and said preset clip types comprises refrain, play and lead at least a in singing;
S2, calculate BPM respectively to each snatch of music;
S3, will merge, obtain the BPM of music to be identified to the BPM that each snatch of music calculates.
According to one preferred embodiment of the present invention; In said step S1, also comprise: the refrain part that determines said music to be analyzed; It is main song part that before the starting point of said refrain part first set in the duration scope, and second after the terminal point of said refrain part played part between setting in the duration scope and being.
According to one preferred embodiment of the present invention, the refrain part of differentiating said music to be analyzed specifically comprises:
S11, said music to be analyzed is carried out the branch frame handle;
S12, extract the overtone characteristic of each frame;
S13, utilize the overtone characteristic of each frame to determine to belong to each frame of refrain part.
According to one preferred embodiment of the present invention, said step S13 specifically comprises:
Compare by frame and preset refrain overtone threshold value in order, if all reach refrain overtone threshold value since the overtone characteristic of M1 frame continuous P 1 frame, then determining said M1 frame is the starting point of refrain part; All be lower than refrain overtone threshold value if the M2+1 frame begins the overtone characteristic of continuous P 2 frames, then determining the M2 frame is the terminal point of refrain part, and said M2 is greater than said M1, and said P1 and P2 are preset positive integer; Perhaps,
According to the overtone characteristic each frame is carried out cluster, according to the time sequencing refrain of each frame corresponding row's time class be identified as the refrain part.
According to one preferred embodiment of the present invention, comprise also that between said step S1 and step S2 the snatch of music that identifies is carried out voice eliminates processing, is specially:
For monaural snatch of music, utilize rejection filter that the filtering signals in the voice frequency band range is fallen; For the snatch of music of two-channel, can be with the signals reverse of one of them sound channel in L channel or the R channel, on the signal of another sound channel that is added to, thus the snatch of music behind the voice that is eliminated.
According to one preferred embodiment of the present invention, in said step S2, specifically carry out following steps to each snatch of music:
S21, the autocorrelation function of current snatch of music is carried out peak value searching, the music of adjudicating under the current snatch of music is clapped system;
S22, to the detection a little that pauses of current snatch of music;
S23, according to the time interval and the court verdict of said step S21 that pause between point, calculate the BPM of the beat number of per minute as current snatch of music.
According to one preferred embodiment of the present invention, said step S21 specifically comprises:
A1, the autocorrelation function of current snatch of music is carried out peak value searching, confirm the frame number l that each beat comprises;
A2, calculate the value of autocorrelation function respectively, and judge to be whether the frame number of local maximum satisfies preset requirement, x is preset positive integer at the integral multiple frame place of xl;
If the judged result of the said steps A 2 of A3 is for being, confirm that then current snatch of music belongs to x and claps system.
According to one preferred embodiment of the present invention, the signal energy amplitude that in said step S22, detects current snatch of music is lower than the position of preset amplitude low spot, with detected position as the pause point.
According to one preferred embodiment of the present invention, said step S23 specifically comprises:
According to the little joint number between the time interval between the pause point and pause point, confirm the time span of each trifle; According to the court verdict of said step S21, confirm the beat number that each trifle comprises; The beat number that comprises according to each trifle and the time span of each trifle are calculated the BPM of the beat number of per minute as current snatch of music; Perhaps,
Determine the beat number that each trifle comprises according to the court verdict of said step S21; According to the definite beat number that pauses and comprise between point of little joint number that pauses between point; In the time interval between beat number that utilizing pauses comprises between point and pause point, calculate the BPM of the beat number of per minute as current snatch of music.
According to one preferred embodiment of the present invention, after said step S23, also comprise:
Judge whether the BPM that calculates exceeds preset reasonable value scope, if continue execution in step S24;
If the BPM that S24 calculates is lower than the minimum of said reasonable value scope, on duty with the BPM that calculates then with 2 BPM as the current snatch of music after optimizing; If the BPM that calculates is higher than the mxm. of said reasonable value scope, then with the value of the BPM that calculates divided by 2 BPM as the current snatch of music after optimizing.
According to one preferred embodiment of the present invention, said step S3 specifically comprises:
According to the weights of the snatch of music of each clip types, confirm the weights sum of the snatch of music that the value of each BMP is corresponding respectively, confirm the BPM of the value of the BPM that corresponding weights sum is the highest as said music to be identified; Perhaps,
According to the weights of the snatch of music of each clip types, the BPM of each snatch of music is carried out weighting ask on average, obtain the BPM of said music to be identified.
A kind ofly analyze the device that music per minute beat is counted BPM, this device comprises:
The fragment selected cell is used for identifying from music to be analyzed the snatch of music of N preset clip types, and N is preset positive integer, and said preset clip types comprises refrain, play and lead at least a in singing;
The BPM computing unit is used for calculating BPM respectively to each snatch of music;
The BPM integrated unit is used for the BPM that calculates to each snatch of music is merged, and obtains the BPM of music to be identified.
According to one preferred embodiment of the present invention, this device also comprises:
The type identification unit; Be used to determine the refrain part of said music to be analyzed; Differentiating in the said refrain starting point partly first setting duration scope before is main song part; Differentiate second after the terminal point of said refrain part and set in the duration scope and play part between being, will differentiate the result and offer said fragment selected cell.
According to one preferred embodiment of the present invention, said type identification unit specifically comprises:
Divide the frame subelement, be used for that said music to be analyzed is carried out the branch frame and handle;
The feature extraction subelement is used to extract the overtone characteristic of each frame;
Refrain recognin unit is used to utilize the overtone characteristic of each frame to determine each frame that belongs to the refrain part.
According to one preferred embodiment of the present invention; Said refrain recognin unit compares by frame and preset refrain overtone threshold value in order; If all reach refrain overtone threshold value since the overtone characteristic of M1 frame continuous P 1 frame, then determining said M1 frame is the starting point of refrain part; All be lower than refrain overtone threshold value if the M2+1 frame begins the overtone characteristic of continuous P 2 frames, then determining the M2 frame is the terminal point of refrain part, and said M2 is greater than said M1, and said P1 and P2 are preset positive integer; Perhaps,
According to the overtone characteristic each frame is carried out cluster, according to the time sequencing refrain of each frame corresponding row's time class be identified as the refrain part.
According to one preferred embodiment of the present invention, this device also comprises:
Voice is eliminated the unit, is used for the snatch of music that said fragment selected cell is selected is carried out offering said BPM computing unit after voice is eliminated processing, is specially:
For monaural snatch of music, utilize rejection filter that the filtering signals in the voice frequency band range is fallen; For the snatch of music of two-channel, can be with the signals reverse of one of them sound channel in L channel or the R channel, on the signal of another sound channel that is added to, thus the snatch of music behind the voice that is eliminated.
According to one preferred embodiment of the present invention, said BPM computing unit specifically comprises:
Clap system's judgement subelement, be used for the autocorrelation function of snatch of music is carried out peak value searching, the music under the judgement snatch of music is clapped system;
The pause detection sub-unit is used for the snatch of music detection a little that pauses;
The BPM computation subunit, being used for according to the time interval and the said bat that pause between point is the court verdict of judgement subelement, calculates the BPM of the beat number of per minute as current snatch of music.
According to one preferred embodiment of the present invention, said bat system judgement subelement specifically comprises:
The peak value searching module is used for the autocorrelation function of snatch of music is carried out peak value searching, confirms the frame number l that each beat comprises;
The peak value judge module is used for calculating respectively the value of autocorrelation function at the integral multiple frame place of xl, and judges to be whether the frame number of local maximum satisfies preset requirement, x is preset positive integer;
Bat is a determination module, if the judged result that is used for said peak value judge module is for being to confirm that then snatch of music belongs to x and claps system.
According to one preferred embodiment of the present invention, the signal energy amplitude that said pause detection sub-unit detects current snatch of music is lower than the position of preset amplitude low spot, with detected position as the pause point.
According to one preferred embodiment of the present invention, said BPM computation subunit is confirmed the time span of each trifle specifically according to the little joint number between the time interval between the pause point and pause point; According to said bat is the court verdict of judgement subelement, confirms the beat number that each trifle comprises; The beat number that comprises according to each trifle and the time span of each trifle are calculated the BPM of the beat number of per minute as current snatch of music; Perhaps,
According to said bat is that the court verdict of adjudicating subelement is determined the beat number that each trifle comprises; According to the definite beat number that pauses and comprise between point of little joint number that pauses between point; In the time interval between beat number that utilizing pauses comprises between point and pause point, calculate the BPM of the beat number of per minute as current snatch of music.
According to one preferred embodiment of the present invention, said BPM computing unit also comprises:
BPM optimizes subelement; Be used to judge whether the BPM that said BPM computation subunit calculates exceeds preset reasonable value scope; If; Continue to carry out: if the said BPM that calculates is lower than the minimum of said reasonable value scope, then will the said BPM that calculates on duty with 2 BPM as the current snatch of music after the optimization; If the said BPM that calculates is higher than the mxm. of said reasonable value scope, then with the value of the said BPM that calculates divided by 2 BPM as the current snatch of music after optimizing.
According to one preferred embodiment of the present invention; Said BPM integrated unit is specifically according to the weights of the snatch of music of each clip types; The weights sum of the snatch of music of the value of definite each BMP correspondence is confirmed the BPM of the value of the BPM that corresponding weights sum is the highest as said music to be identified respectively; Perhaps,
According to the weights of the snatch of music of each clip types, the BPM of each snatch of music is carried out weighting ask on average, obtain the BPM of said music to be identified.
Can find out by above technical scheme; The present invention identifies the snatch of music of N preset clip types from snatch of music to be analyzed; Specifically be to have adopted analyzing BPM and have the refrain that embodies than high precision, playing and lead at least a in singing; After calculating BPM to each snatch of music respectively, the BPM of each snatch of music is merged, obtain the BPM of music to be identified.The compare mode of existing random extraction snatch of music of this mode has improved the accuracy that BPM analyzes.
[description of drawings]
The method flow diagram of the analysis music BPM that Fig. 1 provides for the embodiment of the invention one;
The method flow diagram of the BPM of the computational music fragment that Fig. 2 provides for the embodiment of the invention two;
Clap the method flow diagram of system under the judgement snatch of music that Fig. 3 provides for the embodiment of the invention two;
The structure drawing of device of the analysis music BPM that Fig. 4 provides for the embodiment of the invention three;
The structural drawing of the BPM computing unit that Fig. 5 provides for the embodiment of the invention four.
[embodiment]
In order to make the object of the invention, technical scheme and advantage clearer, describe the present invention below in conjunction with accompanying drawing and specific embodiment.
Embodiment one,
The method flow diagram of the analysis music BPM that Fig. 1 provides for the embodiment of the invention one, as shown in Figure 1, this method can may further comprise the steps:
Step 101: from music to be analyzed, identify the snatch of music of N preset clip types, N is preset positive integer, and preset clip types comprises refrain, play and lead at least a in singing.
With the pop music is example, and music can be divided into prelude, main song, refrain usually, play and ending etc., and wherein refrain part timing is the strongest, and dynamics is maximum, is the optimal music fragment of analyzing BPM.Between play part and do not have voice to have only musical accompaniment usually, analyze the effect of BPM and take second place.Main song part timing and dynamics be all not as good as the refrain part, and analytical effect is not as the refrain part and play part, but also can be used as the snatch of music of analysis BPM.
Need to prove, when selecting snatch of music, can only select refrain, play and lead a kind of in singing, for example select more than one refrain fragment.Also can select refrain, play and lead multiple in the song, for example select to play between n1 refrain fragment, n2 fragment and n3 and lead and sing fragment, wherein n1, n2 and n3 are positive integer.
When carrying out the identification of snatch of music, need carry out refrain, play and lead the differentiation of song.Common most pop music for refrain, the order of playing and lead song all is: main song, refrain and play.Therefore, when differentiating, need confirm the starting point and the terminal point of refrain, can think that then it is to play between the second setting duration scope after main song, the terminal point in refrain is that first before the starting point in refrain set in the duration scope.Wherein the first setting duration and the second setting duration are empirical value, and for example the first setting duration can be chosen 20s, and second sets duration can choose 5s to the value between the 10s.
Compare with playing with main song, the difference of refrain part is mainly reflected on the overtone characteristic, can utilize the overtone characteristic that refrain is differentiated, and wherein the overtone characteristic can adopt chroma (Chroma) characteristic to characterize.Particularly, can at first carry out the branch frame to music and handle, extract the overtone characteristic of each frame, the overtone characteristic of each frame is determined the refrain part then.The method of wherein utilizing the overtone characteristic that refrain is differentiated has a lot, can adopt but is not limited to following two kinds:
Mode one, compare by frame and preset refrain overtone threshold value in order, if all reach refrain overtone threshold value since the overtone characteristic of M1 frame continuous P 1 frame, then determining this M1 frame is the refrain starting point; If the overtone characteristic of M2+1 frame continuous P 2 frames after the M1 frame all is lower than refrain overtone threshold value, then determining this M2 frame is the refrain terminal point.Wherein P1 and P2 are preset positive integer.
The mode of mode two, employing feature clustering; According to the overtone characteristic each frame is carried out cluster; Because music is according to prelude, main song, refrain, an order of playing and ending up; Therefore after the cluster, according to the corresponding row of time sequencing refrain institute of each frame time type be identified as the refrain part, for example being positioned at middle cluster is exactly the refrain part.The method of cluster can adopt LBG etc.
Some discrete frames that should be positioned at other parts when cluster, may occur and arrived the refrain part by cluster; This can revise according to the time of each frame equally; For example after carrying out cluster; Find that it is one type that the 13rd frame, the 20th to 40 frame are gathered, can find out that obvious the 13rd frame does not belong to this cluster.
Step 102: the snatch of music that identifies is carried out voice eliminate processing.
Voice is eliminated to refer to and from music signal, is removed people's acoustic intelligence, only keeps musical accompaniment information, and purpose is to get rid of the influence that voice is analyzed BPM.This step mainly is the processing of doing to the snatch of music of main song and refrain.
When carrying out the voice elimination,, can utilize the frequency domain information of snatch of music for monaural music; The common frequency band broad of musical background, people's sonic-frequency band is narrower, can utilize rejection filter; Filtering signals in the voice frequency band range is fallen, thereby reach the purpose that voice is eliminated.
Music for two-channel; Because the amplitude of voice in left and right acoustic channels is identical with phase place; And the frequency spectrum of musical instrument accompaniment is different often in left and right acoustic channels with phase place, therefore, and can be with the signals reverse of one of them sound channel in the left and right acoustic channels; On the another one that the is added to sound channel, thus the signal behind the voice that is eliminated.
Need to prove that this step is in order further to improve the performed step of accuracy that BPM analyzes, and is not the necessary step of the present invention.
Step 103: calculate BPM respectively to each snatch of music.
At this; Can adopt the mode of existing computational music fragment BPM when calculating BPM to each snatch of music, i.e. the autocorrelation function of computational music fragment is because the beat information of music has periodically; Autocorrelation function has periodically equally; In the time interval that the periodic quantity of autocorrelation function is exactly corresponding between two beats, therefore, the periodic quantity of the autocorrelation function through extracting snatch of music is calculated the BPM of this snatch of music.
Preferably, the BPM computing method that the embodiment of the invention provides a kind of " judgement+pause of two bat/triple times detects " specifically will specifically describe in embodiment two.
Step 104: will merge the BPM that obtains music to be identified to the BPM that each snatch of music calculates.
The method that in this step the BPM of each snatch of music is merged can include but not limited to following several kinds:
Mode one, adopt the method for ballot,, confirm the weights sum of the snatch of music that the value of each BPM is corresponding respectively, confirm the BPM of the value of the BPM that corresponding weights sum is the highest as music to be identified according to the weights of the snatch of music of each preset clip types.
Give an example, the snatch of music of suppose refrain, playing and lead song has one respectively, and the weights of the snatch of music of refrain are 0.4; Between the weights of the snatch of music played be 0.3, the weights of the snatch of music of main song are 0.3, the BPM that calculates to the snatch of music of refrain is 100; The BPM that calculates to a snatch of music of playing is 80, and the BPM that calculates to the main snatch of music of singing is 120, and the weights sum of the snatch of music of 100 correspondences is 0.4 so; Maximum, the BPM that confirms music to be identified is 100.For another example, the BPM that calculates to the snatch of music of refrain is 100, and a BPM who calculates to a snatch of music of playing is 80; The BPM that calculates to the main snatch of music of singing is 80; The weights sum of the snatch of music of 80 correspondences is 0.4+0.4 so, maximum, and the BPM that confirms music to be identified is 80.
Mode two, according to weights to the snatch of music of each clip types, the BPM of each snatch of music is carried out weighting asks on average, obtain the BPM of music to be identified.
Can further avoid snatch of music to select the improper low defective of BPM precision that causes effectively through the mode of above-mentioned fusion.
Embodiment two,
Usually musical rhythm is strong, and when the rhythm drumbeat was very tangible, the rhythm of signal periodically also clearly adopted existing BPM computing method to come the BPM of computational music fragment can obtain good effect.But a little less than musical background, even do not have the time of accompaniment, this periodicity is very not obvious, if adopt existing method often to be difficult to obtain BPM value accurately.Adopt the mode of " triple time/two clap judgement+pause and detect " to can be good at addressing this problem in this embodiment.
The method flow diagram of the BPM of the computational music fragment that Fig. 2 provides for the embodiment of the invention two, as shown in Figure 2, this method can may further comprise the steps:
Step 201: the autocorrelation function to snatch of music carries out peak value searching, adjudicates the affiliated music of this snatch of music and claps system.
The concrete implementation method of this step can be referring to Fig. 3, can may further comprise the steps 301 to step 303:
Step 301: the autocorrelation function to snatch of music carries out peak value searching, confirms the frame number l that each beat comprises.
In this step, at first determine the autocorrelation function of snatch of music, the autocorrelation function of asking for music signal is that prior art is not described in detail in this.After determining autocorrelation function, need carry out peak value searching to autocorrelation function in order to confirm the frame number l that each beat comprises, the pairing sampling frame number of peak value is exactly the frame number l that will seek.For example, suppose that the autocorrelation function horizontal ordinate span of snatch of music is 0～1000 frame,, determine horizontal ordinate and get 60 o'clock auto-correlation function value maximums,, confirm that promptly the frame number that each beat comprises is 60 so l is 60 through autocorrelation function is carried out peak value searching.
Because music signal has been carried out break-in operation, inevitably spectral leakage can appear, and so-called spectral leakage is exactly on the Frequency point that energy originally should not occur, to still have energy to occur.Signal is blocked shortly more; Spectral leakage is serious more; In order to reduce the influence that precision produced of spectral leakage to peak value searching, can be further to autocorrelation function stack window function, for example the Gaussian window function of snatch of music; Thereby the spectrum component that the Sidelobe Suppression of utilizing window function leaks out plays the effect of spectral smoothing.
When selecting the Gaussian window function, because the Gaussian window function is by average μ and variances sigma
2Two parameter determining, wherein average μ has determined the position that Gaussian window function maximal value occurs, and variances sigma
2Determined the main lobe width of Gaussian window function, rule of thumb the BPM of most of music is distributed in 80～120, so the average μ of Gaussian window function is chosen as 80～120.Variances sigma
2Choice relation to main lobe width, when selected variances sigma
2The number percent that back BPM value drops in the interval [μ-σ, μ+σ] surpasses 70%, and the number percent that drops on [μ-2 σ, μ+2 σ] surpasses 95%, and therefore, the span of σ can be 40～60.The process of stack window function directly multiplies each other autocorrelation function and window function in fact exactly, is not described in detail in this.
Step 302: calculate the value of autocorrelation function respectively, and judge it is whether the frame number of local maximum satisfies preset requirement at xnl frame place, wherein n be 1,2,3......, x is preset positive integer.
That is to say, calculate the value of autocorrelation function at xl integral multiple frame place respectively, preferably, the bat of common pop music system is at two bat systems or triple time, so x can get 2 or 3.
Step 303:, confirm that then this snatch of music belongs to x and claps system if the judged result in the step 302 is for being.
Give an example; Getting 2 and 3 with x is example; At first calculate the value of autocorrelation function at 2l, 4l, 6l... frame place; Judge in these frames it is whether the frame number of local maximum satisfies preset requirement, wherein preset requirement can be that frame number surpasses preset amount threshold or ratio surpasses preset proportion threshold value etc.; Calculate the value of autocorrelation function, judge in these frames it is whether the frame number of local maximum satisfies preset requirement at 3l, 6l, 9l... frame place.If the frame number that in 2l, 4l, the 6l... frame is local maximum satisfies preset requirement, then rule out this music and judge that belonging to two claps system; If the frame number that in 3l, 6l, the 9l... frame is local maximum satisfies preset requirement, then rule out this snatch of music and judge that belonging to triple time is.
In addition, in practical application since the bat of most music system belong to two clap system or triple time system; In order to reduce operand; Can only calculate the value of autocorrelation function, judge in these frames it is whether the frame number of local maximum satisfies preset requirement at 3l, 6l, 9l... frame place, if; Then rule out this snatch of music and belong to triple time system, belong to two bat systems otherwise all adjudicate this snatch of music.
Continuation is referring to Fig. 2.
Step 202: to the snatch of music detection a little that pauses.
Signal itself to snatch of music in this step carries out the energy amplitude detection; Music for weak accompaniment; All of short duration pause can be arranged between each lyrics; Can all reach minimum by discharge amplitude at this moment, and these appearance positions of pausing point are that periodically the length in cycle is exactly the duration of each lyrics.
In this step, the signal energy amplitude that detects snatch of music is lower than the position of preset amplitude low spot, with these positions as the pause point.The position of preset amplitude low spot can be chosen through empirical value.
Step 203:, calculate the beat number of per minute according to the time interval between the pause point and the court verdict of step 201.
Concrete computing method can may further comprise the steps S1 to S3:
S1, according to the time interval between point and the little joint number between point of pausing of pausing, determine the time span of each trifle.
Between the common pause point is the length of lyrics, and lyrics comprise 4 trifles, can obtain the time span of each trifle divided by the little joint number between the pause point with the time interval between the pause point.
S2, according to the court verdict of step 202, promptly be that system is clapped by triple time system or two, determine the beat number that each trifle comprises.
Belong to triple time system if determine this snatch of music, then each trifle comprises 3 beats, belongs to two bat systems if determine this snatch of music, and then each trifle comprises 2 beats.
The time span of S3, the beat number that comprises according to each trifle and each trifle just can calculate the beat number of per minute, the i.e. BPM of snatch of music.
Need to prove; Above-mentioned steps S1 to step S3 only be wherein a kind of implementation method that the embodiment of the invention provides; Certainly can also adopt other implementations, for example: determine the beat number that each trifle comprises according to the court verdict of step 202, then according to the definite beat number that pauses and comprise between point of little joint number that pauses between point; Last combination again pauses the time interval between point, determines the beat number of per minute.
Preferably, after step S3, can also further be optimized processing, judge promptly whether BPM exceeds preset reasonable value scope the BPM that calculates, if, if be lower than the minimum of zone of reasonableness, then with this BPM on duty with 2 as the BPM after optimizing; If be higher than the mxm. of zone of reasonableness, then with the value of this BPM divided by 2 as the BPM after optimizing.If the result of calculation of step S3 just in time falls within the preset reasonable value scope, then need not to be optimized processing.
For example, supposing that zone of reasonableness is 40～120, if the BPM that calculates at step S3 is 30, then probably is also to exist a beat omission survey to occur between adjacent two beats, therefore should be with BPM on duty with 2, obtain 60 as correct BPM.If the BPM that calculates at step S3 is 180, then probably be to have detected a beat between adjacent two beats more, therefore should obtain 90 as correct BPM with the value of BPM divided by 2.This optimization is to carry out to the ambiguity that music has a beat, with 2 times of optimizations of its called after.
More than be the detailed description that method provided by the present invention is carried out, be described in detail with four pairs of devices provided by the present invention of embodiment below in conjunction with embodiment three.
Embodiment three,
The structure drawing of device of the analysis music BPM that Fig. 4 provides for the embodiment of the invention three, as shown in Figure 4, this device can comprise: fragment selected cell 400, BPM computing unit 410 and BPM integrated unit 420.
Fragment selected cell 400 is used for identifying from music to be analyzed the snatch of music of N preset clip types, and N is preset positive integer, and preset clip types comprises refrain, play and lead at least a in singing.
Wherein refrain part timing is the strongest, and dynamics is maximum, is the optimal music fragment of analyzing BPM.Between play part and do not have voice to have only musical accompaniment usually, analyze the effect of BPM and take second place.Main song part timing and dynamics be all not as good as the refrain part, and analytical effect is not as the refrain part and play part, but also can be used as the snatch of music of analysis BPM.
When selecting snatch of music, can only select refrain, play and lead a kind of in singing, for example select more than one refrain fragment.Also can select refrain, play and lead multiple in the song, for example select to play between n1 refrain fragment, n2 fragment and n3 and lead and sing fragment, wherein n1, n2 and n3 are positive integer.
BPM integrated unit 420 is used for the BPM that calculates to each snatch of music is merged, and obtains the BPM of music to be identified.
Particularly, BPM integrated unit 420 can this employing but is not limited to following two kinds of amalgamation modes:
Mode one, adopt the method for ballot,, confirm the weights sum of the snatch of music that the value of each BMP is corresponding respectively, confirm the BPM of the value of the BPM that corresponding weights sum is the highest as music to be identified according to the weights of the snatch of music of each clip types.
Mode two, according to the weights of the snatch of music of each clip types, the BPM of each snatch of music is carried out weighting asks on average, obtain the BPM of music to be identified.
In order to carry out the identification of snatch of music type; This device can also comprise: type identification unit 430; Be used to determine the refrain part of music to be analyzed; It is the main part of singing that before the differentiation refrain starting point partly first set in the duration scope, and second after the terminal point of differentiation refrain part played part between setting in the duration scope and being, will differentiate the result and offer fragment selected cell 400.
Wherein, type identification unit 430 can specifically comprise: divide frame subelement 431, feature extraction subelement 432 and refrain recognin unit 433.
Dividing 431 pairs of music to be analyzed of frame subelement to carry out the branch frame handles.
Refrain recognin unit 433 utilizes the overtone characteristic of each frame to determine to belong to each frame of refrain part.Particularly, can adopt but be not limited to following dual mode:
Mode one, can be in order compare by frame and preset refrain overtone threshold value, if all reach refrain overtone threshold value since the overtone characteristic of M1 frame continuous P 1 frame, then determining the M1 frame is the starting point of refrain part; All be lower than refrain overtone threshold value if the M2+1 frame begins the overtone characteristic of continuous P 2 frames, then determining the M2 frame is the terminal point of refrain part, and M2 is greater than M1, and P1 and P2 are preset positive integer.
Mode two, each frame is carried out cluster according to the overtone characteristic, according to the time sequencing refrain of each frame corresponding row's time class be identified as the refrain part.
In order further to eliminate the influence that voice is analyzed BPM; Improve precision; This device can also comprise: voice is eliminated unit 440; The snatch of music that is used for that fragment selected cell 400 is selected carries out offering BPM computing unit 410 after voice is eliminated processing, mainly is the processing of doing to the snatch of music of main song and refrain.Be specially:, utilize rejection filter that the filtering signals in the voice frequency band range is fallen for monaural snatch of music; For the snatch of music of two-channel, can be with the signals reverse of one of them sound channel in L channel or the R channel, on the signal of another sound channel that is added to, thus the snatch of music behind the voice that is eliminated.
Embodiment four,
The BPM computing unit can adopt existing BPM account form to come the BPM of computational music fragment, but for a little less than the musical background, even do not have the time of accompaniment, if adopt existing mode often to be difficult to obtain BPM value accurately.Adopt the mode of " triple time/two clap judgement+pause and detect " to can be good at addressing this problem in this embodiment.The structural drawing of the BPM computing unit that Fig. 5 provides for the embodiment of the invention four, as shown in Figure 5, this BPM computing unit specifically comprises: clap system judgement subelement 500, pause detection sub-unit 510 and BPM computation subunit 520.
The autocorrelation function of clapping 500 pairs of snatch of musics of system's judgement subelement carries out peak value searching, and the music under the judgement snatch of music is clapped system.The autocorrelation function of wherein asking for music signal is that prior art is not described in detail in this.
This bat system judgement subelement 500 can specifically comprise: peak value searching module 501, peak value judge module 502 and bat are determination module 503.
The autocorrelation function of 501 pairs of snatch of musics of peak value searching module carries out peak value searching, confirms the frame number l that each beat comprises.Particularly, after determining autocorrelation function, need carry out peak value searching to autocorrelation function in order to confirm the frame number l that each beat comprises, the pairing sampling frame number of peak value is exactly the frame number l that will seek.For example, suppose that the autocorrelation function horizontal ordinate span of snatch of music is 0～1000 frame,, determine horizontal ordinate and get 60 o'clock auto-correlation function value maximums,, confirm that promptly the frame number that each beat comprises is 60 so l is 60 through autocorrelation function is carried out peak value searching.
Because music signal has been carried out break-in operation, inevitably spectral leakage can appear, and so-called spectral leakage is exactly on the Frequency point that energy originally should not occur, to still have energy to occur.Signal is blocked shortly more; Spectral leakage is serious more; In order to reduce the influence that precision produced of spectral leakage to peak value searching, can be further to autocorrelation function stack window function, for example the Gaussian window function of snatch of music; Thereby the spectrum component that the Sidelobe Suppression of utilizing window function leaks out plays the effect of spectral smoothing.
When selecting the Gaussian window function, because the Gaussian window function is by average μ and variances sigma
2Two parameter determining, wherein average μ has determined the position that Gaussian window function maximal value occurs, and variances sigma
2Determined the main lobe width of Gaussian window function, rule of thumb the BPM of most of music is distributed in 80～120, so the average μ of Gaussian window function is chosen as 80～120.Variances sigma
2Choice relation to main lobe width, when selected variances sigma
2The number percent that back BPM value drops in the interval [μ-σ, μ+σ] surpasses 70%, and the number percent that drops on [μ-2 σ, μ+2 σ] surpasses 95%, and therefore, the span of σ can be 40～60.The process of stack window function directly multiplies each other autocorrelation function and window function in fact exactly, is not described in detail in this.
Peak value judge module 502 calculates the value of autocorrelation function at the integral multiple frame place of xl respectively, and judges to be whether the frame number of local maximum satisfies preset requirement, x is preset positive integer.
Preferably, the bat of common pop music system is at two bat systems or triple time, so x can get 2 or 3.
If bat be the judged result of determination module 503 peak value judge modules for being, confirm that then snatch of music belongs to x and claps system.
Give an example; Getting 2 and 3 with x is example; At first calculate the value of autocorrelation function at 2l, 4l, 6l... frame place; Judge in these frames it is whether the frame number of local maximum satisfies preset requirement, wherein preset requirement can be that frame number surpasses preset amount threshold or ratio surpasses preset proportion threshold value etc.; Calculate the value of autocorrelation function, judge in these frames it is whether the frame number of local maximum satisfies preset requirement at 3l, 6l, 9l... frame place.If the frame number that in 2l, 4l, the 6l... frame is local maximum satisfies preset requirement, then rule out this music and judge that belonging to two claps system; If the frame number that in 3l, 6l, the 9l... frame is local maximum satisfies preset requirement, then rule out this snatch of music and judge that belonging to triple time is.
In addition, in practical application since the bat of most music system belong to two clap system or triple time system; In order to reduce operand; Can only calculate the value of autocorrelation function, judge in these frames it is whether the frame number of local maximum satisfies preset requirement at 3l, 6l, 9l... frame place, if; Then rule out this snatch of music and belong to triple time system, belong to two bat systems otherwise all adjudicate this snatch of music.
The detection a little that pauses of 510 pairs of snatch of musics of pause detection sub-unit.Particularly, the signal energy amplitude that detects current snatch of music is lower than the position of preset amplitude low spot, with detected position as the pause point.
Particularly, BPM computation subunit 520 can adopt but be not limited to following dual mode:
Mode one, according to the time interval between point and the little joint number between point of pausing of pausing, confirm the time span of each trifle; According to the court verdict of clapping system's judgement subelement 500, confirm the beat number that each trifle comprises; The beat number that comprises according to each trifle and the time span of each trifle are calculated the BPM of the beat number of per minute as current snatch of music.
Mode two, determine the beat number that each trifle comprises according to the court verdict of clapping system's judgement subelement 500; According to the definite beat number that pauses and comprise between point of little joint number that pauses between point; In the time interval between beat number that utilizing pauses comprises between point and pause point, calculate the BPM of the beat number of per minute as current snatch of music.
In addition; The BPM computing unit can also comprise: BPM optimizes subelement 530; Be used to judge whether the BPM that BPM computation subunit 520 calculates exceeds preset reasonable value scope; If continue to carry out: if the BPM that calculates is lower than the minimum of reasonable value scope, on duty with the BPM that calculates then with 2 BPM as the current snatch of music after the optimization; If the BPM that calculates is higher than the mxm. of reasonable value scope, then with the value of the BPM that calculates divided by 2 BPM as the current snatch of music after optimizing.
Can find out that by above description method and apparatus provided by the invention possesses following advantage:
1) the present invention identifies the snatch of music of N preset clip types from snatch of music to be analyzed; Specifically be to have adopted analyzing BPM and have the refrain that embodies than high precision, playing and lead at least a in singing; After calculating BPM to each snatch of music respectively; The BPM of each snatch of music is merged, obtain the BPM of music to be identified.The compare mode of existing random extraction snatch of music of this mode has improved the accuracy that BPM analyzes.
2) the present invention has introduced in the process of analyzing BPM the life of snatch of music has been separated, and has alleviated life to the influence that BPM analyzes, and has further improved the accuracy that BPM analyzes.
3) the present invention also further provides the mode of a kind of " triple time/two clap judgement+pause and detect " to come the BPM value of computational music fragment; Thereby the snatch of music to weak accompaniment also can calculate the BPM value more exactly, thereby has further improved the accuracy that BPM analyzes.
The above is merely preferred embodiment of the present invention, and is in order to restriction the present invention, not all within spirit of the present invention and principle, any modification of being made, is equal to replacement, improvement etc., all should be included within the scope that the present invention protects.
Claims (22)
1. analyze the method that music per minute beat is counted BPM for one kind, it is characterized in that this method comprises:
S1, from music to be analyzed, identify the snatch of music of N preset clip types, N is preset positive integer, and said preset clip types comprises refrain, play and lead at least a in singing;
S2, calculate BPM respectively to each snatch of music;
S3, will merge, obtain the BPM of music to be identified to the BPM that each snatch of music calculates.
2. method according to claim 1; It is characterized in that; In said step S1, also comprise: the refrain part that determines said music to be analyzed; It is main song part that before the starting point of said refrain part first set in the duration scope, and second after the terminal point of said refrain part played part between setting in the duration scope and being.
3. method according to claim 2 is characterized in that, the refrain part of differentiating said music to be analyzed specifically comprises:
S11, said music to be analyzed is carried out the branch frame handle;
S12, extract the overtone characteristic of each frame;
S13, utilize the overtone characteristic of each frame to determine to belong to each frame of refrain part.
4. method according to claim 3 is characterized in that, said step S13 specifically comprises:
Compare by frame and preset refrain overtone threshold value in order, if all reach refrain overtone threshold value since the overtone characteristic of M1 frame continuous P 1 frame, then determining said M1 frame is the starting point of refrain part; All be lower than refrain overtone threshold value if the M2+1 frame begins the overtone characteristic of continuous P 2 frames, then determining the M2 frame is the terminal point of refrain part, and said M2 is greater than said M1, and said P1 and P2 are preset positive integer; Perhaps,
According to the overtone characteristic each frame is carried out cluster, according to the time sequencing refrain of each frame corresponding row's time class be identified as the refrain part.
5. method according to claim 1 is characterized in that, comprises also that between said step S1 and step S2 the snatch of music that identifies is carried out voice eliminates processing, is specially:
For monaural snatch of music, utilize rejection filter that the filtering signals in the voice frequency band range is fallen; For the snatch of music of two-channel, can be with the signals reverse of one of them sound channel in L channel or the R channel, on the signal of another sound channel that is added to, thus the snatch of music behind the voice that is eliminated.
6. method according to claim 1 is characterized in that, in said step S2, specifically carries out following steps to each snatch of music:
S21, the autocorrelation function of current snatch of music is carried out peak value searching, the music of adjudicating under the current snatch of music is clapped system;
S22, to the detection a little that pauses of current snatch of music;
S23, according to the time interval and the court verdict of said step S21 that pause between point, calculate the BPM of the beat number of per minute as current snatch of music.
7. method according to claim 6 is characterized in that, said step S21 specifically comprises:
A1, the autocorrelation function of current snatch of music is carried out peak value searching, confirm the frame number l that each beat comprises;
A2, calculate the value of autocorrelation function respectively, and judge to be whether the frame number of local maximum satisfies preset requirement, x is preset positive integer at the integral multiple frame place of xl;
If the judged result of the said steps A 2 of A3 is for being, confirm that then current snatch of music belongs to x and claps system.
8. method according to claim 6 is characterized in that, the signal energy amplitude that in said step S22, detects current snatch of music is lower than the position of preset amplitude low spot, with detected position as the pause point.
9. method according to claim 6 is characterized in that, said step S23 specifically comprises:
According to the little joint number between the time interval between the pause point and pause point, confirm the time span of each trifle; According to the court verdict of said step S21, confirm the beat number that each trifle comprises; The beat number that comprises according to each trifle and the time span of each trifle are calculated the BPM of the beat number of per minute as current snatch of music; Perhaps,
Determine the beat number that each trifle comprises according to the court verdict of said step S21; According to the definite beat number that pauses and comprise between point of little joint number that pauses between point; In the time interval between beat number that utilizing pauses comprises between point and pause point, calculate the BPM of the beat number of per minute as current snatch of music.
10. method according to claim 6 is characterized in that, after said step S23, also comprises:
Judge whether the BPM that calculates exceeds preset reasonable value scope, if continue execution in step S24;
If the BPM that S24 calculates is lower than the minimum of said reasonable value scope, on duty with the BPM that calculates then with 2 BPM as the current snatch of music after optimizing; If the BPM that calculates is higher than the mxm. of said reasonable value scope, then with the value of the BPM that calculates divided by 2 BPM as the current snatch of music after optimizing.
11. method according to claim 1 is characterized in that, said step S3 specifically comprises:
According to the weights of the snatch of music of each clip types, confirm the weights sum of the snatch of music that the value of each BMP is corresponding respectively, confirm the BPM of the value of the BPM that corresponding weights sum is the highest as said music to be identified; Perhaps,
According to the weights of the snatch of music of each clip types, the BPM of each snatch of music is carried out weighting ask on average, obtain the BPM of said music to be identified.
12. analyze the device that music per minute beat is counted BPM for one kind, it is characterized in that this device comprises:
The fragment selected cell is used for identifying from music to be analyzed the snatch of music of N preset clip types, and N is preset positive integer, and said preset clip types comprises refrain, play and lead at least a in singing;
The BPM computing unit is used for calculating BPM respectively to each snatch of music;
The BPM integrated unit is used for the BPM that calculates to each snatch of music is merged, and obtains the BPM of music to be identified.
13. device according to claim 12 is characterized in that, this device also comprises:
The type identification unit; Be used to determine the refrain part of said music to be analyzed; Differentiating in the said refrain starting point partly first setting duration scope before is main song part; Differentiate second after the terminal point of said refrain part and set in the duration scope and play part between being, will differentiate the result and offer said fragment selected cell.
14. device according to claim 13 is characterized in that, said type identification unit specifically comprises:
Divide the frame subelement, be used for that said music to be analyzed is carried out the branch frame and handle;
The feature extraction subelement is used to extract the overtone characteristic of each frame;
Refrain recognin unit is used to utilize the overtone characteristic of each frame to determine each frame that belongs to the refrain part.
15. device according to claim 14; It is characterized in that; Said refrain recognin unit compares by frame and preset refrain overtone threshold value in order; If all reach refrain overtone threshold value since the overtone characteristic of M1 frame continuous P 1 frame, then determining said M1 frame is the starting point of refrain part; All be lower than refrain overtone threshold value if the M2+1 frame begins the overtone characteristic of continuous P 2 frames, then determining the M2 frame is the terminal point of refrain part, and said M2 is greater than said M1, and said P1 and P2 are preset positive integer; Perhaps,
According to the overtone characteristic each frame is carried out cluster, according to the time sequencing refrain of each frame corresponding row's time class be identified as the refrain part.
16. device according to claim 12 is characterized in that, this device also comprises:
Voice is eliminated the unit, is used for the snatch of music that said fragment selected cell is selected is carried out offering said BPM computing unit after voice is eliminated processing, is specially:
For monaural snatch of music, utilize rejection filter that the filtering signals in the voice frequency band range is fallen; For the snatch of music of two-channel, can be with the signals reverse of one of them sound channel in L channel or the R channel, on the signal of another sound channel that is added to, thus the snatch of music behind the voice that is eliminated.
17. device according to claim 12 is characterized in that, said BPM computing unit specifically comprises:
Clap system's judgement subelement, be used for the autocorrelation function of snatch of music is carried out peak value searching, the music under the judgement snatch of music is clapped system;
The pause detection sub-unit is used for the snatch of music detection a little that pauses;
The BPM computation subunit, being used for according to the time interval and the said bat that pause between point is the court verdict of judgement subelement, calculates the BPM of the beat number of per minute as current snatch of music.
18. device according to claim 17 is characterized in that, said bat is that the judgement subelement specifically comprises:
The peak value searching module is used for the autocorrelation function of snatch of music is carried out peak value searching, confirms the frame number l that each beat comprises;
The peak value judge module is used for calculating respectively the value of autocorrelation function at the integral multiple frame place of xl, and judges to be whether the frame number of local maximum satisfies preset requirement, x is preset positive integer;
Bat is a determination module, if the judged result that is used for said peak value judge module is for being to confirm that then snatch of music belongs to x and claps system.
19. device according to claim 17 is characterized in that, the signal energy amplitude that said pause detection sub-unit detects current snatch of music is lower than the position of preset amplitude low spot, with detected position as the pause point.
20. device according to claim 17 is characterized in that, said BPM computation subunit is confirmed the time span of each trifle specifically according to the little joint number between the time interval between the pause point and pause point; According to said bat is the court verdict of judgement subelement, confirms the beat number that each trifle comprises; The beat number that comprises according to each trifle and the time span of each trifle are calculated the BPM of the beat number of per minute as current snatch of music; Perhaps,
According to said bat is that the court verdict of adjudicating subelement is determined the beat number that each trifle comprises; According to the definite beat number that pauses and comprise between point of little joint number that pauses between point; In the time interval between beat number that utilizing pauses comprises between point and pause point, calculate the BPM of the beat number of per minute as current snatch of music.
21. device according to claim 17 is characterized in that, said BPM computing unit also comprises:
BPM optimizes subelement; Be used to judge whether the BPM that said BPM computation subunit calculates exceeds preset reasonable value scope; If; Continue to carry out: if the said BPM that calculates is lower than the minimum of said reasonable value scope, then will the said BPM that calculates on duty with 2 BPM as the current snatch of music after the optimization; If the said BPM that calculates is higher than the mxm. of said reasonable value scope, then with the value of the said BPM that calculates divided by 2 BPM as the current snatch of music after optimizing.
22. device according to claim 12; It is characterized in that; Said BPM integrated unit is specifically according to the weights of the snatch of music of each clip types; The weights sum of the snatch of music of the value of definite each BMP correspondence is confirmed the BPM of the value of the BPM that corresponding weights sum is the highest as said music to be identified respectively; Perhaps,
According to the weights of the snatch of music of each clip types, the BPM of each snatch of music is carried out weighting ask on average, obtain the BPM of said music to be identified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110415218.XA CN102568454B (en) | 2011-12-13 | 2011-12-13 | A kind of method and apparatus analyzing music BPM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110415218.XA CN102568454B (en) | 2011-12-13 | 2011-12-13 | A kind of method and apparatus analyzing music BPM |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102568454A true CN102568454A (en) | 2012-07-11 |
| CN102568454B CN102568454B (en) | 2015-08-05 |
Family
ID=46413714
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110415218.XA Active CN102568454B (en) | 2011-12-13 | 2011-12-13 | A kind of method and apparatus analyzing music BPM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102568454B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104091591A (en) * | 2013-10-15 | 2014-10-08 | 腾讯科技（深圳）有限公司 | Audio processing method and device |
| CN104091595A (en) * | 2013-10-15 | 2014-10-08 | 腾讯科技（深圳）有限公司 | Audio processing method and device |
| CN105280196A (en) * | 2015-11-19 | 2016-01-27 | 科大讯飞股份有限公司 | Method and system for detecting chorus |
| CN108172210A (en) * | 2018-02-01 | 2018-06-15 | 福州大学 | A kind of performance harmony generation method based on song rhythm |
| CN108389568A (en) * | 2018-05-07 | 2018-08-10 | 深圳乐华高科实业有限公司 | A kind of karaoke OK system and its application method of the automatic voice that disappears |
| CN109545249A (en) * | 2018-11-23 | 2019-03-29 | 广州酷狗计算机科技有限公司 | A kind of method and device handling music file |
| CN111508457A (en) * | 2020-04-14 | 2020-08-07 | 上海影卓信息科技有限公司 | Music beat detection method and system |
| CN112562747A (en) * | 2015-06-22 | 2021-03-26 | 玛诗塔乐斯有限公司 | Method for determining start and its position in digital signal, digital signal processor and audio system |
| CN113066512A (en) * | 2021-03-24 | 2021-07-02 | 平安科技（深圳）有限公司 | Buddhism music recognition method, device, equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002116754A (en) * | 2000-07-31 | 2002-04-19 | Matsushita Electric Ind Co Ltd | Tempo extraction device, tempo extraction method, tempo extraction program and recording medium |
| CN1941073A (en) * | 2005-09-26 | 2007-04-04 | 三星电子株式会社 | Apparatus and method of canceling vocal component in an audio signal |
| CN101512636A (en) * | 2006-09-11 | 2009-08-19 | 惠普开发有限公司 | Computational music-tempo estimation |
| US20100011939A1 (en) * | 2008-07-16 | 2010-01-21 | Honda Motor Co., Ltd. | Robot |
| CN101640057A (en) * | 2009-05-31 | 2010-02-03 | 北京中星微电子有限公司 | Audio and video matching method and device therefor |
-
2011
- 2011-12-13 CN CN201110415218.XA patent/CN102568454B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002116754A (en) * | 2000-07-31 | 2002-04-19 | Matsushita Electric Ind Co Ltd | Tempo extraction device, tempo extraction method, tempo extraction program and recording medium |
| CN1941073A (en) * | 2005-09-26 | 2007-04-04 | 三星电子株式会社 | Apparatus and method of canceling vocal component in an audio signal |
| CN101512636A (en) * | 2006-09-11 | 2009-08-19 | 惠普开发有限公司 | Computational music-tempo estimation |
| US20100011939A1 (en) * | 2008-07-16 | 2010-01-21 | Honda Motor Co., Ltd. | Robot |
| CN101640057A (en) * | 2009-05-31 | 2010-02-03 | 北京中星微电子有限公司 | Audio and video matching method and device therefor |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104091595B (en) * | 2013-10-15 | 2017-02-15 | 广州酷狗计算机科技有限公司 | Audio processing method and device |
| CN104091595A (en) * | 2013-10-15 | 2014-10-08 | 腾讯科技（深圳）有限公司 | Audio processing method and device |
| CN104091591A (en) * | 2013-10-15 | 2014-10-08 | 腾讯科技（深圳）有限公司 | Audio processing method and device |
| CN104091591B (en) * | 2013-10-15 | 2016-01-27 | 腾讯科技（深圳）有限公司 | A kind of audio-frequency processing method and device |
| CN112562747A (en) * | 2015-06-22 | 2021-03-26 | 玛诗塔乐斯有限公司 | Method for determining start and its position in digital signal, digital signal processor and audio system |
| CN105280196B (en) * | 2015-11-19 | 2019-03-19 | 科大讯飞股份有限公司 | Refrain detection method and system |
| CN105280196A (en) * | 2015-11-19 | 2016-01-27 | 科大讯飞股份有限公司 | Method and system for detecting chorus |
| CN108172210A (en) * | 2018-02-01 | 2018-06-15 | 福州大学 | A kind of performance harmony generation method based on song rhythm |
| CN108172210B (en) * | 2018-02-01 | 2021-03-02 | 福州大学 | Singing harmony generation method based on singing voice rhythm |
| CN108389568A (en) * | 2018-05-07 | 2018-08-10 | 深圳乐华高科实业有限公司 | A kind of karaoke OK system and its application method of the automatic voice that disappears |
| CN108389568B (en) * | 2018-05-07 | 2023-11-10 | 深圳乐华高科实业有限公司 | Karaoke system capable of automatically eliminating voice |
| CN109545249A (en) * | 2018-11-23 | 2019-03-29 | 广州酷狗计算机科技有限公司 | A kind of method and device handling music file |
| CN109545249B (en) * | 2018-11-23 | 2020-11-03 | 广州酷狗计算机科技有限公司 | Method and device for processing music file |
| CN111508457A (en) * | 2020-04-14 | 2020-08-07 | 上海影卓信息科技有限公司 | Music beat detection method and system |
| CN113066512A (en) * | 2021-03-24 | 2021-07-02 | 平安科技（深圳）有限公司 | Buddhism music recognition method, device, equipment and storage medium |
| CN113066512B (en) * | 2021-03-24 | 2024-02-23 | 平安科技（深圳）有限公司 | Buddhism music identification method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102568454B (en) | 2015-08-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102568454A (en) | Method and device for analyzing music BPM (Beat Per Minutes) | |
| US6518492B2 (en) | System and method of BPM determination | |
| Degara et al. | Reliability-informed beat tracking of musical signals | |
| EP2816550A1 (en) | Audio signal analysis | |
| Dressler | An Auditory Streaming Approach for Melody Extraction from Polyphonic Music. | |
| Mauch et al. | Timbre and Melody Features for the Recognition of Vocal Activity and Instrumental Solos in Polyphonic Music. | |
| Chen et al. | Electric Guitar Playing Technique Detection in Real-World Recording Based on F0 Sequence Pattern Recognition. | |
| CN102543052A (en) | Method and device for analyzing musical BPM | |
| Zhou et al. | Music onset detection based on resonator time frequency image | |
| CN109712600B (en) | Beat identification method and device | |
| Uhle et al. | Estimation of tempo, micro time and time signature from percussive music | |
| Manilow et al. | Predicting algorithm efficacy for adaptive multi-cue source separation | |
| Zhang et al. | Main melody extraction from polyphonic music based on modified Euclidean algorithm | |
| Zhang et al. | Melody extraction from polyphonic music using particle filter and dynamic programming | |
| US7276656B2 (en) | Method for music analysis | |
| CN101093660B (en) | Musical note syncopation method and device based on detection of double peak values | |
| Wu et al. | A supervised learning method for tempo estimation of musical audio | |
| Marolt | On finding melodic lines in audio recordings | |
| Marolt | Gaussian Mixture Models For Extraction Of Melodic Lines From Audio Recordings. | |
| De Mulder et al. | An auditory model based transcriber of vocal queries | |
| Wu | Musical tempo octave error reducing based on the statistics of tempogram | |
| Li et al. | Pitch detection in polyphonic music using instrument tone models | |
| De Mulder et al. | Recent improvements of an auditory model based front-end for the transcription of vocal queries | |
| Özaslan et al. | Attack based articulation analysis of nylon string guitar | |
| Paiva et al. | An auditory model based approach for melody detection in polyphonic musical recordings |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |