CN106691382B - Snore detection method and device based on time-frequency similarity - Google Patents

Snore detection method and device based on time-frequency similarity Download PDF

Info

Publication number
CN106691382B
CN106691382B CN201611220380.5A CN201611220380A CN106691382B CN 106691382 B CN106691382 B CN 106691382B CN 201611220380 A CN201611220380 A CN 201611220380A CN 106691382 B CN106691382 B CN 106691382B
Authority
CN
China
Prior art keywords
segment
snore
time
voiced
sound
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.)
Active
Application number
CN201611220380.5A
Other languages
Chinese (zh)
Other versions
CN106691382A (en
Inventor
竹东翔
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sebiong Technologies Beijing Co ltd
Original Assignee
Sebiong Technologies Beijing Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Sebiong Technologies Beijing Co ltd filed Critical Sebiong Technologies Beijing Co ltd
Priority to CN201611220380.5A priority Critical patent/CN106691382B/en
Publication of CN106691382A publication Critical patent/CN106691382A/en
Application granted granted Critical
Publication of CN106691382B publication Critical patent/CN106691382B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/48Other medical applications
    • A61B5/4806Sleep evaluation
    • A61B5/4818Sleep apnoea

Landscapes

  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Physics & Mathematics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

The invention discloses a snore detecting method and a snore detecting device based on time-frequency similarity, wherein the method comprises the following steps: acquiring a collected snore audio signal, and performing noise estimation on the snore audio signal and then performing sound fragment extraction by using an endpoint detection technology; and calculating the Euclidean distance between the time domain energy or the frequency domain energy of the current voiced segment and the last voiced segment, comparing the Euclidean distance with a threshold value, and judging the snore state according to the comparison result. The snore detecting method is characterized in that the average frequency spectrum energy and the time domain energy of the sound segments are used, the time-frequency similarity of two adjacent sound segments is used for detecting the snore, the operation cost is reduced, and the requirements of hardware realization and real-time detection are met; meanwhile, the short-time change of the snore is ignored, the snore is detected from the periodicity of the snore in a long time, good marking of snore data is not needed, and most snore situations can be adapted.

Description

Snore detection method and device based on time-frequency similarity
Technical Field
The invention belongs to the technical field of sound signal detection, and particularly relates to a snore detecting method and device based on time-frequency similarity.
Background
Obstructive apnea syndrome (OSAHS) occurs at a rate of about 3% to 4%, and is clinically referred to as "snoring". OSAHS seriously affects the life quality of patients, is easy to cause the decline of learning and memory ability, low working efficiency, frequent traffic accidents and the disorder of endocrine and endothelial systems, and is easy to cause sudden death caused by night respiratory insufficiency and cardiovascular and cerebrovascular diseases under extreme conditions. Therefore, the detection of the snore signal has very important significance for tracking the sleep process and judging the sleep quality.
The existing snore signal detection generally utilizes a polysomnography monitoring system, and the method has complex and expensive equipment, and the 'invasive' monitoring is easy to cause physical discomfort. Among the simple OSAHS diagnostic methods, many related studies are made on detection by using snore signals. In addition, when the existing snore detecting technology processes the recording data in the real scene, the following defects exist: the method depends on the effectiveness of the features and the reliability of a classification model, and the extraction process of the common features in the prior art is very complex, the operation cost is high, and the requirements of hardware realization and real-time detection are difficult to meet; the reliability of the classification model depends on a large amount of snore data with good labeling information, which is difficult to obtain in a real scene; secondly, the method for extracting the features frame by frame ignores the periodicity of the snore in a longer time span, and the effectiveness of the method is difficult to generalize to most snores.
Disclosure of Invention
In order to solve the technical defects, the snore detection method directly uses the average frequency spectrum energy and the time domain energy in the sound segments as the characteristics, uses the time-frequency similarity of two adjacent sound segments to detect the snore, reduces the operation cost, and easily meets the requirements of hardware implementation and real-time detection; meanwhile, the invention ignores the short-time change of the snore, and starts to detect the snore periodically in a long time, does not need to label the snore data well, and can adapt to most snore situations.
The invention provides a snore detecting method based on time-frequency similarity, which comprises the following steps:
acquiring a collected snore audio signal, and performing noise estimation on the snore audio signal and then performing sound fragment extraction by using an endpoint detection technology;
and calculating the Euclidean distance between the time domain energy or the frequency domain energy of the current voiced segment and the last voiced segment, comparing the Euclidean distance with a threshold value, and judging the snore state according to the comparison result.
Further, the step of acquiring the collected snore audio signals, and performing noise estimation on the snore audio signals and then extracting the sound fragments by using an endpoint detection technology comprises
Obtaining a starting point of the snore audio signal;
and judging whether the starting point has a sound signal, if so, taking the starting time as a sound judging starting point, judging whether the duration time of the sound segment meets a soundless judging condition, if so, acquiring the duration time of the soundless segment, and otherwise, acquiring the end point of the sound segment.
Further, the step of acquiring the collected snore audio signals, performing noise estimation on the snore audio signals and then extracting the sound fragments by using an endpoint detection technology further comprises the step of
If the starting end point has no sound signal, the starting time is used as a soundless judgment starting point, whether the soundless segment meets the soundless judgment condition or not is judged, if yes, the duration time of the soundless segment is obtained, and if not, the end point of the wireless segment is obtained.
Further, the step of calculating the Euclidean distance between the time domain energy or the frequency domain energy of the current voiced segment and the last voiced segment, comparing the Euclidean distance with a threshold value, and judging the snore state according to the comparison result comprises
Obtaining the duration of each voiced segment, calculating the evaluation duration of all voiced segments, and then calculating the average frequency domain energy of the voiced segments in the average time;
arranging the voiced segments according to the time sequence to obtain a time domain energy sequence of the corresponding voiced segments;
and calculating the Euclidean distance between the current voiced segment and the time domain energy of the front voiced segment and the rear voiced segment, comparing the Euclidean distance with a threshold value, judging whether the Euclidean distance between the current voiced segment and the front voiced segment and the rear voiced segment is not less than the threshold value, and judging that the current voiced segment is the snore segment if the Euclidean distance is not less than the threshold value.
The invention also provides a snore detecting device based on time-frequency similarity, which comprises
The segment extraction module is used for acquiring the collected snore audio signals, and performing noise estimation on the snore audio signals and then performing sound segment extraction by using an endpoint detection technology;
the snore judging module is used for calculating the Euclidean distance between the time domain energy or the frequency domain energy of the current sound segment and the last sound segment, comparing the Euclidean distance with a threshold value, and judging the snore state according to the comparison result.
Further, the fragment extraction module comprises
The starting submodule is used for acquiring a starting point of the snore audio signal;
the segment judgment submodule is used for judging whether the starting point has a sound signal or not, if so, the starting time is used as the sound judgment starting point, and whether the duration time of the sound segment meets the soundless judgment condition or not is judged, if so, the duration time of the soundless segment is obtained, and if not, the ending point of the sound segment is obtained;
or,
if the starting end point has no sound signal, the starting time is used as a soundless judgment starting point, whether the soundless segment meets the soundless judgment condition or not is judged, if yes, the duration time of the soundless segment is obtained, and if not, the end point of the wireless segment is obtained.
Further, the snore judging module comprises
The frequency domain characteristic extraction unit is used for acquiring the duration of each voiced segment, calculating the evaluation duration of all the voiced segments, and then calculating the average frequency domain energy of the voiced segments in the average time;
the time domain feature extraction unit is used for arranging the voiced segments according to the time sequence to obtain the time domain energy sequence of the corresponding voiced segments;
and the judging unit is used for calculating the Euclidean distance between the current voiced segment and the time domain energy of the front voiced segment and the rear voiced segment, comparing the Euclidean distance with a threshold value, judging whether the Euclidean distance between the current voiced segment and the front voiced segment is not less than the threshold value or not, and judging that the current voiced segment is the snore segment if the Euclidean distance is not less than the threshold value.
In summary, the snore detection method is characterized in that the average frequency spectrum energy and the time domain energy of the sound segments are used, the time-frequency similarity of two adjacent sound segments is used for snore detection, the operation cost is reduced, and the requirements of hardware implementation and real-time detection are met; meanwhile, the short-time change of the snore is ignored, the snore is detected from the periodicity of the snore in a long time, good marking of snore data is not needed, and most snore situations can be adapted.
Drawings
In order to more clearly illustrate the embodiments of the present application or technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present invention, and other drawings can be obtained by those skilled in the art according to the drawings.
FIG. 1 is a schematic flow chart of a snore detecting method based on time-frequency similarity according to the present invention;
FIG. 2 is a schematic flow chart of the judgment of the sound segment and the soundless segment in the snore detecting method based on time-frequency similarity according to the present invention;
fig. 3 is a block diagram structure diagram of the snore detecting device based on time-frequency similarity according to the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, those skilled in the art will now describe the present invention in further detail with reference to the accompanying drawings.
The present invention will be described in further detail below with reference to specific embodiments and with reference to the attached drawings.
As shown in fig. 1, a snore detecting method based on time-frequency similarity includes the following steps:
s101, acquiring a collected snore audio signal, and performing noise estimation on the snore audio signal and then performing sound segment extraction by using an endpoint detection technology;
the invention uses the optimal filter and the minimum statistic method to estimate the noise of the snore audio signal, and in the specific implementation, the snore audio signal only comprises the snore signal P and the noise N, and the snore signal P is close to 0 in the snore inhaling process, so that the snore signal recursion can be expressed as the following formula:
P(t)=αP(t-1)+(1-α)(P(t)+N(t))
wherein α is a smoothing factor of the audio signal, and a new α is estimated by using a minimum statistic method over time, and then the noise strength N is estimated by using the above formula, and the calculation formula of α is as follows:
Figure GDA0002590251520000041
the invention uses a Bayes model based on signal-to-noise ratio to detect the end point of the snore audio signal, the noise intensity is obtained by the noise estimation method, the signal intensity can directly count the amplitude of the snore audio signal, and finally, the calculation formula for verifying the signal-to-noise ratio is as follows:
Figure GDA0002590251520000042
estimation of prior signal-to-noise ratio during endpoint detectionSmoothing was performed with α ═ 0.96, assuming voiced segments are H1The silent segment is H0And the calculation formula of the endpoint detection is as follows, when the calculation result of the current formula is more than 4 and the posterior signal-to-noise ratio is more than 5, the voice is judged, and otherwise, the voice is judged to be silent.
Figure GDA0002590251520000051
S102, calculating the Euclidean distance between the time domain energy or the average frequency domain energy of the current voiced segment and the last voiced segment, comparing the Euclidean distance with a threshold value, and judging the snore state according to the comparison result.
In the sound segment extraction module of the snore audio signal, the length of the extracted sound segment is at least more than 0.5 second, and the sound segment is insensitive to the slight change of the snore in the sound segment, but the average time domain and frequency domain energy is directly used for similarity calculation, so that the frequency spectrum details of each frame of sound signal are not required to be stored, the relevant properties of the signal are not required to be analyzed frame by frame, the storage and time consumption are greatly reduced, the snore state judgment cost is further reduced, and the snore state judgment method is suitable for being operated in a terminal with limited low power consumption resources. Meanwhile, the traditional snore detection technology based on classification needs a large amount of snore data with good labeled information to perform supervised learning, the data is difficult to obtain and does not have good generalization capability, and the snore state judgment is performed by using the similarity of the sound fragments, so that the dependence on labeled data is avoided, and the data acquisition cost is reduced.
As shown in fig. 2, further S101 includes
Obtaining a starting point of the snore audio signal;
and judging whether the starting point has a sound signal, if so, taking the starting time as a sound judging starting point, judging whether the duration time of the sound segment meets a soundless judging condition, if so, acquiring the duration time of the soundless segment, and otherwise, acquiring the end point of the sound segment. Further, S101 further includes taking the starting time as a silent judgment starting point if the starting end point has no sound signal, and judging whether a sound judgment condition is satisfied in a silent segment, if so, acquiring the duration of the sound segment, otherwise, acquiring the end point of the wireless segment.
In specific implementation, the sound judging condition optionally includes whether to mute or not and whether the length of the sound fragment is smaller than the preset length or not; the silence judgment condition optionally comprises whether sound exists and the length of the silence segment is less than a threshold value; in specific implementation, the length of the voiced segment (i.e., the shortest time) is set to 0.5 second, and the length of the unvoiced segment (i.e., the shortest duration) is set to 0.05 second.
The invention firstly carries out noise estimation on the snore audio signal, utilizes an end point detection technology to detect the snore audio signal, namely, the end point detection of a starting signal judges whether the starting point is voiced, if so, the starting point is determined to be the starting point of the voiced segment, otherwise, the starting point of the unvoiced segment is determined to be the starting point of the unvoiced segment, and then whether the unvoiced segment and the voiced segment are included in the unvoiced segment duration and the voiced segment duration or not is respectively judged according to the voiced judgment condition and the unvoiced judgment condition, thereby obtaining the end point of the voiced or invisible segment. In specific implementation, the judgment condition for analyzing the snore audio signal is whether the end point detection result of the current frame is mute or voiced, and whether the duration length of the current voiced segment or unvoiced segment exceeds a threshold value. The method aims to judge whether the sound segment is the snore segment or not, so that if the sound state of the current frame is judged to be 'sound segment starting' or 'sound segment continuing', the time-frequency energy is updated on the current sound segment to calculate the similarity; and if the sound state of the current frame is judged to be 'silent segment continuous' and the length of the silent segment exceeds the threshold value, the snore detection is carried out on the previous sound segment.
Further, S102 includes
Obtaining the duration of each voiced segment, calculating the evaluation duration of all voiced segments, and then calculating the average frequency domain energy of the voiced segments in the average time;
arranging the voiced segments according to the time sequence to obtain a time domain energy sequence of the corresponding voiced segments;
and calculating the Euclidean distance between the current voiced segment and the time domain energy of the front voiced segment and the rear voiced segment, comparing the Euclidean distance with a threshold value, judging whether the Euclidean distance between the current voiced segment and the front voiced segment and the rear voiced segment is not less than the threshold value, and judging that the current voiced segment is the snore segment if the Euclidean distance is not less than the threshold value.
As shown in fig. 3, the invention further provides a snore detecting device based on time-frequency similarity, which comprises a segment extracting module 10 and a snore judging module 20.
Wherein,
the segment extraction module 10 is configured to acquire the collected snore audio signals, perform noise estimation on the snore audio signals, and then perform voiced segment extraction by using an endpoint detection technology; in specific implementation, the snore audio signal can be selected by using a mobile device (such as a mobile phone, a pad or a notebook), a special sleep medical device or a sound collecting device and other devices to collect the sleep sound of the human body. The invention utilizes the conventional equipment to collect the sound, thereby reducing the cost of sound collection, and simultaneously utilizes the fragment extraction module to extract the sound fragments of the collected snore audio signals, thereby avoiding the cost of using a special polysomnography monitoring system.
The snore judging module 20 is configured to calculate a euclidean distance between the time domain energy or the frequency domain energy of the current voiced segment and the previous voiced segment, compare the euclidean distance with a threshold, and judge the snore state according to the comparison result.
Further, the fragment extraction module comprises
The starting submodule is used for acquiring a starting point of the snore audio signal;
the segment judgment submodule is used for judging whether the starting point has a sound signal or not, if so, the starting time is used as the sound judgment starting point, and whether the duration time of the sound segment meets the soundless judgment condition or not is judged, if so, the duration time of the soundless segment is obtained, and if not, the ending point of the sound segment is obtained; or, if the starting end point has no sound signal, the starting time is used as a soundless judgment starting point, whether the soundless segment meets the soundless judgment condition or not is judged, if yes, the duration time of the soundless segment is obtained, and if not, the end point of the wireless segment is obtained.
Further, the snore judging module comprises
The frequency domain characteristic extraction unit is used for acquiring the duration of each voiced segment, calculating the evaluation duration of all the voiced segments, and then calculating the average frequency domain energy of the voiced segments in the average time;
the time domain feature extraction unit is used for arranging the voiced segments according to the time sequence to obtain the time domain energy sequence of the corresponding voiced segments;
and the judging unit is used for calculating the Euclidean distance between the current voiced segment and the time domain energy of the front voiced segment and the rear voiced segment, comparing the Euclidean distance with a threshold value, judging whether the Euclidean distance between the current voiced segment and the front voiced segment is not less than the threshold value or not, and judging that the current voiced segment is the snore segment if the Euclidean distance is not less than the threshold value.
While certain exemplary embodiments of the present invention have been described above by way of illustration only, it will be apparent to those of ordinary skill in the art that the described embodiments may be modified in various different ways without departing from the spirit and scope of the invention. Accordingly, the drawings and description are illustrative in nature and should not be construed as limiting the scope of the invention.

Claims (3)

1. A snore detecting method based on time-frequency similarity is characterized by comprising the following steps:
acquiring a collected snore audio signal, performing noise estimation on the snore audio signal, and then performing sound segment extraction by using an endpoint detection technology, wherein the noise estimation of the snore audio signal is performed by using an optimal filter and a minimum statistic method;
calculating the Euclidean distance between the time domain energy or the frequency domain energy of the current voiced segment and the last voiced segment, comparing the Euclidean distance with a threshold value, and judging the snore state according to the comparison result;
the steps of acquiring the collected snore audio signals, performing noise estimation on the snore audio signals and then extracting the sound fragments by utilizing an endpoint detection technology comprise
Obtaining a starting point of the snore audio signal;
judging whether the starting point has a sound signal, if so, taking the starting time as a sound judgment starting point, and judging whether the duration time of the sound segment meets a soundless judgment condition, if so, acquiring the duration time of the soundless segment, otherwise, acquiring the end point of the sound segment;
calculating the Euclidean distance between the time domain energy or the frequency domain energy of the current voiced segment and the last voiced segment, comparing the Euclidean distance with a threshold value, and judging the snore state according to the comparison result
Obtaining the duration of each voiced segment, calculating the evaluation duration of all voiced segments, and then calculating the average frequency domain energy of the voiced segments in the average time;
arranging the voiced segments according to the time sequence to obtain a time domain energy sequence of the corresponding voiced segments;
and calculating the Euclidean distance between the current voiced segment and the time domain energy of the front voiced segment and the rear voiced segment, comparing the Euclidean distance with a threshold value, judging whether the Euclidean distance between the current voiced segment and the front voiced segment and the rear voiced segment is not less than the threshold value, and judging that the current voiced segment is the snore segment if the Euclidean distance is not less than the threshold value.
2. The time-frequency similarity-based snore detecting method according to claim 1, wherein the acquiring of the collected snore audio signals, the noise estimation of the snore audio signals and the extraction of the sound segments by using an endpoint detection technology further comprise
If the starting end point has no sound signal, the starting time is used as a soundless judgment starting point, whether the soundless segment meets the soundless judgment condition or not is judged, if yes, the duration time of the soundless segment is obtained, and if not, the end point of the wireless segment is obtained.
3. A snore detecting device based on time-frequency similarity is characterized by comprising
The segment extraction module is used for acquiring the collected snore audio signals, performing noise estimation on the snore audio signals and then performing sound segment extraction by using an endpoint detection technology, wherein the noise estimation of the snore audio signals is performed by using an optimal filter and a minimum statistic method;
the snore judging module is used for calculating the Euclidean distance between the time domain energy or the frequency domain energy of the current sound segment and the last sound segment, comparing the Euclidean distance with a threshold value, and judging the snore state according to the comparison result;
the fragment extraction module comprises
The starting submodule is used for acquiring a starting point of the snore audio signal;
the segment judgment submodule is used for judging whether the starting point has a sound signal or not, if so, the starting time is used as the sound judgment starting point, and whether the duration time of the sound segment meets the soundless judgment condition or not is judged, if so, the duration time of the soundless segment is obtained, and if not, the ending point of the sound segment is obtained;
or,
if the starting end point has no sound signal, the starting time is used as a soundless judgment starting point, whether the soundless segment meets the soundless judgment condition or not is judged, if yes, the duration time of the soundless segment is obtained, and if not, the end point of the wireless segment is obtained;
the snore judging module comprises
The frequency domain characteristic extraction unit is used for acquiring the duration of each voiced segment, calculating the evaluation duration of all the voiced segments, and then calculating the average frequency domain energy of the voiced segments in the average time;
the time domain feature extraction unit is used for arranging the voiced segments according to the time sequence to obtain the time domain energy sequence of the corresponding voiced segments;
and the judging unit is used for calculating the Euclidean distance between the current voiced segment and the time domain energy of the front voiced segment and the rear voiced segment, comparing the Euclidean distance with a threshold value, judging whether the Euclidean distance between the current voiced segment and the front voiced segment is not less than the threshold value or not, and judging that the current voiced segment is the snore segment if the Euclidean distance is not less than the threshold value.
CN201611220380.5A 2016-12-26 2016-12-26 Snore detection method and device based on time-frequency similarity Active CN106691382B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201611220380.5A CN106691382B (en) 2016-12-26 2016-12-26 Snore detection method and device based on time-frequency similarity

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201611220380.5A CN106691382B (en) 2016-12-26 2016-12-26 Snore detection method and device based on time-frequency similarity

Publications (2)

Publication Number Publication Date
CN106691382A CN106691382A (en) 2017-05-24
CN106691382B true CN106691382B (en) 2020-12-15

Family

ID=58903444

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201611220380.5A Active CN106691382B (en) 2016-12-26 2016-12-26 Snore detection method and device based on time-frequency similarity

Country Status (1)

Country Link
CN (1) CN106691382B (en)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107374223A (en) * 2017-08-28 2017-11-24 赛博龙科技(北京)有限公司 A kind of closed-loop system, device and intelligent pillow for detecting, analyzing sleep
WO2019127088A1 (en) * 2017-12-27 2019-07-04 深圳和而泰数据资源与云技术有限公司 Snore recognition method and snore-stopping device
CN108938175B (en) * 2018-07-25 2024-02-09 深圳市华信物联传感技术有限公司 Snore relieving system and method
CN108917086B (en) * 2018-07-28 2020-07-31 奥克斯空调股份有限公司 Sleep mode automatic control method and air conditioner
CN109745011B (en) * 2019-02-20 2021-12-03 华为终端有限公司 User sleep respiration risk monitoring method, terminal and computer readable medium
CN111603335A (en) * 2019-02-25 2020-09-01 胡福民 Sleep posture adjustment system
CN111685735B (en) * 2020-06-08 2022-01-21 清华大学 Sleep apnea monitoring method and device based on snore signals
CN111933181B (en) * 2020-07-10 2022-05-17 北京理工大学 Snore feature extraction and detection method and device based on complex order derivative processing
CN113314143B (en) * 2021-06-07 2024-01-30 南京优博一创智能科技有限公司 Method and device for judging apnea and electronic equipment
CN113421586B (en) * 2021-06-18 2024-05-07 南京优博一创智能科技有限公司 Dream recognition method and device and electronic equipment
CN114027801B (en) * 2021-12-17 2022-09-09 广东工业大学 Method and system for recognizing sleep snore and restraining snore
CN115206329B (en) * 2022-09-15 2023-01-24 深圳市倍轻松科技股份有限公司 Method, device, electronic equipment and storage medium for determining snore signals

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102429662A (en) * 2011-11-10 2012-05-02 大连理工大学 Screening system for sleep apnea syndrome in home environment
CN102499637A (en) * 2011-09-26 2012-06-20 大连理工大学 Obstructive sleep apnea-hypopnea syndrome screening method and device thereof
US20140350426A1 (en) * 2009-09-14 2014-11-27 Sleep Methods, Inc. System and method for anticipating the onset of an obstructive sleep apnea event
US20140358189A1 (en) * 2009-10-20 2014-12-04 Adi Mashiach Device and Method for Snoring Detection and Control
US20140371821A1 (en) * 2013-06-17 2014-12-18 Adi Mashiach Residual Signal Feedback-Based Control of an Implant
CN104739412A (en) * 2013-12-29 2015-07-01 中国移动通信集团公司 Method and equipment for monitoring sleep apnea
US20160150993A1 (en) * 2014-02-28 2016-06-02 Powell Mansfield, Inc. Systems, methods and devices for sensing emg activity
CN105662417A (en) * 2015-12-31 2016-06-15 沈阳迈思医疗科技有限公司 Control method and device for identifying snores based on pressure signal characteristics
CN106128458A (en) * 2016-08-31 2016-11-16 成都市和平科技有限责任公司 A kind of home voice control system based on speech recognition technology and method

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140350426A1 (en) * 2009-09-14 2014-11-27 Sleep Methods, Inc. System and method for anticipating the onset of an obstructive sleep apnea event
US20140358189A1 (en) * 2009-10-20 2014-12-04 Adi Mashiach Device and Method for Snoring Detection and Control
CN102499637A (en) * 2011-09-26 2012-06-20 大连理工大学 Obstructive sleep apnea-hypopnea syndrome screening method and device thereof
CN102429662A (en) * 2011-11-10 2012-05-02 大连理工大学 Screening system for sleep apnea syndrome in home environment
US20140371821A1 (en) * 2013-06-17 2014-12-18 Adi Mashiach Residual Signal Feedback-Based Control of an Implant
CN104739412A (en) * 2013-12-29 2015-07-01 中国移动通信集团公司 Method and equipment for monitoring sleep apnea
US20160150993A1 (en) * 2014-02-28 2016-06-02 Powell Mansfield, Inc. Systems, methods and devices for sensing emg activity
CN105662417A (en) * 2015-12-31 2016-06-15 沈阳迈思医疗科技有限公司 Control method and device for identifying snores based on pressure signal characteristics
CN106128458A (en) * 2016-08-31 2016-11-16 成都市和平科技有限责任公司 A kind of home voice control system based on speech recognition technology and method

Also Published As

Publication number Publication date
CN106691382A (en) 2017-05-24

Similar Documents

Publication Publication Date Title
CN106691382B (en) Snore detection method and device based on time-frequency similarity
Lu et al. Speakersense: Energy efficient unobtrusive speaker identification on mobile phones
TWI474317B (en) Signal processing apparatus and signal processing method
US11315591B2 (en) Voice activity detection method
CN100505040C (en) Audio frequency splitting method for changing detection based on decision tree and speaking person
US20140067388A1 (en) Robust voice activity detection in adverse environments
WO2011044853A1 (en) Method and device for realizing trace of background noise in communication system
CN102469978B (en) Noise reduction of breathing signals
CN108697328B (en) Snore identification method and snore stopping device
CN103198838A (en) Abnormal sound monitoring method and abnormal sound monitoring device used for embedded system
CN104739412A (en) Method and equipment for monitoring sleep apnea
CN105679312A (en) Phonetic feature processing method of voiceprint identification in noise environment
CN110942784A (en) Snore classification system based on support vector machine
CN104361887A (en) Quick acoustic event detection system under traveling noise environment
CN103730124A (en) Noise robustness endpoint detection method based on likelihood ratio test
WO2014168022A1 (en) Signal processing device, signal processing method, and signal processing program
CN113314143A (en) Apnea judgment method and device and electronic equipment
CN109377982A (en) A kind of efficient voice acquisition methods
CN101656070B (en) Voice detection method
Sarkar et al. Automatic speech segmentation using average level crossing rate information
US20220346705A1 (en) Apparatus and method for sleep-disordered breathing estimation based on sound analysis
JP5765338B2 (en) Voice processing apparatus and method of operating voice processing apparatus
TWI684912B (en) Voice wake-up apparatus and method thereof
CN110047471A (en) Voice awakening method
CN111128244B (en) Short wave communication voice activation detection method based on zero crossing rate detection

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant