CN109166592A - HRTF frequency-division section linear regression method based on physiological parameter - Google Patents
HRTF frequency-division section linear regression method based on physiological parameter Download PDFInfo
- Publication number
- CN109166592A CN109166592A CN201810894053.0A CN201810894053A CN109166592A CN 109166592 A CN109166592 A CN 109166592A CN 201810894053 A CN201810894053 A CN 201810894053A CN 109166592 A CN109166592 A CN 109166592A
- Authority
- CN
- China
- Prior art keywords
- principal component
- physiological parameter
- frequency
- hrtf
- frequency range
- 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
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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
- G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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
Abstract
The present invention provides a kind of HRTF frequency-division section linear regression method based on physiological parameter, the HRTF of different subjects carry out principal component analysis under same orientation in selection CIPIC database first, obtain the principal component under the orientation；Then, the correlations of principal component and the every physiological parameter of different subjects at different frequencies are calculated；The relevant physiological parameter to low frequency, intermediate frequency and high frequency section is obtained by statistical method；The linear fit for carrying out low frequency, intermediate frequency and high band respectively finally by the physiological parameter of selection carries out phase reconstruction by minimum phase method.It is different that the present invention influences different physiological parameters, application relativity analysis, it obtains in different frequency range physiological parameter influential on HRTF, linear regression is carried out using the physiological parameter, the HRTF synthesized using the present invention is theoretically relatively sharp, while also more accurate to the result of each frequency range progress linear regression recurrence respectively.
Description
Technical field
The present invention relates to sound sources to divide message area, especially a kind of HRTF linear regression method based on physiological parameter.
Background technique
Head related transfer function (Head Related Transfer Function, HRTF) is to describe free field sound
The sound that source issues reaches the frequency domain acoustics transmitting of ears after the scattering of the physiological structures such as head, auricle, trunk and reflection
Function.Corresponding a pair of the HRTF in each sound source spatial position, they are the sides of the sound source distance, sound source at center to the end under normal circumstances
Parallactic angle, the elevation angle, the function of frequency.Since the physiological structure and size of Different Individual are different, and HRTF and physiological structure and
Size is closely related, thus is the physical quantity with obvious personalized features.
Influence of the filter action of the physiological structures such as trunk, head and auricle to acoustical signal is more significant, that is to say, that raw
Reason structure, which has HRTF, more to be significantly affected.The physiologic factor for influencing the HRTF of different direction is different, simultaneously for same
HRTF, in different frequency scope, influence factor is not also identical.Research shows that helix outer circle plays in high band localization of sound
Conclusive effect；Helix outer circle is responsible for guiding the sound above ear, and nest is then responsible for guidance from ear in helix
The sound of lower section.But, the structure feature of ear only plays a leading role to voice signal in high band (being greater than 5kHz).Psychological sound
It is learning studies have shown that the scattering on head and the longitudinal register for being reflected into low-frequency range (lower than 3kHz) sound of trunk provide
Clue.
Document " Martens W L.Principal components analysis and resynthesis of
spectral cues to perceived direction.In:Proceedings of 1987 International
Computer Music Conference,1987,274-281".Announce a kind of utilization human figure feature enemy's associated delivery
Function carries out personalized method, i.e., using the correlation enemy related transfer function between basic function and human body measurements into
Row approximate evaluation.In this article, physiological parameter needed for author has selected building model by experience." head correlation passes document
The personalization method of delivery function ".The method of physiological parameter needed for disclosing the new building model of one kind.This method pass through by
All physiological parameters that CIPIC database provides carry out correlation analysis, choose incoherent physiological parameter as building model
Required physiological parameter.The physiological parameter being gained through experience passes through correlation point theoretically without persuasion property by comparison
The physiological parameter that analysis obtains theoretically have it is certain it is theoretical rely on, however the correlation analysis only considered physiological parameter it
Between correlation, without considering the correlation between physiological parameter and HRTF.Simultaneously as the physiology influenced on different frequency range
Parameter is different, and it is also undesirable to carry out full frequency band linear regression by physiological parameter, therefore in order to preferably physiological parameter be applied to carry out
HRTF reconstruct.Frequency-division section linear regression is carried out to HRTF using frequency dividing phase method.
Summary of the invention
For overcome the deficiencies in the prior art, the present invention provides a kind of HRTF frequency-division section linear regression based on physiological parameter
Method.
The step of the technical solution adopted by the present invention to solve the technical problems, is as follows:
Step 1: all subjects are chosen in the HRTF in a certain determining orientation, and principal component analysis is carried out to the HRTF in the orientation,
Obtain the principal component in the orientation:
(1) choosing all subjects is being the HRTF composition vector D of (θ, φ) at azimuthij, wherein i is subject sequence, j
For frequency serial number, subject number is m, and wherein θ is azimuth, and φ is pitch angle；
(2) to DijCarry out following standardization:
Wherein,For DijHead phase by standardization
Close transmission function；
(3) principal component analysis is carried out to the head related transfer function after standardization:
Wherein, Pm×nIt is the score matrix of m × n of principal component analysis；W is the matrix of loadings of n × n of principal component analysis, T
The transposition of representing matrix；
Step 2: by all physiological parameters of subjects all in CIPIC database, respectively with obtained in step 1 it is main at
Divide and carry out correlation analysis, obtains correlation coefficient rik, rikIndicate the phase relation of i-th of human body physiological parameter and k-th of principal component
Number, i.e. i-th subject sequence of i-th of human body physiological parameter；
Step 3: determining different frequency range related physiological parameters；
To any one principal components, correlation coefficient charts are inquired, the related coefficient of the principal component Yu all physiological parameters is obtained,
The related coefficient of the relevant physiological parameter of different frequency is arranged according to descending, selection is greater than preceding t significant works of related coefficient
For the candidate items of k-th of principal component；
(1) any one candidate physiological parameter is selected, calculates separately candidate's physiological parameter in the different degree of different frequency range
I:
Wherein, NnkIndicate r-th of physiological parameter in time for the appearance that the frequency range f of k-th of principal component is the section a~b
Number, in the section a~b, which occurs once adding 1, does not occur then adding 0, αkIt is total for kth Principal Component Explanation initial data
The percentage of variance, wherein 1 and c refers to which principal component, i.e. first principal component to c principal component；
(2) descending arrangement is carried out by different degree to the physiological parameter in the candidate items in different frequency respectively, before selection
The q major physiological parameter as the frequency range under the orientation；
(3) common physiology number of parameters in side frequency is more than total raw by the difference according to the physiological parameter for influencing HRTF
50% frequency-portions of reason parameter are divided into a frequency range, if a certain frequency is unable to satisfy common physiology parameter in side frequency
Quantity is more than the 50% of total physiological parameter, then is divided into side frequency in the lower frequency range of frequency, by obtaining for principal component
Sub-matrix P is divided according to frequency range and is carried out segment processing, is divided into d Matrix dividing:
P=[P1, P2 ..., Pd] (4)
Step 4: approximate, the approximate principal component scores square of acquisition is carried out to P using the major physiological parameter of each frequency range of selection
Battle arrayThe reconstruct data of head related transfer function are obtained using the approximate score vector of each frequency range, detailed step is as follows:
(1) score vector in principal component analysis passes through key physiological parameters linear regression, regression equation formula (5) table
Show:
Wherein,Indicate d-th of Matrix dividing in principal component,For the physiological parameter square selected from step 3
Battle array；Ad(q+1)×nFor the coefficient of human body measurements；M is subject number；N indicates different frequencies；Ad indicates the people at frequency range d
The coefficient matrix of bulk measurement project；
(2) according to principle of least square method, coefficient matrices A i is obtained by formula (2) and formula (5):
Head related transfer function is standardized to obtain by formula (6):
Wherein,Indicate standardization head related transfer functionApproximation；
When knowing the physiological parameter of subject, the approximate HRTF of the subject is obtained by formula (7).
The beneficial effects of the present invention are the physiological parameter for for HRTF, influencing different is different, application relativity
Analysis, obtains in different frequency range physiological parameter influential on HRTF, carries out linear regression using the physiological parameter.Using the party
Method synthesis HRTF it is theoretically relatively sharp, while respectively to each frequency range carry out linear regression recurrence result also more subject to
Really.
Detailed description of the invention
Fig. 1 is that the present invention is based on the HRTF frequency-division section linear regression flow charts of physiological parameter.
Fig. 2 is orientation of the present invention (0 °, 0 °) principal component scores figure.
Fig. 3 is orientation of the present invention (0 °, 0 °) first principal component and physiological parameter dependency graph, and wherein Fig. 3 (a) is left ear the
One principal component and physiological parameter dependency graph, Fig. 3 (b) are auris dextra first principal component and physiological parameter dependency graph.
Fig. 4 is the present invention is based on the HRTF frequency-division section linear regression of physiological parameter as a result, wherein Fig. 4 (a) is left ear frequency spectrum
Regression result, Fig. 4 (b) are auris dextra frequency spectrum regression result.
Specific embodiment
Present invention will be further explained below with reference to the attached drawings and examples.
Since the physiological parameter being gained through experience inherently lacks persuasion property, obtained by comparison by correlation analysis
Physiological parameter theoretically have it is certain it is theoretical rely on, however the correlation analysis only considered the phase between physiological parameter
Guan Xing, without considering the correlation between physiological parameter and HRTF.Simultaneously as not on the physiological parameter of different frequency range influence
Together, also undesirable by physiological parameter progress full frequency band linear regression, therefore in order to preferably apply physiological parameter to carry out HRTF weight
Structure.This patent proposes that a kind of use divides phase method to HRTF progress frequency-division section linear regression.It selects in CIPIC database first
The HRTF of different subjects carries out principal component analysis under same orientation, obtains the principal component under the orientation；Then, it calculates in different frequencies
The correlations of principal component and the every physiological parameter of different subjects under rate；It is obtained by statistical method to low frequency, intermediate frequency and height
The relevant physiological parameter of frequency part；Carry out the line of low frequency, intermediate frequency and high band respectively finally by the physiological parameter of selection
Property fitting, pass through minimum phase method carry out phase reconstruction.
Specific steps of the invention are as follows:
Step 1: all subjects are chosen in the HRTF in a certain determining orientation, and principal component analysis is carried out to the HRTF in the orientation,
Obtain the principal component in the orientation:
(1) choosing all subjects is being the HRTF composition vector D of (θ, φ) at azimuthij, wherein i is subject sequence, j
For frequency serial number, subject number is m, and wherein θ is azimuth, and φ is pitch angle；
(2) to DijCarry out following standardization:
Wherein,For DijHead phase by standardization
Close transmission function；
(3) principal component analysis is carried out to the head related transfer function after standardization:
Wherein, Pm×nIt is the score matrix of m × n of principal component analysis；W is the matrix of loadings of n × n of principal component analysis, T
The transposition of representing matrix；
Step 2: by all physiological parameters of subjects all in CIPIC database, respectively with the master that is obtained in step 1
Ingredient carries out correlation analysis, obtains correlation coefficient rik, rikIndicate that i-th of human body physiological parameter is related to k-th principal component
Coefficient, i.e. i-th subject sequence of i-th of human body physiological parameter；
Step 3: determining different frequency range related physiological parameters；
Physiological parameter due to influencing different frequency range has significant difference, therefore to any one principal component, inquire related coefficient
Table obtains the related coefficient of the principal component Yu all physiological parameters, and the related coefficient of the relevant physiological parameter of different frequency is pressed
It is arranged according to descending, selects the preceding t significant candidate items as k-th of principal component greater than related coefficient；
(1) any one candidate physiological parameter is selected, calculates separately candidate's physiological parameter in the different degree of different frequency range
I:
Wherein, NnkIndicate r-th of physiological parameter in time for the appearance that the frequency range f of k-th of principal component is the section a~b
Number, in the section a~b, which occurs once adding 1, does not occur then adding 0, αkIt is total for kth Principal Component Explanation initial data
The percentage of variance, wherein 1 and c refers to which principal component, i.e. first principal component to c principal component；
(2) descending arrangement is carried out by different degree to the physiological parameter in the candidate items in different frequency respectively, before selection
The q major physiological parameter as the frequency range under the orientation；
(3) common physiology number of parameters in side frequency is more than total raw by the difference according to the physiological parameter for influencing HRTF
50% frequency-portions of reason parameter are divided into a frequency range, if a certain frequency is unable to satisfy common physiology parameter in side frequency
Quantity is more than the 50% of total physiological parameter, then is divided into side frequency in the lower frequency range of frequency, by obtaining for principal component
Sub-matrix P is divided according to frequency range and is carried out segment processing, is divided into d Matrix dividing:
P=[P1, P2 ..., Pd] (4)
Step 4: approximate, the approximate principal component scores square of acquisition is carried out to P using the major physiological parameter of each frequency range of selection
Battle arrayThe reconstruct data of head related transfer function are obtained using the approximate score vector of each frequency range, detailed step is as follows；
(1) score vector in principal component analysis passes through key physiological parameters linear regression, regression equation formula (5) table
Show:
Wherein,Indicate d-th of Matrix dividing in principal component,For the physiological parameter square selected from step 3
Battle array；Ad(q+1)×nFor the coefficient of human body measurements；M is subject number；N indicates different frequencies；Ad indicates the people at frequency range d
The coefficient matrix of bulk measurement project；
(2) according to principle of least square method, coefficient matrices A i is obtained by formula (2) and formula (5):
Head related transfer function is standardized to obtain by formula (6):
Wherein,Indicate standardization head related transfer functionApproximation；
When knowing the physiological parameter of subject, the approximate HRTF of the subject is obtained by formula (7).
Embodiment is as follows:
Step 1: choosing HRTF of all subjects at particular orientation (0 °, 0 °), principal component is carried out to the HRTF in the orientation
Analysis, obtains the principal component in the orientation.
(1) all subjects are chosen and constitute vector D at azimuth for the HRTF of (0 °, 0 °)ij, wherein i is subject sequence, and j is
Frequency serial number, subject number are m.
(2) to DijIt is standardized.
Wherein,Indicate that i-th of subject passes through at frequency j
Cross the head related transfer function of standardization.
(3) principal component analysis is carried out to the head related transfer function after standardization.
Wherein, Pm×nIt is the score matrix of principal component analysis；W is the matrix of loadings of principal component analysis；Referring to Fig. 2, principal component
Analyze percent 90 or more the preceding ten Principal Component Explanation initial data obtained.
Step 2: by all physiological parameters of subjects all in CIPIC database, respectively with the master that is obtained in step 1
Ingredient carries out correlation analysis, obtains correlation coefficient rik。rikIndicate that i-th of human body physiological parameter is related to k-th principal component
Coefficient.First principal component and physiological parameter relative coefficient figure refer to Fig. 3.
Step 3: determining different frequency range related physiological parameters；
Physiological parameter due to influencing different frequency range has significant difference, therefore to any one principal component, inquire related coefficient
Table obtains the related coefficient of the principal component Yu all physiological parameters, and the related coefficient of the relevant physiological parameter of different frequency is pressed
It is arranged according to descending, selects the preceding t significant candidate items as k-th of principal component greater than related coefficient.CIPIC database
Middle subject quantity is 35, and physiological parameter 37 is tieed up totally.Selecting confidence interval in the present invention is 0.05, therefore the correlation of relevant item
Coefficient is selected as 0.32.
(1) a candidate physiological parameter is selected, calculates separately candidate's physiological parameter in the different degree I of different frequency range:
Wherein, NnkIndicate r-th of physiological parameter in time for the appearance that the frequency range f of k-th of principal component is the section a~b
Number.In the section a~b, which occurs once adding 1, does not occur then adding 0, αkIt is total for kth Principal Component Explanation initial data
The percentage of variance, wherein 1 and c refers to which principal component, i.e. first principal component to c principal component.
(2) descending arrangement is carried out by different degree to the physiological parameter in the candidate items in different frequency respectively, before selection
The q major physiological parameter as the frequency range under the orientation；
(3) foundation influences the difference of the major physiological parameter of HRTF, is more than by common physiology number of parameters in side frequency
50% frequency-portions of total physiological parameter are divided into a frequency range, if a certain frequency is unable to satisfy common physiology in side frequency
Number of parameters is more than the 50% of total physiological parameter, then is divided into side frequency in the lower frequency range of frequency.By principal component
Score matrix P carry out segment processing according to frequency range division rule, be divided into d Matrix dividing:
P=[P1, P2 ..., Pd] (4)
It in this example, can be 1-37,38-88 and 89-100 three parts by frequency partition with reference to Fig. 3 (a)；With reference to Fig. 3
It (b), can be 1-37,38-80 and 81-100 three parts by frequency partition.
Step 4: approximate, the approximate principal component scores square of acquisition is carried out to P using the major physiological parameter of each frequency range of selection
Battle arrayThe reconstruct data of head related transfer function are obtained using the approximate score vector of each frequency range；
(1) score vector in principal component analysis passes through key physiological parameters linear regression, regression equation formula (5) table
Show:
Wherein,Indicate d-th of frequency range in principal component,For the physiological parameter matrix selected in step 3；
Ad(q+1)×nFor the coefficient of human body measurements；M is the number of subject；N indicates different frequencies；Ad indicates the human body at frequency range d
The coefficient matrix of measure the item；
(2) according to principle of least square method, coefficient matrices A d is obtained by formula (2) and formula (5):
Head related transfer function is standardized to obtain by formula (6):
Wherein,Indicate standardization head related transfer functionApproximation.
With reference to Fig. 4 (a) and 4 (b), using the frequency-division section linear regression that frequency-division section linear regression model (LRM) carries out, with reference to Fig. 4
(a) with 4 (b) it is found that regression result and original HRTF error are smaller by the HRTF frequency-division section linear regression realized of the present invention, with
Document " Martens W L.Principal components analysis and resynthesis of spectral
cues to perceived direction.In:Proceedings of 1987 International Computer
Music Conference, 1987,274-281 ", the method for middle proposition are compared, and method of the invention can be realized preferably
HRTF regression analysis.
Claims (1)
1. a kind of HRTF frequency-division section linear regression method based on physiological parameter, it is characterised in that include the following steps:
Step 1: choosing all subjects in the HRTF in a certain determining orientation, principal component analysis is carried out to the HRTF in the orientation, is obtained
The principal component in the orientation:
(1) choosing all subjects is being the HRTF composition vector D of (θ, φ) at azimuthij, wherein i is subject sequence, and j is frequency
Serial number, subject number are m, and wherein θ is azimuth, and φ is pitch angle；
(2) to DijCarry out following standardization:
Wherein, For DijHead associated delivery by standardization
Function；
(3) principal component analysis is carried out to the head related transfer function after standardization:
Wherein, Pm×nIt is the score matrix of m × n of principal component analysis；W is the matrix of loadings of n × n of principal component analysis, and T is indicated
The transposition of matrix；
Step 2: by all physiological parameters of subjects all in CIPIC database, respectively with the principal component that is obtained in step 1 into
Row correlation analysis obtains correlation coefficient rik, rikIndicate the related coefficient of i-th of human body physiological parameter and k-th of principal component,
I.e. i-th subject sequence of i-th of human body physiological parameter；
Step 3: determining different frequency range related physiological parameters；
To any one principal component, correlation coefficient charts are inquired, the related coefficient of the principal component Yu all physiological parameters is obtained, it will not
The related coefficient of the relevant physiological parameter of same frequency is arranged according to descending, is selected a greater than the preceding t of related coefficient significant as kth
The candidate items of a principal component；
(1) any one candidate physiological parameter is selected, calculates separately candidate's physiological parameter in the different degree I of different frequency range:
Wherein, NnkIndicate r-th of physiological parameter k-th of principal component frequency range f be the section a~b appearance number,
In the section a~b, which occurs once adding 1, does not occur then adding 0, αkFor kth Principal Component Explanation initial data population variance
Percentage, wherein 1 and c refers to which principal component, i.e. first principal component to c principal component；
(2) descending arrangement is carried out by different degree to the physiological parameter in the candidate items in different frequency respectively, q work before selection
For the major physiological parameter of the frequency range under the orientation；
(3) difference according to the physiological parameter for influencing HRTF joins common physiology number of parameters in side frequency more than total physiology
50% several frequency-portions are divided into a frequency range, if a certain frequency is unable to satisfy common physiology number of parameters in side frequency
It more than the 50% of total physiological parameter, is then divided into side frequency in the lower frequency range of frequency, by the score square of principal component
Battle array P is divided according to frequency range and is carried out segment processing, is divided into d Matrix dividing:
P=[P1, P2 ..., Pd] (4)
Step 4: approximate, the approximate principal component scores matrix of acquisition is carried out to P using the major physiological parameter of each frequency range of selection
The reconstruct data of head related transfer function are obtained using the approximate score vector of each frequency range, detailed step is as follows:
(1) score vector in principal component analysis is indicated by key physiological parameters linear regression, regression equation with formula (5):
Wherein,Indicate d-th of Matrix dividing in principal component,For the physiological parameter matrix selected from step 3；
Ad(q+1)×nFor the coefficient of human body measurements；M is subject number；N indicates different frequencies；Ad indicates that human body is surveyed at frequency range d
Quantifier purpose coefficient matrix；
(2) according to principle of least square method, coefficient matrices A i is obtained by formula (2) and formula (5):
Head related transfer function is standardized to obtain by formula (6):
Wherein,Indicate standardization head related transfer functionApproximation；
When knowing the physiological parameter of subject, the approximate HRTF of the subject is obtained by formula (7).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810894053.0A CN109166592B (en) | 2018-08-08 | 2018-08-08 | HRTF (head related transfer function) frequency division band linear regression method based on physiological parameters |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810894053.0A CN109166592B (en) | 2018-08-08 | 2018-08-08 | HRTF (head related transfer function) frequency division band linear regression method based on physiological parameters |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109166592A true CN109166592A (en) | 2019-01-08 |
| CN109166592B CN109166592B (en) | 2023-04-18 |
Family
ID=64895367
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810894053.0A Active CN109166592B (en) | 2018-08-08 | 2018-08-08 | HRTF (head related transfer function) frequency division band linear regression method based on physiological parameters |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109166592B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110493701A (en) * | 2019-07-16 | 2019-11-22 | 西北工业大学 | HRTF personalization method based on sparse principal component analysis |
| CN113938767A (en) * | 2021-12-16 | 2022-01-14 | 深圳市鑫正宇科技有限公司 | Bone conduction intelligent sleep-aiding device based on neural network |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2007110519A2 (en) * | 2006-03-28 | 2007-10-04 | France Telecom | Method and device for efficient binaural sound spatialization in the transformed domain |
| CN103621110A (en) * | 2011-05-09 | 2014-03-05 | Dts（英属维尔京群岛）有限公司 | Room characterization and correction for multi-channel audio |
| CN104284291A (en) * | 2014-08-07 | 2015-01-14 | 华南理工大学 | Headphone dynamic virtual replaying method based on 5.1 channel surround sound and implementation device thereof |
| CN105786764A (en) * | 2014-12-19 | 2016-07-20 | 天津安腾冷拔钢管有限公司 | Calculation method and device for obtaining personalized head-related transfer function (HRTF) |
| CN106162499A (en) * | 2016-07-04 | 2016-11-23 | 大连理工大学 | The personalized method of a kind of related transfer function and system |
| WO2017041922A1 (en) * | 2015-09-07 | 2017-03-16 | 3D Sound Labs | Method and system for developing a head-related transfer function adapted to an individual |
| CN106682203A (en) * | 2016-12-30 | 2017-05-17 | 西北工业大学 | Three-dimensional physiological parameter based HRTF (head related transfer function) personal matching method |
| CN107864440A (en) * | 2016-07-08 | 2018-03-30 | 奥迪康有限公司 | Hearing assistance system including EEG records and analysis system |
-
2018
- 2018-08-08 CN CN201810894053.0A patent/CN109166592B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2007110519A2 (en) * | 2006-03-28 | 2007-10-04 | France Telecom | Method and device for efficient binaural sound spatialization in the transformed domain |
| CN103621110A (en) * | 2011-05-09 | 2014-03-05 | Dts（英属维尔京群岛）有限公司 | Room characterization and correction for multi-channel audio |
| CN104284291A (en) * | 2014-08-07 | 2015-01-14 | 华南理工大学 | Headphone dynamic virtual replaying method based on 5.1 channel surround sound and implementation device thereof |
| CN105786764A (en) * | 2014-12-19 | 2016-07-20 | 天津安腾冷拔钢管有限公司 | Calculation method and device for obtaining personalized head-related transfer function (HRTF) |
| WO2017041922A1 (en) * | 2015-09-07 | 2017-03-16 | 3D Sound Labs | Method and system for developing a head-related transfer function adapted to an individual |
| CN106162499A (en) * | 2016-07-04 | 2016-11-23 | 大连理工大学 | The personalized method of a kind of related transfer function and system |
| CN107864440A (en) * | 2016-07-08 | 2018-03-30 | 奥迪康有限公司 | Hearing assistance system including EEG records and analysis system |
| CN106682203A (en) * | 2016-12-30 | 2017-05-17 | 西北工业大学 | Three-dimensional physiological parameter based HRTF (head related transfer function) personal matching method |
Non-Patent Citations (8)
| Title |
|---|
| NAOYA INOUE ET AL.: "HRTF Modeling using Physical Features", 《FORUM ACUSTICUM 2005 BUDAPEST》 * |
| WANGLEI ET AL.: "New Method for Synthesizing Personalized Head-Related Transfer Function", 《2016 IEEE》 * |
| WILLINRN L. MARTENS ET AL.: "PRINCIPAL COMPONENTS ANALYSIS AND RESYNTHESIS OF SPECTRAL CUES TO PERCEIVED DIRECTION", 《1987 ICMC PROCEEDINGS》 * |
| YUANCHENG LUO ET AL.: "Gaussian Process Models for HRTF based Sound-Source Localization and Active-Learning", 《IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING》 * |
| 张斌等: "一种头相关传输函数的双维度稀疏压缩方法", 《电声技术》 * |
| 曾向阳: "头相关传递函数的个人化方法", 《电声技术》 * |
| 谢菠荪等: "头相关传输函数的低频特性", 《声学学报》 * |
| 黄婉秋等: "基于多维生理参数的头相关传递函数个人化方法", 《西北工业大学学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110493701A (en) * | 2019-07-16 | 2019-11-22 | 西北工业大学 | HRTF personalization method based on sparse principal component analysis |
| CN110493701B (en) * | 2019-07-16 | 2020-10-27 | 西北工业大学 | HRTF (head related transfer function) personalization method based on sparse principal component analysis |
| CN113938767A (en) * | 2021-12-16 | 2022-01-14 | 深圳市鑫正宇科技有限公司 | Bone conduction intelligent sleep-aiding device based on neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109166592B (en) | 2023-04-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Super nested arrays: Linear sparse arrays with reduced mutual coupling—Part I: Fundamentals | |
| Ando et al. | Auditory and visual sensations | |
| CN101263741B (en) | Method of and device for generating and processing parameters representing HRTFs | |
| CN1860526B (en) | Encoding audio signals | |
| KR102544923B1 (en) | A system and a processing method for customizing audio experience | |
| CN109166592A (en) | HRTF frequency-division section linear regression method based on physiological parameter | |
| CN107731238A (en) | The coding method of multi-channel signal and encoder | |
| CN108596016A (en) | A kind of personalized head-position difficult labor modeling method based on deep neural network | |
| CN110517701A (en) | A kind of microphone array voice enhancement method and realization device | |
| CN106682203A (en) | Three-dimensional physiological parameter based HRTF (head related transfer function) personal matching method | |
| Vepa | Join cost for unit selection speech synthesis | |
| DE112020000883T5 (en) | Indoor positioning system | |
| Li et al. | Primitive auditory memory is correlated with spatial unmasking that is based on direct-reflection integration | |
| Jin et al. | Neural system identification model of human sound localization | |
| Kuusinen et al. | Investigation of auditory distance perception and preferences in concert halls by using virtual acoustics | |
| Lu et al. | Personalization of head-related transfer function based on sparse principle component analysis and sparse representation of 3D anthropometric parameters | |
| El Badawy et al. | Direction of arrival with one microphone, a few legos, and non-negative matrix factorization | |
| Hawkins et al. | Ultrasonic tracking of juvenile cod by means of a large spaced hydrophone array | |
| Okwori et al. | An empirical investigation of Malthusian population theory in Nigeria | |
| KR20190122839A (en) | Multi-channel signal encoding and decoding method and codec | |
| Klenova et al. | Duet structure provides information about pair identity in the red-crowned crane (Grus japonensis) | |
| Yao et al. | An individualization approach for head-related transfer function in arbitrary directions based on deep learning | |
| Kosała | Calculation models for acoustic analysis of St. Elizabeth of Hungary Church in Jaworzno Szczakowa | |
| CN110493701A (en) | HRTF personalization method based on sparse principal component analysis | |
| Talagala et al. | Novel head related transfer function model for sound source localisation |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |