CN109493874A - A kind of live pig cough sound recognition methods based on convolutional neural networks - Google Patents
A kind of live pig cough sound recognition methods based on convolutional neural networks Download PDFInfo
- Publication number
- CN109493874A CN109493874A CN201811402994.4A CN201811402994A CN109493874A CN 109493874 A CN109493874 A CN 109493874A CN 201811402994 A CN201811402994 A CN 201811402994A CN 109493874 A CN109493874 A CN 109493874A
- Authority
- CN
- China
- Prior art keywords
- sound
- cough
- neural networks
- convolutional neural
- live pig
- 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
- 206010011224 Cough Diseases 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 20
- 238000012549 training Methods 0.000 claims description 19
- 238000012360 testing method Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 6
- 210000002569 neuron Anatomy 0.000 claims description 6
- 230000005236 sound signal Effects 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 241001269238 Data Species 0.000 claims description 2
- 208000023504 respiratory system disease Diseases 0.000 abstract description 8
- 201000010099 disease Diseases 0.000 abstract description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 abstract description 4
- 238000009395 breeding Methods 0.000 abstract description 3
- 208000024891 symptom Diseases 0.000 abstract description 3
- 230000003115 biocidal effect Effects 0.000 abstract description 2
- 239000003814 drug Substances 0.000 abstract description 2
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 239000013598 vector Substances 0.000 description 5
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 4
- 241000282887 Suidae Species 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 229910052742 iron Inorganic materials 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000005549 size reduction Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 208000018569 Respiratory Tract disease Diseases 0.000 description 1
- 206010039101 Rhinorrhoea Diseases 0.000 description 1
- 238000009360 aquaculture Methods 0.000 description 1
- 244000144974 aquaculture Species 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000036760 body temperature Effects 0.000 description 1
- 230000001488 breeding effect Effects 0.000 description 1
- 239000004568 cement Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000005034 decoration Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 210000005069 ears Anatomy 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000009408 flooring Methods 0.000 description 1
- 239000006260 foam Substances 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 208000010753 nasal discharge Diseases 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 230000004580 weight loss Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L17/00—Speaker identification or verification techniques
- G10L17/26—Recognition of special voice characteristics, e.g. for use in lie detectors; Recognition of animal voices
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
- G10L25/18—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being spectral information of each sub-band
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Evolutionary Computation (AREA)
- Acoustics & Sound (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Signal Processing (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Image Analysis (AREA)
Abstract
The live pig cough sound recognition methods based on convolutional neural networks that the invention discloses a kind of, belongs to field of voice signal.Respiratory disease is to influence one of the most common disease of pig-breeding industry, the serious development for restricting live pig industry, cough is one of the manifest symptom of live pig respiratory disease, the early warning to respiratory disease may be implemented with identification by the monitoring to cough sound, to effectively reduce disease's spread and the use of antibiotic medicine.Existing live pig cough sound recognition methods has that accuracy of identification is low, it is particularly evident particularly with intensive large-scale pig house, the invention proposes a kind of, and the convolutional neural networks model based on AlexNet identifies cough sound, using the sound spectrograph of voice signal as the input of network, network can automatically extract the further feature of sound spectrograph, compared to conventional sound identification method, this method can effectively promote cough sound and whole accuracy of identification.
Description
Technical field
The invention belongs to field of voice signal, and in particular to a kind of live pig cough sound based on convolutional neural networks
Recognition methods.
Background technique
Respiratory disease is to influence one of the most common disease of pig-breeding industry, and respiratory disease can cause live pig to breathe
Difficult, body temperature repeatedly, cough, have sharp ears it is rubescent, partially there is the symptoms such as spit out white foams, eye nasal discharge increases, while respiratory tract
Disease is easy to infect, and less serious case can cause live pig weight loss, and serious person will lead to live pig death, bring huge damage to raiser
It loses, therefore early warning is carried out to respiratory disease and is had great importance.One of the obvious symptom of respiratory disease performance
It is cough, the early warning to respiratory disease can be realized with identification by the monitoring to cough sound, to effectively reduce disease
The use of diffusion and antibiotic medicine.
Currently, live pig cough sound recognition methods mainly includes dynamic time warping (DTW), vector quantization (VQ), Fuzzy C
The combination and improvement of mean cluster (FCM), hidden Markov models (HMM), artificial neural network (ANN) and various algorithms
It needs artificial to extract voice signal property vector as input and carry out Modulation recognition and identification, common sound Deng, these algorithms
Signal characteristic vector includes power spectral density (PSD), mel cepstrum coefficients (MFCC), linear prediction residue error (LPCC) etc.,
These feature vectors are for there is preferable discrimination between certain human speech signals, but such as sound certain in pig house
The discriminations such as cough, shriek and the metallic sound that bites are not high, and nicety of grading is not high when as feature vector.These common coughs
Sound identification method of coughing obtains that live pig cough sound Classification and Identification effect is preferable in laboratory environments, but intensive big
In type farm, under complicated breeding environment, since noise jamming is bigger, recognition effect is poor.
Summary of the invention
The present invention against the above deficiency place, provide a kind of live pig cough sound identification side based on convolutional neural networks
Method, its technical solution is as follows:
It by voice signal in pick up facility and the pig house of data acquisition equipment acquisition and saves into audio file, passes through people
The method of work Auditory identification identifies effective audio signal in pig house, and individually saves into audio file, the sound that will identify that
Frequency signal is divided into two classes, and one kind is live pig cough sound, and another kind of is non-live pig cough sound, generates the language of two class voice signals
Spectrogram is simultaneously saved into picture format, two class data is respectively classified into two datasets, one is used as training set, and one as test
Collection, is input to convolutional neural networks for training set data and corresponding data label and is trained to network, completes network training
After process, voice signal sound spectrograph to be identified in test set is inputted into trained convolutional neural networks, output result is two
The label of class signal can identify cough sound according to label.
The pick up facility frequency range is not less than 100Hz-16,000Hz, and audio signal sample frequency is lower than 32,
000Hz。
The sound spectrograph of the two class voice signal of generation simultaneously saves mode at picture format are as follows: first by voice signal
Preemphasis processing is carried out, pre-emphasis transfer function is H (z)=1-az-1, the value range of pre emphasis factor a is 0.9 < a < 1.0,
Then the signal after preemphasis is subjected to sub-frame processing, each frame signal length range is 10ms-30ms, and frame moves length and is set as frame
The half of signal length carries out bandpass filtering, filter frequency range 100Hz-16,000Hz, to filtering to each frame signal
Rear each frame signal carries out time-domain windowed, and window function can be Hanning window, hamming window, kaiser window or Blackman window, so
Fourier transformation is carried out to each frame signal afterwards, logarithm operation is taken to the result after Fourier transformation and takes out positive frequency part number
According to, sort to form time-frequency two-dimensional sound spectrograph according to time series, by sound spectrograph save at 227 × 227 × 3 sizes RGB color
Picture.
The feature of the training set data and corresponding data label are as follows: training set data will include all types of sound
Sound signal, data label are defined as 0 and 1, can be with customized representative data type: 0 represents cough, and 1 represents non-cough
Sound 1 represents cough, and 0 represents non-cough.
The model that the convolutional neural networks are selected is the Alexnet model of fine tuning, first five layer of the model is convolutional layer,
Three layers are full articulamentum afterwards, and one LRN layers and maximum pond layer are all connected with after the first two convolutional layer, and the full articulamentum of the first two includes
4096 neurons, the last one full articulamentum include 2 neurons, compare the size of two neuron output datas, and value is big
Definition be 1, be worth small definition be 0, be compared with the data label of input, it is identical as input label i.e. indicate belong to input
The category signal of definition.
The invention has the benefit that using the three-dimensional sound spectrograph comprising time domain and frequency domain information as the defeated of classifier
Enter, more information can be provided for improving accuracy of identification, cough sound classification is carried out using convolutional neural networks, by more
A convolutional layer can automatically extract the further feature of input signal, strong antijamming capability, with regular speech signal sorting algorithm phase
Than the present invention can effectively improve live pig cough sound accuracy of identification.
Detailed description of the invention
Fig. 1 is that acquisition voice data experimental facilities lays figure in pig house
Fig. 2 is the voice signal segment time-domain diagram acquired in pig house
Fig. 3 is the sound spectrograph of alternative sounds signal
Fig. 4 is the AlexNet convolutional neural networks model structure of fine tuning
Cough, non-cough and whole classification accuracy when Fig. 5 is different test sets
Specific embodiment
Specific embodiments of the present invention will be described in further detail with example with reference to the accompanying drawing.Following instance is used for
Illustrate the present invention, but is not intended to limit the scope of the invention.
Live pig cough sound and other cry data are collected in Harbin, Heilongjiang Province Acheng District commercial aquaculture field
In one fattening house, live pig is in the fattening stage that the average monthly age is 5 half a months, average weight 60kg.Pig house having a size of
27.5 meters long × 13.7 meters wide × 3.2 meters of height share 6 industrial negative-pressure air fans in house and work, include 21 fences in house,
Wherein only having in 13 fences has pig, has 10 pigs in average each column, altogether 126 pigs, each fence is by 1.1 meters of high iron railings
Column besieged city, floor half are concrete floor, and half is slatted floor.
Voice signal in pig house is acquired by the microphone that a frequency range is 100Hz-16kHz, microphone connecting pen
Remember this computer sound card, recorded by recording software, microphone is fixed on apart from the position on 1.4 meters of ground, big apart from pig back
General 0.8 meter of height, microphone acquires voice data experiment photo in pig house as shown in Figure 1, Sampling with sound card rate is 44.1kHz, resolution
Rate is 16bits.8 days data of continuous acquisition, wherein the voice signal of any one section of acquisition is as shown in Fig. 2, by manually extracting
Method be extracted 4480 sections of different voice signals altogether, including 2703 sections of cough sounds and 1777 sections of non-cough sounds,
Non-cough sound includes 1463 sections of crys, 184 sections of metallic sounds, 99 sections of water flow sound and 31 sections of people's voices.Metallic sound is that have work
Personnel rub between spade and cement flooring and iron railing the sound of sending during cleaning up excrement.
The voice signal manually extracted is pre-processed, first progress preemphasis, pre-emphasis filter transfer function H
(z)=1-az-1, pre emphasis factor a value is common parameter 0.9375.Then data are subjected to sub-frame processing, a frame is taken to believe
Number length is 20ms, and frame moves that length is 10ms namely every two frame signal is overlapped 10ms, and the signal after framing is filtered and is added
Window, filter frequency range are the 10 rank butterworth filters of 100Hz-16kHz, and window function is selected as Hanning window.To each
Frame signal carries out Fourier transformation and takes logarithm, takes positive frequency part, and sequence obtains sound spectrograph in temporal sequence.Alternative sounds letter
Number sound spectrograph it is as shown in Figure 3, wherein C1, C2 and C3 represent the sound spectrograph of different cough sounds, and CL1 and CL2 represent metallic sound
Sound spectrograph, W represents the sound spectrograph of water flow sound, the sound spectrograph of H representative's voice, and S1-S5 represents the language spectrum of different shrieks
Figure, abscissa indicate the time, and unit is the second, and ordinate indicates frequency, and unit is kHz.
As shown in figure 4, network includes 8 layers, COVN1-CONV5 is the AlexNet convolutional neural networks model structure of fine tuning
Convolutional layer, FC6-FC8 are full articulamentum, and the RGB sound spectrograph and corresponding data label that input signal is 227 × 227 × 3 are false
If cough label is 1, non-cough label is 0.In the 1st layer, using 96 having a size of 11 × 11, convolution that step-length is 4
Filter makes input signal size be down to 55 × 55, then constructs maximum pond layer with one 3 × 3 filter, makes convolutional layer
Size reduction is to 27 × 27 × 96.Then the convolution for executing one 5 × 5 again at the 2nd layer, after filling, the ruler of the result of output
Very little is 27 × 27 × 256；Then maximum pond is carried out again.Execute once convolution identical with previous step, phase again in the 3rd layer
Same filling, the size of obtained result are 13 × 13 × 384；2 times 3 × 3 identical convolution are done in the 4th, 5 layer again, finally
Primary maximum pond is carried out again, as a result size reduction to 6 × 6 × 256.Then this result is input to full articulamentum, finally
The category result of a prediction is generated to the time-frequency figure of the input.Then the prediction result is calculated according to the label classification of input
With the error of concrete class label, the forward reasoning operation in an iteration is completed.Next, using the side of stochastic gradient descent
Method constantly reduces loss, and the update of parameter is carried out using backpropagation, completes the backward learning of an iteration.In model
After convergence, one group of optimum set of parameters of network is obtained, the training process of model is completed.After training model, verifying is concentrated
Time-frequency figure is input in network, and is predicted using classification of the model of preservation to audio image.In this stage, before only executing
To propagation, and the prediction classification with maximum probability is final recognition result.
The model after fine tuning is realized in emulation using python language and TensorFlow deep learning frame.Entire instruction
Practice and the process of test is carried out in batches on the GPU of two pieces of 1080 Ti of model GeForce GTX, batch size is
64, the number of iterations is set as 200.Each iteration includes two processes of training and test, in training process each time, using random
Gradient descent algorithm carrys out training pattern.For all network layers, it is trained using identical learning rate, it is initial to learn
Rate is 0.001, and learning rate is set as current value when validation error tends towards stability under current learning rate and obtains 1/10, and at end
It is reduced three times before only.
In simulation process, use 30% (1344) of all voice signals as test set, remaining signal is as instruction
Practice collection, training set quantity successively increases to 70% (3136) from 10% (448), and simulation result is as shown in table 1, when different training sets
Cough, non-cough and whole accuracy of identification are whole as shown in figure 5, can be seen that the increase with training set quantity from table 1 and Fig. 5
Body accuracy of identification is gradually increased, and cough accuracy of identification reaches highest when training set quantity is 70%, is 97.5%.
The classification performance of the algorithm proposed by the present invention of table 1
The above described is only a preferred embodiment of the present invention, be not intended to limit the present invention in any form, though
So the present invention has been disclosed as a preferred embodiment, and however, it is not intended to limit the invention, any technology people for being familiar with this profession
Member, in the range of not departing from technical solution of the present invention, when the technology contents using the disclosure above are modified or are repaired
Decorations are the equivalent embodiment of equivalent variations, but without departing from the technical solutions of the present invention, according to the technical essence of the invention
To any simple modification, equivalent change and modification made by above example, all of which are still within the scope of the technical scheme of the invention.
Claims (5)
1. a kind of live pig cough sound recognition methods based on convolutional neural networks, it is characterised in that: pass through pick up facility sum number
It according to voice signal in the pig house of acquisition equipment acquisition and saves into audio file, method for distinguishing is known by artificial hearing and identifies pig
Effective audio signal in giving up, and audio file is individually saved into, the audio signal that will identify that is divided into two classes, and one kind is live pig
Cough sound, another kind of is non-live pig cough sound, generates the sound spectrograph of two class voice signals and saves at picture format, by two
Class data are respectively classified into two datasets, and one is used as training set, and one is used as test set, by training set data and corresponding number
Convolutional neural networks are input to according to label to be trained network, it, will be to be identified in test set after completing network training process
Voice signal sound spectrograph inputs trained convolutional neural networks, and output result is the label of two class signals, according to label
Cough sound is identified.
2. a kind of live pig cough sound recognition methods based on convolutional neural networks according to claim 1, feature exist
In: the pick up facility frequency range is not less than 100Hz-16,000Hz, and audio signal sample frequency is lower than 32,000Hz.
3. a kind of live pig cough sound recognition methods based on convolutional neural networks according to claim 1, feature exist
In: the two class voice signal of generation sound spectrograph and save mode at picture format are as follows: voice signal is carried out first
Preemphasis processing, pre-emphasis transfer function are H (z)=1-az-1, the value range of pre emphasis factor a is 0.9 < a < 1.0, then
Signal after preemphasis is subjected to sub-frame processing, each frame signal length range is 10ms-30ms, and frame moves length and is set as frame signal
The half of length carries out bandpass filtering, filter frequency range 100Hz-16,000Hz, to filtered to each frame signal
Each frame signal carries out time-domain windowed, and window function can be Hanning window, hamming window, kaiser window or Blackman window, then right
Each frame signal carries out Fourier transformation, takes logarithm operation to the result after Fourier transformation and takes out positive frequency part data,
It sorts to form time-frequency two-dimensional sound spectrograph according to time series, sound spectrograph is saved into the RGB color figure at 227 × 227 × 3 sizes
Piece.
4. a kind of live pig cough sound recognition methods based on convolutional neural networks according to claim 1, feature exist
In: the feature of the training set data and corresponding data label are as follows: training set data will include that all types of sound are believed
Number, data label is defined as 0 and 1, and can be with customized representative data type: 0 represents cough, and 1 represents non-cough, or
Person 1 represents cough, and 0 represents non-cough.
5. a kind of live pig cough sound recognition methods based on convolutional neural networks according to claim 1, feature exist
Be the Alexnet model of fine tuning in: the model that the convolutional neural networks are selected, first five layer of the model is convolutional layer, rear three
Layer is full articulamentum, and one LRN layers and maximum pond layer are all connected with after the first two convolutional layer, and the full articulamentum of the first two includes 4096
A neuron, the last one full articulamentum include 2 neurons, compare the size of two neuron output datas, are worth and big determine
Justice is 1, and being worth small definition is 0, is compared with the data label of input, identical as input label to indicate to belong to input definition
Category signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811402994.4A CN109493874A (en) | 2018-11-23 | 2018-11-23 | A kind of live pig cough sound recognition methods based on convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811402994.4A CN109493874A (en) | 2018-11-23 | 2018-11-23 | A kind of live pig cough sound recognition methods based on convolutional neural networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109493874A true CN109493874A (en) | 2019-03-19 |
Family
ID=65697417
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811402994.4A Pending CN109493874A (en) | 2018-11-23 | 2018-11-23 | A kind of live pig cough sound recognition methods based on convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109493874A (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110338092A (en) * | 2019-07-01 | 2019-10-18 | 河南牧业经济学院 | A kind of pig Activity recognition method and system based on sound |
| CN110532424A (en) * | 2019-09-26 | 2019-12-03 | 西南科技大学 | A kind of lungs sound tagsort system and method based on deep learning and cloud platform |
| CN110782901A (en) * | 2019-11-05 | 2020-02-11 | 深圳大学 | Method, storage medium and device for identifying voice of network telephone |
| CN111626093A (en) * | 2020-03-27 | 2020-09-04 | 国网江西省电力有限公司电力科学研究院 | Electric transmission line related bird species identification method based on sound power spectral density |
| CN111653273A (en) * | 2020-06-09 | 2020-09-11 | 杭州叙简科技股份有限公司 | Out-hospital pneumonia preliminary identification method based on smart phone |
| CN111860203A (en) * | 2020-06-28 | 2020-10-30 | 北京小龙潜行科技有限公司 | Abnormal pig identification device, system and method based on image and audio mixing |
| CN111951812A (en) * | 2020-08-26 | 2020-11-17 | 杭州情咖网络技术有限公司 | Animal emotion recognition method and device and electronic equipment |
| CN112259127A (en) * | 2020-09-24 | 2021-01-22 | 上海荷福人工智能科技（集团）有限公司 | Cough and sneeze monitoring and identifying method |
| CN112562739A (en) * | 2020-11-20 | 2021-03-26 | 山东泛在智能科技有限公司 | Large-scale feeding type broiler respiratory disease sound data analysis system |
| CN112863667A (en) * | 2021-01-22 | 2021-05-28 | 杭州电子科技大学 | Lung sound diagnosis device based on deep learning |
| CN112885333A (en) * | 2021-01-13 | 2021-06-01 | 上海英粤汽车科技有限公司 | Animal health state diagnosis instrument through voice recognition technology |
| CN113011302A (en) * | 2021-03-11 | 2021-06-22 | 国网电力科学研究院武汉南瑞有限责任公司 | Thunder signal identification system and method based on convolutional neural network |
| CN113096691A (en) * | 2021-03-22 | 2021-07-09 | 深圳市安保科技有限公司 | Detection method, device, equipment and computer storage medium |
| CN113488071A (en) * | 2021-07-16 | 2021-10-08 | 河南牧原智能科技有限公司 | Pig cough recognition method, device, equipment and readable storage medium |
| CN113566948A (en) * | 2021-07-09 | 2021-10-29 | 中煤科工集团沈阳研究院有限公司 | Fault audio recognition and diagnosis method for robot coal pulverizer |
| CN113804767A (en) * | 2021-08-16 | 2021-12-17 | 东南大学 | Bolt failure detection method |
| CN113870896A (en) * | 2021-09-27 | 2021-12-31 | 动者科技(杭州)有限责任公司 | Motion sound false judgment method and device based on time-frequency graph and convolutional neural network |
| CN114209302A (en) * | 2021-12-08 | 2022-03-22 | 赵永源 | Cough detection method based on data uncertainty learning |
| CN114330454A (en) * | 2022-01-05 | 2022-04-12 | 东北农业大学 | Live pig cough sound identification method based on DS evidence theory fusion characteristics |
| CN115273908A (en) * | 2022-08-05 | 2022-11-01 | 东北农业大学 | Live pig cough sound identification method based on classifier fusion |
| CN117423342A (en) * | 2023-10-27 | 2024-01-19 | 东北农业大学 | Pig abnormal state monitoring method and system based on edge calculation |
| CN117423342B (en) * | 2023-10-27 | 2024-06-07 | 东北农业大学 | Pig abnormal state monitoring method and system based on edge calculation |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104217722A (en) * | 2014-08-22 | 2014-12-17 | 哈尔滨工程大学 | Dolphin whistle signal spectrum contour extraction method |
| CN106847262A (en) * | 2016-12-28 | 2017-06-13 | 华中农业大学 | A kind of porcine respiratory disease automatic identification alarm method |
| CN107799114A (en) * | 2017-04-26 | 2018-03-13 | 珠海智牧互联科技有限公司 | A kind of pig cough sound recognition methods and system |
-
2018
- 2018-11-23 CN CN201811402994.4A patent/CN109493874A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104217722A (en) * | 2014-08-22 | 2014-12-17 | 哈尔滨工程大学 | Dolphin whistle signal spectrum contour extraction method |
| CN106847262A (en) * | 2016-12-28 | 2017-06-13 | 华中农业大学 | A kind of porcine respiratory disease automatic identification alarm method |
| CN107799114A (en) * | 2017-04-26 | 2018-03-13 | 珠海智牧互联科技有限公司 | A kind of pig cough sound recognition methods and system |
Non-Patent Citations (3)
| Title |
|---|
| HUI-HUI WANG，等: "Audio Signals Encoding for Cough Classification Using Convolutional Neural Networks: A Comparative Study", 《2015 IEEE INTERNATIONAL CONFERENCE ON BIOINFORMATICS AND BIOMEDICINE》 * |
| 尹永: "基于Mel倒谱参数的咳嗽声识别", 《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》 * |
| 龚永杰: "基于声音识别技术的猪咳嗽音识别方法的研究", 《万方数据库》 * |
Cited By (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110338092A (en) * | 2019-07-01 | 2019-10-18 | 河南牧业经济学院 | A kind of pig Activity recognition method and system based on sound |
| CN110532424A (en) * | 2019-09-26 | 2019-12-03 | 西南科技大学 | A kind of lungs sound tagsort system and method based on deep learning and cloud platform |
| CN110782901A (en) * | 2019-11-05 | 2020-02-11 | 深圳大学 | Method, storage medium and device for identifying voice of network telephone |
| CN110782901B (en) * | 2019-11-05 | 2021-12-24 | 深圳大学 | Method, storage medium and device for identifying voice of network telephone |
| CN111626093B (en) * | 2020-03-27 | 2023-12-26 | 国网江西省电力有限公司电力科学研究院 | Method for identifying related bird species of power transmission line based on sound power spectral density |
| CN111626093A (en) * | 2020-03-27 | 2020-09-04 | 国网江西省电力有限公司电力科学研究院 | Electric transmission line related bird species identification method based on sound power spectral density |
| CN111653273A (en) * | 2020-06-09 | 2020-09-11 | 杭州叙简科技股份有限公司 | Out-hospital pneumonia preliminary identification method based on smart phone |
| CN111860203A (en) * | 2020-06-28 | 2020-10-30 | 北京小龙潜行科技有限公司 | Abnormal pig identification device, system and method based on image and audio mixing |
| CN111860203B (en) * | 2020-06-28 | 2023-09-01 | 北京小龙潜行科技有限公司 | Abnormal pig identification device, system and method based on image and audio mixing |
| CN111951812A (en) * | 2020-08-26 | 2020-11-17 | 杭州情咖网络技术有限公司 | Animal emotion recognition method and device and electronic equipment |
| CN112259127A (en) * | 2020-09-24 | 2021-01-22 | 上海荷福人工智能科技（集团）有限公司 | Cough and sneeze monitoring and identifying method |
| CN112562739A (en) * | 2020-11-20 | 2021-03-26 | 山东泛在智能科技有限公司 | Large-scale feeding type broiler respiratory disease sound data analysis system |
| CN112885333A (en) * | 2021-01-13 | 2021-06-01 | 上海英粤汽车科技有限公司 | Animal health state diagnosis instrument through voice recognition technology |
| CN112863667B (en) * | 2021-01-22 | 2023-10-24 | 杭州电子科技大学 | Lung sound diagnostic device based on deep learning |
| CN112863667A (en) * | 2021-01-22 | 2021-05-28 | 杭州电子科技大学 | Lung sound diagnosis device based on deep learning |
| CN113011302A (en) * | 2021-03-11 | 2021-06-22 | 国网电力科学研究院武汉南瑞有限责任公司 | Thunder signal identification system and method based on convolutional neural network |
| CN113011302B (en) * | 2021-03-11 | 2022-04-01 | 国网电力科学研究院武汉南瑞有限责任公司 | Thunder signal identification system and method based on convolutional neural network |
| CN113096691A (en) * | 2021-03-22 | 2021-07-09 | 深圳市安保科技有限公司 | Detection method, device, equipment and computer storage medium |
| CN113566948A (en) * | 2021-07-09 | 2021-10-29 | 中煤科工集团沈阳研究院有限公司 | Fault audio recognition and diagnosis method for robot coal pulverizer |
| CN113488071A (en) * | 2021-07-16 | 2021-10-08 | 河南牧原智能科技有限公司 | Pig cough recognition method, device, equipment and readable storage medium |
| CN113804767B (en) * | 2021-08-16 | 2022-11-04 | 东南大学 | Bolt failure detection method |
| CN113804767A (en) * | 2021-08-16 | 2021-12-17 | 东南大学 | Bolt failure detection method |
| CN113870896A (en) * | 2021-09-27 | 2021-12-31 | 动者科技(杭州)有限责任公司 | Motion sound false judgment method and device based on time-frequency graph and convolutional neural network |
| CN114209302A (en) * | 2021-12-08 | 2022-03-22 | 赵永源 | Cough detection method based on data uncertainty learning |
| CN114209302B (en) * | 2021-12-08 | 2024-03-26 | 赵永源 | Cough detection method based on data uncertainty learning |
| CN114330454A (en) * | 2022-01-05 | 2022-04-12 | 东北农业大学 | Live pig cough sound identification method based on DS evidence theory fusion characteristics |
| CN115273908A (en) * | 2022-08-05 | 2022-11-01 | 东北农业大学 | Live pig cough sound identification method based on classifier fusion |
| CN117423342A (en) * | 2023-10-27 | 2024-01-19 | 东北农业大学 | Pig abnormal state monitoring method and system based on edge calculation |
| CN117423342B (en) * | 2023-10-27 | 2024-06-07 | 东北农业大学 | Pig abnormal state monitoring method and system based on edge calculation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109493874A (en) | A kind of live pig cough sound recognition methods based on convolutional neural networks | |
| CN107492382B (en) | Voiceprint information extraction method and device based on neural network | |
| CN102982809B (en) | Conversion method for sound of speaker | |
| CN110534132A (en) | A kind of speech-emotion recognition method of the parallel-convolution Recognition with Recurrent Neural Network based on chromatogram characteristic | |
| CN108899051A (en) | A kind of speech emotion recognition model and recognition methods based on union feature expression | |
| CN110390955A (en) | A kind of inter-library speech-emotion recognition method based on Depth Domain adaptability convolutional neural networks | |
| CN103996155A (en) | Intelligent interaction and psychological comfort robot service system | |
| CN110111797A (en) | Method for distinguishing speek person based on Gauss super vector and deep neural network | |
| CN108694949A (en) | Method for distinguishing speek person and its device based on reorder super vector and residual error network | |
| Chen et al. | Tone Classification in Mandarin Chinese Using Convolutional Neural Networks. | |
| Zhao et al. | DNN-HMM based acoustic model for continuous pig cough sound recognition | |
| CN106611604A (en) | An automatic voice summation tone detection method based on a deep neural network | |
| CN107039036A (en) | A kind of high-quality method for distinguishing speek person based on autocoding depth confidence network | |
| CN108877812B (en) | Voiceprint recognition method and device and storage medium | |
| CN109461441A (en) | A kind of Activities for Teaching Intellisense method of adaptive, unsupervised formula | |
| CN104992707A (en) | Cleft palate voice glottal stop automatic identification algorithm and device | |
| CN112863667A (en) | Lung sound diagnosis device based on deep learning | |
| CN106297769B (en) | A kind of distinctive feature extracting method applied to languages identification | |
| CN116842460A (en) | Cough-related disease identification method and system based on attention mechanism and residual neural network | |
| CN113850013B (en) | Ship radiation noise classification method | |
| CN111090726A (en) | NLP-based electric power industry character customer service interaction method | |
| Ling | An acoustic model for English speech recognition based on deep learning | |
| Xiao et al. | AMResNet: An automatic recognition model of bird sounds in real environment | |
| Rabiee et al. | Persian accents identification using an adaptive neural network | |
| CN112466284B (en) | Mask voice identification method |
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 |
Application publication date: 20190319 |
|
| RJ01 | Rejection of invention patent application after publication |