CN107170466A - The sound detection method that mops floor based on audio - Google Patents
The sound detection method that mops floor based on audio Download PDFInfo
- Publication number
- CN107170466A CN107170466A CN201710242995.6A CN201710242995A CN107170466A CN 107170466 A CN107170466 A CN 107170466A CN 201710242995 A CN201710242995 A CN 201710242995A CN 107170466 A CN107170466 A CN 107170466A
- Authority
- CN
- China
- Prior art keywords
- sound
- probability
- mops floor
- audio
- channel
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 19
- 230000005021 gait Effects 0.000 claims abstract description 38
- 238000000034 method Methods 0.000 claims abstract description 31
- 238000012549 training Methods 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims abstract description 8
- 210000002683 foot Anatomy 0.000 claims description 27
- 210000001699 lower leg Anatomy 0.000 claims description 21
- 230000015572 biosynthetic process Effects 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 4
- 238000003786 synthesis reaction Methods 0.000 claims description 4
- 210000004744 fore-foot Anatomy 0.000 claims description 2
- 230000002159 abnormal effect Effects 0.000 claims 1
- 230000008569 process Effects 0.000 abstract description 6
- 238000004458 analytical method Methods 0.000 description 10
- 238000012360 testing method Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000004927 fusion Effects 0.000 description 4
- 238000009432 framing Methods 0.000 description 3
- 238000013480 data collection Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000009977 dual effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 230000005236 sound signal Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 206010017577 Gait disturbance Diseases 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 210000002414 leg Anatomy 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000002560 therapeutic procedure Methods 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
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
- G10L25/66—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination for extracting parameters related to health condition
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/112—Gait analysis
-
- 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
-
- 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
- G10L25/45—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of analysis window
-
- 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
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Physiology (AREA)
- Heart & Thoracic Surgery (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Epidemiology (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Dentistry (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention provides the detection method that a kind of step based on audio mops floor sound.This method includes：Sub-frame processing is carried out to the left and right pin dual-channel audio data collected, to obtain corresponding audio frame；Input is used as using the characteristic vector extracted from the audio frame, belong to the probability for the sound that mops floor using grader acquisition audio frame and belong to the probability of normal footsteps, wherein, what the grader was obtained by training, positive sample that training sample includes being used for recognizing normal footsteps, the sample and for the negative sample for the other sound for recognizing non-step of mopping floor for recognizing the sound that mops floor；The probability for the sound that mops floor is belonged to according to each audio frame of acquisition and belongs to the probability of normal footsteps, the corresponding time interval of the sound that mops floor is drawn.The method according to the invention can accurately detect the sound that mops floor in walking process, contribute to gait detection, tumble early warning etc..
Description
Technical field
The present invention relates to Computer Applied Technology field, more particularly to the sound that mops floor based on audio-frequency information detection method.
Background technology
Gait analysis (gait analysis) is a kind of posture by human body when observation or collection walking, draws and divides
Analyse the technology of gait parameter.Generally, gait parameter includes spatial parameter (such as stride, step-length, step width), time parameter (example
Such as cadence, leg speed) and these parameters symmetry, the stability of long term data of left and right pin etc..Gait analysis is in physical culture
It is widely used and studies in terms of motion, medical rehabilitation.
In gait analysis, whether pin is mopped floor medically, and referred to as foot is cleaned up, and the step of normal person lands and liftoff process
Stablize relatively, it is liftoff in swing process to have a sufficient height, and patient's foot-up and during landing along with starting and stop
On difficulty, the situation of scrape along the ground can be produced, and in swing process, can also produced bright due to pin raising height not enough
The aobvious sound mopped floor.The detection that foot is cleaned up has great significance for medical science of recovery therapy, gait detection, tumble early warning etc..
However, in the prior art, gait analysis is typically based on video image, pressure sensor, myoelectricity technology etc., but this
A little equipment have larger invasive to patient, particularly with the event of mopping floor, it is difficult to directly judge from motion sensor.This
Outside, although the existing step detection method based on audio is (for example, Application No. of the king into grade in the prior art
201610971951.2nd, the Chinese patent application of denomination of invention " the step detection method based on two-channel "), but it is not including sentencing
The scheme whether disconnected pin mops floor, in the prior art, also lacks general effective pin and mops floor testing mechanism.
The content of the invention
Therefore, can be accurate based on audio there is provided one kind it is an object of the invention to the defect for overcoming above-mentioned prior art
Detection step mops floor the method for sound.This method comprises the following steps：
Step 1：Sub-frame processing is carried out to the left and right pin dual-channel audio data collected, to obtain corresponding audio frame；
Step 2：Characteristic vector to be extracted from the audio frame is belonged to using grader acquisition audio frame and dragged as input
The probability of ground sound and the probability for belonging to normal footsteps, wherein, what the grader was obtained by training, training sample includes using
In recognizing the positive sample of normal footsteps, mop floor sample and other sound for recognizing non-step for recognizing the sound that mops floor
Negative sample；
Step 3：The probability for the sound that mops floor is belonged to according to each audio frame of acquisition and belongs to the probability of normal footsteps, is drawn
The corresponding time interval of the sound that mops floor.
Preferably, the positive sample is included under normal gait, before being labelled with the heelstrike audio frame of sound and being labelled with
Sole lands the audio frame of sound.
Preferably, the positive sample is included under known normal gait, in left foot channel audio data with each mark
Three audio frames heelstrike centered on the position of sound and three centered on the position of each mark forward roll sound
Three sounds centered on the position of the sound that landed by each mark heel in audio frame, and known right crus of diaphragm channel audio data
Frequency and three audio frames centered on the position of each mark forward roll sound.
Preferably, the sample that mops floor, which is included in, mops floor under gait, is labelled with the heelstrike audio frame of sound and is labelled with
The audio frame of forward roll sound.
Preferably, the sound sample that mops floor include it is known mop floor under gait, in left foot channel audio data with each
Mark three audio frames heelstrike centered on the position of sound and centered on the position of each mark forward roll sound
Three centered on the position of the sound that landed by each mark heel in three audio frames, and known right crus of diaphragm channel audio data
Individual audio and three audio frames centered on the position of each mark forward roll sound.
Preferably, the negative sample includes the forward roll sound and latter step of the back in left foot channel audio data
Heelstrike sound between nine audio frames, and back in right crus of diaphragm channel audio data forward roll sound with after
Nine audio frames between the heelstrike sound of one step.
Preferably, in the step 2, the probability that the audio frame of left foot sound channel and its belong to mops floor constitutes left foot sound channel
Mop floor sound probability curve, the audio frame of right crus of diaphragm sound channel and its belong to the sound that mops floor probability constitute right crus of diaphragm sound channel the sound that mops floor probability
Curve, the audio frame of left foot sound channel and its belong to the normal footsteps that the probability of normal footsteps constitutes left foot sound channel probability it is bent
Line, the audio frame of right crus of diaphragm sound channel and its belong to the normal footsteps that the probability of normal footsteps constitutes right crus of diaphragm sound channel probability it is bent
Line；In the step 3, in addition to the probability curve of the sound that mops floor of left and right pin sound channel is fused into the probability of the comprehensive sound that mops floor
Curve, the probability curve of the normal footsteps of left and right pin sound channel is fused into the probability curve of comprehensive normal footsteps, is based on
Default probability threshold value draws the corresponding time interval of normal footsteps time interval corresponding with the sound that mops floor.
Preferably, the corresponding time interval of normal footsteps is that the probability curve of the normal footsteps of the synthesis is less than
0.5 time interval；The corresponding time interval of the sound that mops floor is that the probability curve of the sound that mops floor of the synthesis is more than for 0.35 time
It is interval.
Compared with prior art, the advantage of the invention is that：Audio frame can accurately be detected according to dual-channel audio data
No is mop floor sound and/or normal footsteps, in addition, the method for the invention based on machine learning can be applied to a variety of different fields
Scape, highly versatile.
Brief description of the drawings
The following drawings only makees schematical description and interpretation to the present invention, is not intended to limit the scope of the present invention, wherein：
Fig. 1 shows the flow of the method according to an embodiment of the invention for training the grader for detecting the sound that mops floor
Figure；
Fig. 2 shows the flow of the method according to an embodiment of the invention for detecting mop floor sound and normal footsteps
Figure；
Fig. 3 shows sub-frame processing mode according to an embodiment of the invention；
Fig. 4 shows dual channel data mark example according to an embodiment of the invention；
Fig. 5 shows the example of the testing result of mop floor sound and normal footsteps according to an embodiment of the invention.
Embodiment
In order to which technical characteristic, purpose and effect to the present invention are more clearly understood from, referring now to accompanying drawing to the present invention
The detection method that mops floor of the step based on audio proposed elaborates.
In order to be clearly understood that the present invention, following patent (or patent application) is incorporated herein by way of quoting in full：
1. king is into the Application No. 201610971951.2, denomination of invention " the step detection method based on two-channel " waited
Chinese patent application.
2. king into wait Application No. 201610517381.X, denomination of invention " a kind of method for setting up gait data collection and
The Chinese patent application of gait analysis method ".
Fig. 1 shows the schematic flow diagram of the detection method that mops floor according to an embodiment of the invention based on audio.Tool
Body comprises the following steps：
Step S110, gathers voice data
By the way that wearable gait data acquisition device is respectively arranged at the pin of left and right, the sound letter produced during collection people's walking
Number, result in dual-channel audio data.
In one embodiment, wearable gait data acquisition device includes that the microphone list of acoustic signals can be gathered
Member.Gait data acquisition device includes left foot gait data acquisition node and right crus of diaphragm gait data acquisition node, and each gait is adopted
Collecting node includes memory cell, microprocessor, power subsystem, wireless transmit/receive units, signal picker, signal projector.Adopting
When collecting data, microprocessor is sent to by signal picker (such as microphone) collected sound signal, the signal gathered and carried out
Processing.
In one embodiment, the acquisition method of dual channel data includes：By left foot gait acquisition node and right crus of diaphragm gait
Acquisition node is separately fixed at the left foot and right crus of diaphragm of tested person.Use two gait collecting device nodes simultaneously on both feet, i.e.,
Left foot gait acquisition node gathers the voice data of left foot, and right crus of diaphragm gait acquisition node gathers the voice data of right crus of diaphragm, so that structure
Into two-channel, the data of left and right pin are subjected to analysis fusion, can be obtained than single pin metering system more accurately information.Specifically
Gait data acquisition node, can be worn on the diverse location of footwear, front side, outside or the rear side of such as upper of a shoe can also by ground
It is sole at forefoot, middle part or at the heel.Preferably, left foot gait acquisition node and right crus of diaphragm gait collection section
Point is worn in the symmetric position of left and right pin.
The specific method of voice data can be found in Chinese patent application " one kind set up gait data collection during for collection walking
Method and gait analysis method " (Chinese invention patent application number CN201610517381.X).
Step S120：Data are cut
Framing adding window is carried out to the dual-channel audio data collected, a series of audio frame is obtained.Point as shown in Figure 3
Frame adding window, under 8000hz audio sample rate, each audio frame includes 200 samples, in framing, between setting consecutive frame
There is the overlapping interval of 120 samples.After framing, in order to reduce the discontinuity of signal at frame starting and ending, to audio
Frame adds Hamming (hamming) window, i.e., add sliding window in voice data, take corresponding audio frame to be used as this implementation with sliding window
The basic investigation unit of example, because in the range of 10~30ms, the spectrum signature of audio and some physical characteristic parameters are basic
Keep constant, therefore the length of window of Hamming window is generally 10ms to 30ms.
Step S130：Extract and select audio frequency characteristics
Feature extraction is carried out to audio frame, to obtain the characteristic vector of the audio frame.According to one embodiment of present invention,
Characteristic vector includes：Auto-correlation coefficient, carry energy (0 to 4kHz) feature, zero-crossing rate, linear predictor coefficient (LPCC features) and
Mel cepstrum coefficients (MFCC) feature.Table 1 shows the composition of characteristic vector in one embodiment, including：10 dimension subband energy
Measure feature, 12 Jan Vermeer cepstrum coefficient features, 12 dimensional linear predictive coefficients, zero-crossing rate and auto-correlation coefficient, 36 are tieed up totally.
Table 1
| Feature | Dimension |
| Auto-correlation coefficient | 1 |
| Sub-belt energy (0~4kHz) | 10 |
| Zero-crossing rate | 1 |
| LPCC | 12 |
| MFCC | 12 |
It should be understood that the specific combination of the dimension and characteristic vector of features described above vector is not unique.Other
In embodiment, characteristic vector can also be that some or all of independent assortment in features described above either being capable of preferable earth's surface
Levy other combinations of features for the information that audio frame is contained.
Step S140：Select training sample
Two sound that the characteristics of typical footsteps lands comprising heel and ball of foot, and the audio data collecting of left and right pin sets
It can be collected in standby comprising corresponding earth signal, but the audio signal of this batter is relatively strong.Therefore, manually marking
When, each sound of step two is marked out in the audio of respective side successively (i.e. after pin according to left and right pin on the channel audio of left and right pin two
With two sound landed with ball of foot) position, as shown in Figure 4.
In the present embodiment, for the detection for realizing normal footsteps, mop floor three kinds of sound and other sound are classified, sample is trained
This include be used for recognize normal footsteps positive sample, for recognize mop floor sound mop floor sample and for recognize non-step its
The negative sample of its sound.
Preferably, under normal gait, on two channel audios of left and right pin centered on the position of each mark, 3 are respectively taken
Frame is as positive sample, so that (correspond to the sound channel of left foot or the sound channel of correspondence right crus of diaphragm) in monophonic, each step correspondence 6
Positive sample, then takes continuously in the centre position (in the middle of the rising tone of previous pin and the first sound of latter pin) of two neighboring step
9 frames as negative sample, have 18 negative samples between such each two step.
It will be understood by those of skill in the art that other audio frames can also be selected, as long as it can be with normal step
Sound is differentiated.
For the collection for the sample that mops floor, simulation can be carried out by normal person and mops floor to obtain the sample that mops floor, the side of mark
Method is identical with the mask method of normal step, for example, under the gait that mops floor, with each mark on two channel audios of left and right pin
Centered on position, 3 frames are respectively taken as positive sample, so that in monophonic, 6 samples that mop floor of each step correspondence.
In one embodiment, the training sample of collection includes the positive sample of 3264 audio frames, the negative sample of 4026 audio frames
The sample that mops floor of sheet and 463 audio frames.
It should be understood that each corresponding positive sample quantity of step, negative sample quantity and the sample size and each of mopping floor
The quantity of the corresponding audio frame of sample can consider training time and the model accuracy obtained to determine, however it is not limited to herein
Listed concrete numerical value.
Step S150：Train sorter model
Utilize above-mentioned positive and negative samples and the sample composition Sample Storehouse that mops floor, it is possible to use the method instruction of computer machine study
Practice grader, for example, SVMs (SVM), Weighted Support Vector, extreme learning machine or weighting extreme learning machine etc..Point
The input of class device is the characteristic vector extracted from each audio frame, and output is whether some audio frame is mop floor sound, normal step
The probability of sound and other sound, for each audio frame, the probability of this three and be 1.
The sound that mops floor is can detect using the grader trained, it is shown in Figure 2, in this embodiment, detection method bag
Include following steps：
Step S210, gathers voice data.
In this step, the step S110 according to Fig. 1 of the audio frame of dual-channel audio data to be detected side
Method is obtained.
Step S220：The probability that audio frame belongs to normal footsteps and the sound that mops floor is obtained using the grader trained.
Each audio frame of dual-channel audio data to be detected is detected using the grader trained, obtains each
Individual audio frame belongs to the probability for the sound that mops floor, and sets up corresponding probability curve.The probability curve refers to that abscissa is audio frame number
(or at the time of representated by audio frame), ordinate is the curve for the probability that correspondence audio frame belongs to the sound that mops floor.The two of left and right pin
Audio data two probability curves of correspondence.Similarly, also can obtain each audio frame belong to normal footsteps probability and its
The probability curve of its sound, and the corresponding probability curve of left and right pin is equally set up, referring to the probability of Fig. 5 normal footsteps illustrated
The probability curve of curve and the sound that mops floor.
Step S230：Smoothing processing probability curve simultaneously recognizes mop floor sound and normal footsteps
In this step, fusion is carried out and smooth to mop floor sound probability curve and the normal footsteps probability curve of left and right pin
Processing, for example, in one embodiment, after based on summation, the probability curve of left and right pin is merged, in order to overcome probability bent
Larger unstability and noise spot that line is present, are smoothed, smoothly with low pass filter (with respect to cut-off frequency 0.1)
There is more obvious " maximum probability " interval in probability curve afterwards, for example, in the probability curve for mopping floor sound that Fig. 5 illustrates, depositing
In the obvious time interval more than 0.35, it therefore, it can find out the continuous interval for exceeding threshold value according to default threshold value, by this
A little intervals are judged to belonging to the interval for the sound that mops floor.
In another embodiment, determine that step is interval based on two-channel maximum probability value method.As a rule, this side sound channel
Voice data be judged as mopping floor sound probability it is bigger, it is possible to more voice datas for relying on this side sound channel, and another
The voice data probability of side plays supplementary function.(sound of table moment identical left and right acoustic channels is referred to for the audio frame of each pair candidate
Frequency frame), larger one of probability is first selected, the probable value of the audio frame position in combined chance curve is then represented with it, this
Sample has just drawn the probability curve of comprehensive left and right pin voice data.To above-mentioned combined chance curve, sought with default probability threshold value
The continuous interval for exceeding threshold value is looked for, these intervals are judged to belonging to the interval for the sound that mops floor.
It is above-mentioned also to can be found in number of patent application for the detailed process that probability curve carries out fusion and smoothing processing
CN201610971951.2 (the step detection method based on two-channel).
And inventor has found in testing, the patient mopped floor for having, the step sound that mops floor has very high probability, Er Qiezheng
Normal footsteps can also obtain higher probability in itself, moreover, at statistics is the position of normal footsteps, may also include and mop floor
Sound.Therefore, in order to accurately distinguish mop floor sound and normal footsteps, in the present invention, the time interval for the sound that mopped floor in judgement,
Statistic analysis is not the sound that mops floor on the time interval of normal footsteps.For example, deterministic process is, first by normal footsteps
Interval on probability curve less than predetermined threshold value 0.5 is considered normal footsteps；It will be more than on the probability curve of normal footsteps
0.5 and the probability curve for the sound that mops floor on be more than 0.35 interval and be considered to mop floor the time interval of sound.In this way can
The enough sound that avoids mopping floor to a certain extent is covered by normal footsteps.Certainly, depending on specific application demand, it can also count together
Normal footsteps and the sound that mops floor.For example, the interval on the probability curve of normal footsteps less than predetermined threshold value 0.5 is considered just
Normal footsteps；The interval for being more than 0.35 on the probability curve for the sound that mops floor is considered to mop floor the time interval of sound.
Fig. 5 shows the schematic diagram of normal footsteps and the sound detection result that mops floor, and wherein abscissa represents each audio frame
Sequence number, ordinate respectively illustrates the probability curve of normal footsteps and the probability curve of sound of mopping floor.Original normal step
Sound probability curve, refer to after the fusion of left and right pin voice data, it is smooth before footsteps probability curve.The probability of normal footsteps is bent
The probability curve of line and the sound that mops floor is the probability curve after left and right pin voice data merges and be smooth.What Fig. 5 illustrated mops floor
The probabilistic determination threshold value of sound is that, more than 0.35, the probabilistic determination threshold value of normal footsteps is less than 0.5.
The above-mentioned sound detection method that mops floor based on audio, will not miss normal footsteps and the sound that mops floor, be called together with higher
Return rate and accuracy rate.
In order to further verify the technique effect of the present invention, sorter model of the inventor based on the present invention is surveyed
Examination.Test data includes：3 Healthy Peoples, 2 abnormal gait patients in 5 meters of long distances, go and back running 4 times.Test knot
As shown in table 2, health does not find the phenomenon that mops floor to fruit per capita, and patient has the more phenomenon that mops floor along with gait.
Table 2
| Healthy People 1 | Healthy People 2 | Healthy People 3 | Patient 1 | Patient 2 |
| 0 | 0 | 0 | 6 times | 8 times |
The present invention can be system, method and/or computer program product.Computer program product can include computer
Readable storage medium storing program for executing, containing for making processor realize the computer-readable program instructions of various aspects of the invention.
Computer-readable recording medium can be to maintain and store the tangible device that the instruction that equipment is used is performed by instruction.
Computer-readable recording medium can for example include but is not limited to storage device electric, magnetic storage apparatus, light storage device, electromagnetism and deposit
Store up equipment, semiconductor memory apparatus or above-mentioned any appropriate combination.The more specifically example of computer-readable recording medium
Sub (non exhaustive list) includes：Portable computer diskette, hard disk, random access memory (RAM), read-only storage (ROM),
Erasable programmable read only memory (EPROM or flash memory), static RAM (SRAM), Portable compressed disk are read-only
Memory (CD-ROM), digital versatile disc (DVD), memory stick, floppy disk, mechanical coding equipment, for example it is stored thereon with instruction
Punch card or groove internal projection structure and above-mentioned any appropriate combination.
It is described above various embodiments of the present invention, described above is exemplary, and non-exclusive, and
It is not limited to disclosed each embodiment.In the case of without departing from the scope and spirit of illustrated each embodiment, for this skill
Many modifications and changes will be apparent from for the those of ordinary skill in art field.The selection of term used herein, purport
Best explaining the principle of each embodiment, practical application or to the technological improvement in market, or making its of the art
Its those of ordinary skill is understood that each embodiment disclosed herein.
Claims (10)
1. a kind of sound detection method that mops floor based on audio, comprises the following steps：
Step 1：Sub-frame processing is carried out to the left and right pin dual-channel audio data collected, to obtain corresponding audio frame；
Step 2：Characteristic vector to be extracted from the audio frame obtains audio frame using grader and belongs to the sound that mops floor as input
Probability and belong to the probability of normal footsteps, wherein, what the grader was obtained by training, training sample includes being used to know
The positive sample of abnormal footsteps, the sample and for the negative sample for the other sound for recognizing non-step of mopping floor for recognizing the sound that mops floor
This；
Step 3：The probability for the sound that mops floor is belonged to according to each audio frame of acquisition and belongs to the probability of normal footsteps, draws and mops floor
The corresponding time interval of sound.
2. according to the method described in claim 1, wherein, the positive sample is included under normal gait, is labelled with heelstrike
The audio frame of sound and the audio frame for being labelled with forward roll sound.
3. method according to claim 2, wherein, the positive sample is included under known normal gait, left foot sound channel sound
Frequency in by each mark three audio frames heelstrike centered on the position of sound and with each mark forward roll
Three audio frames centered on the position of sound, and being landed sound with each mark heel in known right crus of diaphragm channel audio data
Position centered on three audios and by it is each mark forward roll sound position centered on three audio frames.
4. according to the method described in claim 1, wherein, the sample that mops floor, which is included in, to mop floor under gait, is labelled with heel and
The audio frame of ground sound and the audio frame for being labelled with forward roll sound.
5. method according to claim 4, wherein, the sound sample that mops floor mops floor under gait including known, left foot sound
In audio data by each mark three audio frames heelstrike centered on the position of sound and with each mark forefoot
Three audio frames landed centered on the position of sound, and with each mark heel in known right crus of diaphragm channel audio data
Three audios centered on the position of ground sound and three audio frames centered on the position of each mark forward roll sound.
6. according to the method described in claim 1, wherein, the negative sample includes the back in left foot channel audio data
Nine audio frames between forward roll sound and the heelstrike sound of latter step, and it is previous in right crus of diaphragm channel audio data
Nine audio frames between the forward roll sound of step and the heelstrike sound of latter step.
7. method according to any one of claim 1 to 6, wherein, in the step 2, the audio frame of left foot sound channel
And its belong to the sound probability curve that mops floor that the probability that mops floor constitutes left foot sound channel, the audio frame of right crus of diaphragm sound channel and its belong to the sound that mops floor
Probability constitute right crus of diaphragm sound channel the sound that mops floor probability curve, the audio frame of left foot sound channel and its probability for belonging to normal footsteps
Constitute the probability curve of the normal footsteps of left foot sound channel, the audio frame of right crus of diaphragm sound channel and its probability structure for belonging to normal footsteps
Into the probability curve of the normal footsteps of right crus of diaphragm sound channel；
In the step 3, in addition to the probability curve of the sound that mops floor of left and right pin sound channel is fused into the general of the comprehensive sound that mops floor
Rate curve, the probability curve of the normal footsteps of left and right pin sound channel is fused into the probability curve of comprehensive normal footsteps, base
The corresponding time interval of normal footsteps time interval corresponding with the sound that mops floor is drawn in default probability threshold value.
8. method according to claim 7, wherein, the corresponding time interval of normal footsteps is the normal foot of the synthesis
The probability curve of step sound is less than 0.5 time interval；The corresponding time interval of the sound that mops floor is the probability of the sound that mops floor of the synthesis
Curve is more than 0.35 time interval.
9. a kind of computer-readable recording medium, is stored thereon with computer program, wherein, it is real when the program is executed by processor
Existing step according to any one of claim 1 to 8.
10. a kind of computer equipment, including memory, processor and storage can be run on a processor on a memory
Computer program, it is characterised in that realized during the computing device described program any one of claim 1 to 8
Step.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710242995.6A CN107170466B (en) | 2017-04-14 | 2017-04-14 | Mopping sound detection method based on audio |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710242995.6A CN107170466B (en) | 2017-04-14 | 2017-04-14 | Mopping sound detection method based on audio |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107170466A true CN107170466A (en) | 2017-09-15 |
| CN107170466B CN107170466B (en) | 2020-12-29 |
Family
ID=59849006
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710242995.6A Active CN107170466B (en) | 2017-04-14 | 2017-04-14 | Mopping sound detection method based on audio |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107170466B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110364141A (en) * | 2019-06-04 | 2019-10-22 | 杭州电子科技大学 | Elevator typical case's abnormal sound alarm method based on depth single classifier |
| CN113679300A (en) * | 2021-09-13 | 2021-11-23 | 李侃 | Electronic book recommendation method and device for floor sweeping robot |
| CN114550395A (en) * | 2022-04-28 | 2022-05-27 | 常州分音塔科技有限公司 | Sound alarm detection method and device |
| GB2607561A (en) * | 2021-04-08 | 2022-12-14 | Miicare Ltd | Mobility analysis |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002197437A (en) * | 2000-12-27 | 2002-07-12 | Sony Corp | Walking detection system, walking detector, device and walking detecting method |
| US20050094795A1 (en) * | 2003-10-29 | 2005-05-05 | Broadcom Corporation | High quality audio conferencing with adaptive beamforming |
| CN101393660A (en) * | 2008-10-15 | 2009-03-25 | 中山大学 | Intelligent gate inhibition system based on footstep recognition |
| CN102799899A (en) * | 2012-06-29 | 2012-11-28 | 北京理工大学 | Special audio event layered and generalized identification method based on SVM (Support Vector Machine) and GMM (Gaussian Mixture Model) |
| CN103730124A (en) * | 2013-12-31 | 2014-04-16 | 上海交通大学无锡研究院 | Noise robustness endpoint detection method based on likelihood ratio test |
| CN104200815A (en) * | 2014-07-16 | 2014-12-10 | 电子科技大学 | Audio noise real-time detection method based on correlation analysis |
| CN105215542A (en) * | 2015-10-14 | 2016-01-06 | 西北工业大学 | Underwater Acoustic channels method in friction welding (FW) welding process |
| CN106166071A (en) * | 2016-07-04 | 2016-11-30 | 中国科学院计算技术研究所 | The acquisition method of a kind of gait parameter and equipment |
| CN106175778A (en) * | 2016-07-04 | 2016-12-07 | 中国科学院计算技术研究所 | A kind of method setting up gait data collection and gait analysis method |
| CN106531186A (en) * | 2016-10-28 | 2017-03-22 | 中国科学院计算技术研究所 | Footstep detecting method according to acceleration and audio information |
-
2017
- 2017-04-14 CN CN201710242995.6A patent/CN107170466B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002197437A (en) * | 2000-12-27 | 2002-07-12 | Sony Corp | Walking detection system, walking detector, device and walking detecting method |
| US20020107649A1 (en) * | 2000-12-27 | 2002-08-08 | Kiyoaki Takiguchi | Gait detection system, gait detection apparatus, device, and gait detection method |
| US20050094795A1 (en) * | 2003-10-29 | 2005-05-05 | Broadcom Corporation | High quality audio conferencing with adaptive beamforming |
| CN101393660A (en) * | 2008-10-15 | 2009-03-25 | 中山大学 | Intelligent gate inhibition system based on footstep recognition |
| CN102799899A (en) * | 2012-06-29 | 2012-11-28 | 北京理工大学 | Special audio event layered and generalized identification method based on SVM (Support Vector Machine) and GMM (Gaussian Mixture Model) |
| CN103730124A (en) * | 2013-12-31 | 2014-04-16 | 上海交通大学无锡研究院 | Noise robustness endpoint detection method based on likelihood ratio test |
| CN104200815A (en) * | 2014-07-16 | 2014-12-10 | 电子科技大学 | Audio noise real-time detection method based on correlation analysis |
| CN105215542A (en) * | 2015-10-14 | 2016-01-06 | 西北工业大学 | Underwater Acoustic channels method in friction welding (FW) welding process |
| CN106166071A (en) * | 2016-07-04 | 2016-11-30 | 中国科学院计算技术研究所 | The acquisition method of a kind of gait parameter and equipment |
| CN106175778A (en) * | 2016-07-04 | 2016-12-07 | 中国科学院计算技术研究所 | A kind of method setting up gait data collection and gait analysis method |
| CN106531186A (en) * | 2016-10-28 | 2017-03-22 | 中国科学院计算技术研究所 | Footstep detecting method according to acceleration and audio information |
Non-Patent Citations (3)
| Title |
|---|
| BIN SHE: ""Framework of footstep detection in door environment"", 《ICA》 * |
| GEORGE I ZDRAGKAS: ""Gait Analysis and automatic Gait Event Identification using Acceleraters"", 《RESEARCHGATE》 * |
| 王光旭: ""基于表面肌电信号的下肢运动模式识别的研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110364141A (en) * | 2019-06-04 | 2019-10-22 | 杭州电子科技大学 | Elevator typical case's abnormal sound alarm method based on depth single classifier |
| GB2607561A (en) * | 2021-04-08 | 2022-12-14 | Miicare Ltd | Mobility analysis |
| GB2607561B (en) * | 2021-04-08 | 2023-07-19 | Miicare Ltd | Mobility analysis |
| CN113679300A (en) * | 2021-09-13 | 2021-11-23 | 李侃 | Electronic book recommendation method and device for floor sweeping robot |
| CN114550395A (en) * | 2022-04-28 | 2022-05-27 | 常州分音塔科技有限公司 | Sound alarm detection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107170466B (en) | 2020-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107170466A (en) | The sound detection method that mops floor based on audio | |
| CN106166071B (en) | A kind of acquisition method and equipment of gait parameter | |
| CN106531186B (en) | Merge the step detection method of acceleration and audio-frequency information | |
| CN104434128B (en) | A kind of left and right based on plantar nervous arch information foot dynamic identifying method | |
| CN106175778A (en) | A kind of method setting up gait data collection and gait analysis method | |
| CN106456024A (en) | Resting heart rate monitor system | |
| US20240341696A1 (en) | Apparatus for Measuring Degree of Frailty on Basis of Athletic Ability Parameters | |
| CN106419938A (en) | Attention deficit hyperactivity disorder (ADHD) detection method and system based on kinetic energy release estimation | |
| JP2014042757A (en) | Electronic equipment, method for determining swimming form, and swimming form determination program | |
| CN106203484A (en) | A kind of human motion state sorting technique based on classification layering | |
| CN106653058B (en) | Dual-track-based step detection method | |
| CN107480637B (en) | Heart failure based on heart sound feature method by stages | |
| US10426426B2 (en) | Methods and apparatus for performing dynamic respiratory classification and tracking | |
| CN110384503A (en) | Automate the suitable energy test method of body and its system | |
| CN109223006A (en) | A kind of schizophrenia diagnosis system | |
| Martinez-Tabares et al. | Improvement of ECG signal quality measurement using correlation and diversity-based approaches | |
| CN113456060B (en) | Extraction device for motion function characteristic parameters | |
| WO2019001040A1 (en) | Lung cancer detection method and system for asian population | |
| CN113095379A (en) | Human motion state identification method based on wearable six-axis sensing data | |
| CN112914544A (en) | Visceral fat measuring method, device, equipment and storage medium | |
| JP2021074238A (en) | Measuring device and program | |
| Hamidi Rad et al. | Monitoring weekly progress of front crawl swimmers using IMU-based performance evaluation goal metrics | |
| KR101310500B1 (en) | Apparatus and method for determining health using skin features, apparatus and method for generating health sort function | |
| Yadav et al. | Role of breath phase and breath boundaries for the classification between asthmatic and healthy subjects | |
| Zaeni et al. | Detection of the Imbalance Step Length using the Decision Tree |
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 | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20170915 Assignee: Beijing Zhongke Huicheng Technology Co., Ltd. Assignor: Institute of Computing Technology, Chinese Academy of Sciences Contract record no.: 2018110000005 Denomination of invention: Shuffling sound detection method based on voice frequency License type: Common License Record date: 20180222 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EC01 | Cancellation of recordation of patent licensing contract |
Assignee: Beijing Zhongke Huicheng Technology Co., Ltd. Assignor: Institute of Computing Technology, Chinese Academy of Sciences Contract record no.: 2018110000005 Date of cancellation: 20180309 |
|
| EC01 | Cancellation of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20170915 Assignee: Luoyang Zhongke Huicheng Technology Co., Ltd. Assignor: Institute of Computing Technology, Chinese Academy of Sciences Contract record no.: 2018110000009 Denomination of invention: Shuffling sound detection method based on voice frequency License type: Common License Record date: 20180319 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |