CN108596049A - A kind of singing sorting technique based on brain wave - Google Patents
A kind of singing sorting technique based on brain wave Download PDFInfo
- Publication number
- CN108596049A CN108596049A CN201810296369.XA CN201810296369A CN108596049A CN 108596049 A CN108596049 A CN 108596049A CN 201810296369 A CN201810296369 A CN 201810296369A CN 108596049 A CN108596049 A CN 108596049A
- Authority
- CN
- China
- Prior art keywords
- singing
- brain wave
- signal
- rhythm
- feature
- 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.)
- Pending
Links
- 210000004556 brain Anatomy 0.000 title claims abstract description 62
- 238000000034 method Methods 0.000 title claims abstract description 22
- 230000033764 rhythmic process Effects 0.000 claims abstract description 43
- 238000012549 training Methods 0.000 claims abstract description 27
- 238000012545 processing Methods 0.000 claims abstract description 22
- 238000012360 testing method Methods 0.000 claims abstract description 16
- 238000011156 evaluation Methods 0.000 claims abstract description 15
- 239000000284 extract Substances 0.000 claims abstract description 9
- 238000001228 spectrum Methods 0.000 claims description 33
- 238000000605 extraction Methods 0.000 claims description 19
- 239000013598 vector Substances 0.000 claims description 15
- 238000013135 deep learning Methods 0.000 claims description 11
- 238000013136 deep learning model Methods 0.000 claims description 10
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000011551 log transformation method Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 238000012958 reprocessing Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000013480 data collection Methods 0.000 claims description 3
- 238000012417 linear regression Methods 0.000 claims description 3
- 239000000463 material Substances 0.000 claims description 3
- 230000000306 recurrent effect Effects 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 2
- 238000010801 machine learning Methods 0.000 abstract description 7
- 238000001514 detection method Methods 0.000 abstract description 4
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000011161 development Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The singing sorting technique based on brain wave that the invention discloses a kind of.It includes training stage and evaluation and test stage, and the training stage refers to:With eeg signal, the rhythm of singing, the Feature Selection Model of rhythm, style are trained using depth learning technology, feature is extracted according to Feature Selection Model, final training obtains singing disaggregated model;The evaluation and test stage refers to:Brain wave sensor obtains eeg signal, extracts the rhythm of singing, the feature of rhythm, style, is classified to singing using singing disaggregated model.The beneficial effects of the invention are as follows:Brain wave detection, signal processing technology and machine learning method are run, realizes and carries out accurate, automatically evaluation and test to singing, reached the automatic classification purpose to singing.
Description
Technical field
The present invention relates to machine learning correlative technology fields, refer in particular to a kind of singing sorting technique based on brain wave.
Background technology
As the improvement of people's living standards, people grow to even greater heights for the enthusiasm of singing.How effectively and rapidly research
Automatic classification of singing has great significance, and can not only significantly decrease artificial workload, and can improve and listen singing
Efficiency.Artificial evaluation and test needs specially artificial, while needing manually repeatedly to read to provide the classification of profession.With the modern times
The development of science and technology, the development of brain wave technology enter fast traffic lane, in new field using more and more.
Invention content
The present invention be in order to overcome the above deficiencies in the prior art, provide it is a kind of can classify automatically based on
The singing sorting technique of brain wave.
To achieve the goals above, the present invention uses following technical scheme:
A kind of singing sorting technique based on brain wave, including training stage and evaluation and test stage, the training stage refer to
Be:With eeg signal, the rhythm of singing, the Feature Selection Model of rhythm, style are trained using deep learning technology,
Feature is extracted according to Feature Selection Model, final training obtains singing disaggregated model;The evaluation and test stage refers to:Brain wave
Sensor obtain eeg signal, extract the rhythm of singing, the feature of rhythm, style, using singing disaggregated model to sing into
Row classification.
The present invention proposes the automatic classification method of singing, by acquiring the eeg signal of user, by depth
Algorithm extraction is practised in relation to the rhythm sung, the feature of rhythm, style, classifies on singing disaggregated model, obtains final
It sings and classifies.The present invention runs brain wave detection, signal processing technology and machine learning method, realizes and carries out standard to singing
Really, automatically evaluation and test, has reached the automatic classification purpose to singing.
Preferably, the training stage, steps are as follows:
(1) data collection establishes eeg signal language material and singing corpus;Brain wave sensor detects human brain, every time
Brain wave original signal will be collected and be converted to brain wave digital signal;
(2) it utilizes signal processing algorithm to handle brain wave digital signal, obtains spectrum signal;Detailed process is as follows:By brain
Electric wave digital signal carries out segment processing, and frequency-region signal is obtained using Fast Fourier Transform (FFT) to each segment signal, believes frequency domain
Number extraction power spectrum, finally to power spectrum carry out Log transformation, obtain Log power spectrum, i.e. spectrum signal;
(3) spectrum signal is used, the characteristic model of the rhythm, rhythm, style sung using deep learning model training, together
Shi Liyong deep learnings model extracts feature to spectrum signal;
(4) characteristic model obtained using training is to the rhythm of the brain wave digital signal extraction in relation to singing, rhythm, style
Feature, and final singing disaggregated model is trained according to linear regression algorithm.
Preferably, in step (2), segment processing mode is specially:It is 1s per segment length, every section is disposed, to
After move 0.5s, had between adjacent two sections 0.5s overlapping, reprocessing, until being disposed;Fast Fourier Transform (FFT) be from
Dissipate the fast algorithm of Fourier transformation.
Preferably, in step (3), the deep learning model includes deep neural network, convolutional neural networks
And Recognition with Recurrent Neural Network;Deep learning is the branch of machine learning, is that one kind attempting use comprising labyrinth or by multiple non-
Multiple process layers that linear transformation is constituted carry out data the algorithm of higher level of abstraction, and the feature of extraction includes the rhythm sung, section
It plays and style, these are characterized in that deep learning algorithm is automatically learned, later on the corresponding time slice of eeg signal
It is labeled.
Preferably, in step (4), singing disaggregated model is that logistic returns more classification, and logistic returns more
Principle of classification is as follows:The classification of Hypothetical classification has k, and such Y is expressed as one-hot forms, and specifically it corresponds to the label of class
It is 1, remaining category label is 0;Z is obtained using following two formula, Z is a vector, and the maximum value in Z is knot of classifying
Fruit, formula are specific as follows:(i) Y=Ax+b;(ii) Z=1/ (1+exp (- Y));Above-mentioned formula is all vector form, wherein Z is
Vector indicates that final classification, A and b are singing disaggregated model parameters, and A is matrix, and b is scalar, and x is the feature vector of extraction, Y
It is vector, indicates results of intermediate calculations.
Preferably, the evaluation and test stage etch is as follows:
(a) brain wave sensor detects human brain, will collect brain wave original signal every time and be converted to brain wave number
Signal;
(b) utilize signal processing algorithm handle brain wave digital signal, obtain spectrum signal, by brain wave digital signal into
Row segment processing obtains frequency-region signal to each segment signal using Fast Fourier Transform (FFT), extracts power spectrum to frequency-region signal, most
Log transformation is carried out to power spectrum afterwards;
(c) the deep learning model obtained according to training, the spy of the rhythm, rhythm, style sung to spectrum signal extraction
Sign;
(d) the singing disaggregated model obtained using training, and classified to singing according to the feature of extraction.
Preferably, in step (b), segment processing mode is specially:It is 1s per segment length, every section is disposed, to
After move 0.5s, had between adjacent two sections 0.5s overlapping, reprocessing, until being disposed;Fast Fourier Transform (FFT) be from
Dissipate the fast algorithm of Fourier transformation.
The beneficial effects of the invention are as follows:Brain wave detection, signal processing technology and machine learning method are run, is realized pair
It sings and carries out accurate, automatically evaluation and test, reached the automatic classification purpose to singing.
Specific implementation mode
The present invention will be further described With reference to embodiment.
A kind of singing sorting technique based on brain wave, including training stage and evaluation and test stage, the training stage refer to
Be:With eeg signal, the rhythm of singing, the Feature Selection Model of rhythm, style are trained using deep learning technology,
Feature is extracted according to Feature Selection Model, final training obtains singing disaggregated model;The evaluation and test stage refers to:Brain wave
Sensor obtain eeg signal, extract the rhythm of singing, the feature of rhythm, style, using singing disaggregated model to sing into
Row classification.
Wherein:Training stage, steps are as follows:
(1) data collection establishes eeg signal language material and singing corpus;Brain wave sensor detects human brain, every time
Brain wave original signal will be collected and be converted to brain wave digital signal;
(2) it utilizes signal processing algorithm to handle brain wave digital signal, obtains spectrum signal;Detailed process is as follows:By brain
Electric wave digital signal carries out segment processing, and frequency-region signal is obtained using Fast Fourier Transform (FFT) to each segment signal, believes frequency domain
Number extraction power spectrum, finally to power spectrum carry out Log transformation, obtain Log power spectrum, i.e. spectrum signal;Segment processing mode has
Body is:It is 1s per segment length, every section is disposed, and moves backward 0.5s, and 0.5s is had between adjacent two sections and is overlapped, at repetition
Reason, until being disposed;Fast Fourier Transform (FFT) is the fast algorithm of discrete Fourier transform;
(3) spectrum signal is used, the characteristic model of the rhythm, rhythm, style sung using deep learning model training, together
Shi Liyong deep learnings model extracts feature to spectrum signal;Deep learning model includes deep neural network, convolutional Neural net
Network and Recognition with Recurrent Neural Network;Deep learning is the branch of machine learning, is that one kind attempting use comprising labyrinth or by multiple
Multiple process layers that nonlinear transformation is constituted carry out data the algorithm of higher level of abstraction, the feature of extraction include the rhythm sung,
Rhythm and style, these are characterized in that deep learning algorithm is automatically learned, later to the corresponding time slice of eeg signal
On be labeled;
(4) characteristic model obtained using training is to the rhythm of the brain wave digital signal extraction in relation to singing, rhythm, style
Feature, and final singing disaggregated model is trained according to linear regression algorithm.Singing disaggregated model is that logistic recurrence is more
Classification, it is as follows that logistic returns more principles of classification:The classification of Hypothetical classification has k, and such Y is expressed as one-hot forms, tool
Body is that the label of its correspondence class is 1, remaining classification border is denoted as 0;Z is obtained using following two formula, and Z is a vector, in Z
Maximum value is classification results, and formula is specific as follows:(i) Y=Ax+b;(ii) Z=1/ (1+exp (- Y));Above-mentioned formula is all
Vector form, wherein Z is vector, indicates final classification, A and b are singing disaggregated model parameters, and A is matrix, and b is scalar, and x is
The feature vector of extraction, Y are vectors, indicate results of intermediate calculations.
It is as follows to evaluate and test stage etch:
(a) brain wave sensor detects human brain, will collect brain wave original signal every time and be converted to brain wave number
Signal;
(b) utilize signal processing algorithm handle brain wave digital signal, obtain spectrum signal, by brain wave digital signal into
Row segment processing obtains frequency-region signal to each segment signal using Fast Fourier Transform (FFT), extracts power spectrum to frequency-region signal, most
Log transformation is carried out to power spectrum afterwards;Segment processing mode is specially:It is 1s per segment length, every section is disposed, and moves backward
0.5s has between adjacent two sections and is O.5s overlapped, and reprocesses, until being disposed;Fast Fourier Transform (FFT) is direct computation of DFT
The fast algorithm of leaf transformation;
(c) the deep learning model obtained according to training, the spy of the rhythm, rhythm, style sung to spectrum signal extraction
Sign;
(d) the singing disaggregated model obtained using training, and classified to singing according to the feature of extraction.
The present invention proposes the automatic classification method of singing, by acquiring the eeg signal of user, by depth
Algorithm extraction is practised in relation to the rhythm sung, the feature of rhythm, style, classifies on singing disaggregated model, obtains final
It sings and classifies.The present invention runs brain wave detection, signal processing technology and machine learning method, realizes and carries out standard to singing
Really, automatically evaluation and test, has reached the automatic classification purpose to singing.
Claims (7)
1. a kind of singing sorting technique based on brain wave, characterized in that including training stage and evaluation and test stage, the training
Stage refers to:With eeg signal, the feature of the rhythm of singing, rhythm, style is trained to carry using deep learning technology
Modulus type extracts feature according to Feature Selection Model, and final training obtains singing disaggregated model;What the evaluation and test stage referred to
It is:Brain wave sensor obtains eeg signal, extracts the rhythm of singing, the feature of rhythm, style, utilizes singing disaggregated model
Classify to singing.
2. a kind of singing sorting technique based on brain wave according to claim 1, characterized in that the training stage
Steps are as follows:
(1) data collection establishes eeg signal language material and singing corpus;Brain wave sensor detects human brain, will adopt every time
Collection obtains brain wave original signal and is converted to brain wave digital signal;
(2) it utilizes signal processing algorithm to handle brain wave digital signal, obtains spectrum signal;Detailed process is as follows:By brain wave
Digital signal carries out segment processing, obtains frequency-region signal using Fast Fourier Transform (FFT) to each segment signal, is carried to frequency-region signal
Power spectrum is taken, Log transformation finally is carried out to power spectrum, obtains Log power spectrum, i.e. spectrum signal;
(3) spectrum signal is used, the characteristic model of the rhythm, rhythm, style sung using deep learning model training, while profit
Feature is extracted to spectrum signal with deep learning model;
(4) spy of the characteristic model obtained using training to the rhythm of the brain wave digital signal extraction in relation to singing, rhythm, style
Sign, and final singing disaggregated model is trained according to linear regression algorithm.
3. a kind of singing sorting technique based on brain wave according to claim 2, characterized in that in step (2), point
Section processing mode be specially:It is 1s per segment length, every section is disposed, and moves backward 0.5s, 0.5s is had between adjacent two sections
Overlapping, reprocessing, until being disposed;Fast Fourier Transform (FFT) is the fast algorithm of discrete Fourier transform.
4. a kind of singing sorting technique based on brain wave according to claim 2, characterized in that in step (3), institute
The deep learning model stated includes deep neural network, convolutional neural networks and Recognition with Recurrent Neural Network;Deep learning is engineering
The branch of habit, be it is a kind of attempt using the multiple process layers constituted comprising labyrinth or by multiple nonlinear transformation to data into
The feature of the algorithm of row higher level of abstraction, extraction includes the rhythm, rhythm and the style sung, these are characterized in deep learning algorithm certainly
Dynamic study obtains, later to being labeled on the corresponding time slice of eeg signal.
5. a kind of singing sorting technique based on brain wave according to claim 2, characterized in that in step (4), sing
It is that logistic returns more classification to sing disaggregated model, and it is as follows that logistic returns more principles of classification:The classification of Hypothetical classification has k
A, such Y is expressed as one-hot forms, and it is 1 that specifically it, which corresponds to the label of class, remaining category label is 0;Use following two
A formula obtains Z, and Z is a vector, and the maximum value in Z is classification results, and formula is specific as follows:(i) Y=Ax+b;(ii)Z
=1/ (1+exp (- Y));Above-mentioned formula is all vector form, wherein Z is vector, indicates that final classification, A and b are classification of singing
Model parameter, A are matrixes, and b is scalar, and x is the feature vector of extraction, and Y is vector, indicates results of intermediate calculations.
6. a kind of singing sorting technique based on brain wave according to Claims 2 or 3 or 4 or 5, characterized in that described
Evaluation and test stage etch it is as follows:
(a) brain wave sensor detects human brain, will collect brain wave original signal every time and be converted to brain wave digital signal;
(b) it utilizes signal processing algorithm to handle brain wave digital signal, obtains spectrum signal, brain wave digital signal is divided
Section processing, frequency-region signal is obtained to each segment signal using Fast Fourier Transform (FFT), and power spectrum is extracted to frequency-region signal, finally right
Power spectrum carries out Log transformation;
(c) the deep learning model obtained according to training, the feature of the rhythm, rhythm, style sung to spectrum signal extraction;
(d) the singing disaggregated model obtained using training, and classified to singing according to the feature of extraction.
7. a kind of singing sorting technique based on brain wave according to claim 6, characterized in that in step (b), point
Section processing mode be specially:It is 1s per segment length, every section is disposed, and moves backward 0.5s, 0.5s is had between adjacent two sections
Overlapping, reprocessing, until being disposed;Fast Fourier Transform (FFT) is the fast algorithm of discrete Fourier transform.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810296369.XA CN108596049A (en) | 2018-03-30 | 2018-03-30 | A kind of singing sorting technique based on brain wave |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810296369.XA CN108596049A (en) | 2018-03-30 | 2018-03-30 | A kind of singing sorting technique based on brain wave |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108596049A true CN108596049A (en) | 2018-09-28 |
Family
ID=63624382
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810296369.XA Pending CN108596049A (en) | 2018-03-30 | 2018-03-30 | A kind of singing sorting technique based on brain wave |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108596049A (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101015451A (en) * | 2007-02-13 | 2007-08-15 | 电子科技大学 | Music brain electricity analytical method |
CN101690659B (en) * | 2009-09-29 | 2012-07-18 | 华东理工大学 | Brain wave analysis method |
CN102999701A (en) * | 2012-11-28 | 2013-03-27 | 电子科技大学 | Brain wave music generation method |
CN103412646A (en) * | 2013-08-07 | 2013-11-27 | 南京师范大学 | Emotional music recommendation method based on brain-computer interaction |
CA2923979A1 (en) * | 2012-09-14 | 2014-03-20 | Interaxon Inc. | Systems and methods for collecting, analyzing, and sharing bio-signal and non-bio-signal data |
US20150297109A1 (en) * | 2014-04-22 | 2015-10-22 | Interaxon Inc. | System and method for associating music with brain-state data |
CN106803081A (en) * | 2017-01-25 | 2017-06-06 | 东南大学 | A kind of brain electricity sorting technique based on Multi-classifers integrated |
CN107423352A (en) * | 2017-05-25 | 2017-12-01 | 杭州回车电子科技有限公司 | Music recommends method and system |
-
2018
- 2018-03-30 CN CN201810296369.XA patent/CN108596049A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101015451A (en) * | 2007-02-13 | 2007-08-15 | 电子科技大学 | Music brain electricity analytical method |
CN101690659B (en) * | 2009-09-29 | 2012-07-18 | 华东理工大学 | Brain wave analysis method |
CA2923979A1 (en) * | 2012-09-14 | 2014-03-20 | Interaxon Inc. | Systems and methods for collecting, analyzing, and sharing bio-signal and non-bio-signal data |
CN102999701A (en) * | 2012-11-28 | 2013-03-27 | 电子科技大学 | Brain wave music generation method |
CN103412646A (en) * | 2013-08-07 | 2013-11-27 | 南京师范大学 | Emotional music recommendation method based on brain-computer interaction |
US20150297109A1 (en) * | 2014-04-22 | 2015-10-22 | Interaxon Inc. | System and method for associating music with brain-state data |
CN106803081A (en) * | 2017-01-25 | 2017-06-06 | 东南大学 | A kind of brain electricity sorting technique based on Multi-classifers integrated |
CN107423352A (en) * | 2017-05-25 | 2017-12-01 | 杭州回车电子科技有限公司 | Music recommends method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106782602B (en) | Speech emotion recognition method based on deep neural network | |
CN104751163B (en) | The fluoroscopic examination system and method for automatic Classification and Identification are carried out to cargo | |
CN109255044A (en) | A kind of image intelligent mask method based on YOLOv3 deep learning network | |
CN110108992B (en) | Cable partial discharge fault identification method and system based on improved random forest algorithm | |
Piczak | Recognizing bird species in audio recordings using deep convolutional neural networks | |
Huang et al. | Intelligent feature extraction and classification of anuran vocalizations | |
CN107292891A (en) | A kind of detection method of counting of the southern vegetables Severe pests based on machine vision | |
CN106845528A (en) | A kind of image classification algorithms based on K means Yu deep learning | |
CN109343046A (en) | Radar gait recognition method based on multifrequency multiple domain deep learning | |
CN107468260A (en) | A kind of brain electricity analytical device and analysis method for judging ANIMAL PSYCHE state | |
CN104978884A (en) | Teaching system of preschool education profession student music theory and solfeggio learning | |
CN109408660A (en) | A method of the music based on audio frequency characteristics is classified automatically | |
CN108509939A (en) | A kind of birds recognition methods based on deep learning | |
CN108256579A (en) | A kind of multi-modal sense of national identity quantization measuring method based on priori | |
CN109446985A (en) | Multi-angle plants identification method based on vector neural network | |
CN111160438A (en) | Acoustic garbage classification method adopting one-dimensional convolutional neural network | |
CN113116361A (en) | Sleep staging method based on single-lead electroencephalogram | |
Tang et al. | Transound: Hyper-head attention transformer for birds sound recognition | |
Gavhale et al. | Identification of medicinal plant using Machine learning approach | |
Berger et al. | Bird audio detection-dcase 2018 | |
CN108537261A (en) | A kind of composition evaluating method based on brain wave | |
CN108596049A (en) | A kind of singing sorting technique based on brain wave | |
CN108447504A (en) | A kind of oral evaluation method based on brain wave | |
CN116721060A (en) | Method for detecting red palm weevil beetles based on attention mechanism | |
Zhang et al. | A novel insect sound recognition algorithm based on MFCC and CNN |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180928 |