阅读说明：本技术 对人类头部的声学效应进行建模的方法 (Method for modeling the acoustic effect of a human head ) 是由 海伦妮·巴乌 大卫·E·罗博隆姆 于 2020-01-31 设计创作，主要内容包括：提供了一种对人类头部建模的方法。人类头部模型具有宽度和纵横比。纵横比定义独立于人类头部模型的大小的不同头部形状。所述方法包括以下步骤：基于射线追踪和多个半平面部分来形成高频头部模型；将高频头部模型与远场阴影滤波器结合；将远场阴影滤波器与近场补偿滤波器结合,以补偿远场和近场区域之间的声学变化；以及修改人类头部模型的纵横比,以配置从近乎球形变动到非常窄的实施方式的人类头部的可变几何模型。(A method of modeling a human head is provided. The human head model has a width and an aspect ratio. The aspect ratio defines different head shapes independent of the size of the human head model. The method comprises the following steps: forming a high frequency head model based on ray tracing and the plurality of semi-planar portions; combining a high frequency head model with a far field shadow filter; combining a far-field shadow filter with a near-field compensation filter to compensate for acoustic variations between the far-field and near-field regions; and modifying the aspect ratio of the human head model to configure a variable geometry model of the human head that ranges from nearly spherical to very narrow implementations.)

1. A method of modeling a human head, the human head model having a width and an aspect ratio, the aspect ratio defining different head shapes independent of a size of the human head model, the method comprising the steps of:

Forming a high frequency head model based on ray tracing and the plurality of semi-planar portions;

combining the high-frequency head model with a far-field shadow filter;

combining the far-field shadow filter with a near-field compensation filter to compensate for acoustic variations between the far-field region and the near-field region; and

modifying the aspect ratio of the human head model to configure a variable geometry model of the human head ranging from near spherical to very narrow implementations.

2. The method of claim 1, wherein the width of the model of the human head corresponds to a anthropometric width of the human head.

3. The method of claim 1, wherein each of the semi-planar portions forms a polar angle with a forehead semi-plane oriented in a Y-axis direction.

4. The method of claim 1, wherein the generated geometric model of the human head has a substantially oval shape when viewed from the side after interpolation is considered.

5. The method of claim 1, wherein each of the semi-planar portions is formed using a rectangular shape in combination with a semi-circular shape.

6. The method of claim 5, wherein the rectangular shape has a width and a height and the semi-circular shape has a radius, and wherein the radius of the semi-circular shape is equal to half the width of the rectangular shape.

7. The method of claim 1, wherein each of the semi-planar portions is formed using an isosceles trapezoid shape in combination with a semi-circular shape.

8. The method of claim 7, wherein the isosceles trapezoid shape has a width and a height, and the semi-circular shape has a radius, and wherein the radius of the semi-circular shape is greater than half of the width of the isosceles trapezoid shape.

9. The method of claim 1, wherein the far-field shading filter accounts for variability in the shape of the human head model.

10. The method of claim 1, wherein the near-field compensation filter is configured to account for large variations in interaural level differences as a virtual source approaches the human head, particularly at low frequencies and for a single-sided source.

11. A method of modeling a human head, the human head having a width and an aspect ratio, the aspect ratio defining different head shapes independent of a size of a human head model, the method comprising the steps of:

Forming a high frequency head model based on ray tracing and the plurality of semi-planar portions;

combining the high-frequency head model with a far-field shadow filter; and

combining the far-field shadow filter with a near-field compensation filter to compensate for acoustic variations between the far-field region and the near-field region;

wherein a width of the model of the human head corresponds to a anthropometric width of the human head.

12. The method of claim 11, wherein the human head model can vary from nearly spherical to very narrow implementation.

13. The method of claim 11, wherein each of the semi-planar portions forms a polar angle with a forehead semi-plane oriented in a Y-axis direction.

14. The method of claim 11, wherein the generated geometric model of the human head has a substantially oval shape when viewed from the side after interpolation is considered.

15. The method of claim 11, wherein each of the semi-planar portions is formed using a rectangular shape in combination with a semi-circular shape.

16. The method of claim 15, wherein the rectangular shape has a width and a height and the semi-circular shape has a radius, and wherein the radius of the semi-circular shape is equal to 1/2 of the width of the rectangular shape.

17. The method of claim 11, wherein each of the semi-planar portions is formed using an isosceles trapezoid shape in combination with a semi-circular shape.

18. The method of claim 17, wherein the isosceles trapezoid shape has a width and a height, and the semi-circular shape has a radius, and wherein the radius of the semi-circular shape is greater than half of the width of the isosceles trapezoid shape.

19. The method of claim 11, wherein the far-field shading filter accounts for variability in the shape of the human head model.

20. The method of claim 11, wherein the near-field compensation filter is configured to account for large variations in interaural level differences as a virtual source approaches the human head, particularly at low frequencies and for a single-sided source.

Background

A head related transfer function (hereinafter, referred to as "HRTF") is a measure of the interaction of sound waves with the human body as they propagate from a sound source to the human ears. Non-limiting examples of such interactions include sound shadows of the head, sound reflections of the shoulders, or resonances and depressions (notch) caused by the pinna outside the ear.

The HRTF may vary with the direction of the sound source, the distance from the sound source, the frequency of the sound source, and the morphology of the listener. HRTFs can be used in a virtual auditory environment to spatialize sound sources through headphones.

The conventionally measured HRTF can be considered as a "black box" containing all the constituent acoustic processes linked to different body parts. In other examples, structural models of HRTFs attempt to decompose these acoustic processes and model them separately using digital filters and delays. When measurements are not available, the structural model can be used to synthesize the HRTF of the listener.

The acoustic effects of the human head may include delayed arrival times and shadowing of sound waves at the contralateral ear, creating interaural time differences (hereinafter "ITDs") and interaural level differences (hereinafter "ILDs"). The amount of ITDs and ILDs may vary with the direction, distance and frequency of the sound source and the size and shape of the human head. As a non-limiting part of HRTF structural models, modeling of the human head aims to accurately estimate the amount of ITDs and ILDs with respect to the position and frequency of the sound source and the morphology of the human head.

It would be advantageous if a method of modeling the acoustic effects of a human head could be improved.

Disclosure of Invention

It should be appreciated that this summary is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. This summary is not intended to identify key features or essential features of the disclosure, nor is it intended to limit the scope of the methods of modeling the acoustic effects of a human head.

The above objects, and others not specifically enumerated, are achieved by a method of modeling a human head. The human head model has a width and an aspect ratio. The aspect ratio defines different head shapes independent of the size of the human head model. The method comprises the following steps: forming a high frequency head model based on ray tracing and the plurality of semi-planar portions; combining a high frequency head model with a far field shadow filter; combining a far-field shadow filter with a near-field compensation filter to compensate for acoustic variations between the far-field and near-field regions; and modifying the aspect ratio of the human head model to configure a variable geometry model of the human head from a nearly spherical to a very narrow implementation.

The above objects, as well as other objects not specifically enumerated, are also achieved by a method of modeling a human head. The human head model has a width and an aspect ratio. The aspect ratio defines different head shapes independent of the size of the human head model. The method comprises the following steps: forming a high frequency head model based on ray tracing and the plurality of semi-planar portions; combining a high frequency head model with a far field shadow filter; a far field shadow filter is combined with a near field compensation filter to compensate for acoustic variations between the far field and near field regions. The width of the human head model corresponds to the anthropometric width of the human head.

Various objects and advantages of the method of modeling the acoustic effect of a human head will become apparent to those skilled in the art from the following detailed description, when read in light of the accompanying drawings.

Drawings

The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the office upon request and payment of the necessary fee.

Fig. 1 is a diagram illustrating elements used in a method of modeling a human head.

Fig. 2 is a front view of a geometric model of a human head.

Fig. 3A is a diagram illustrating a side view of a geometric model of the human head of fig. 2 and a plurality of superimposed semi-planes.

Fig. 3B is a perspective view of the geometric model of the human head of fig. 2 illustrating polar angles formed by individual half-planes.

Fig. 3C is a diagram illustrating a geometric model of the human head of fig. 2 after consideration of interpolation, the geometric model having an oval shape when viewed from the side.

Fig. 4A is a perspective view of the geometric model of the human head of fig. 3B, illustrating a first embodiment of a half-plane 16a formed by a rectangular shape in combination with a semicircular shape.

Fig. 4B is a front view of a second embodiment of a half plane 16a, the half plane 16a being formed by an isosceles trapezoid shape combined with a semicircular shape.

FIG. 5A is a color drawing depicting models and measurements of Interaural Time Difference (ITD) at different frequencies.

Fig. 5B is a color drawing depicting measured amplitudes of Head Related Transfer Functions (HRTFs) for a mannequin head having a torso for different orientations in a horizontal plane.

Fig. 5C is a color drawing depicting the magnitude of the Head Related Transfer Function (HRTF) for acoustic referencing of a head model for different orientations in the horizontal plane.

Fig. 5D is a color diagram depicting model amplitudes of Head Related Transfer Functions (HRTFs) for different orientations in the horizontal plane.

Fig. 5E is a color drawing depicting the alpha distribution as a function of elevation and azimuth.

Detailed Description

A method of modelling the acoustic effects of a human head will now be described with occasional reference to the detailed description. However, the method of modeling the acoustic effect of a human head may be implemented in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the method of modeling the acoustic effects of a human head to those skilled in the art.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which methods of modeling the acoustic effects of a human head belong. The terminology used herein in the description of the method of modeling the acoustic effect of the human head is for the purpose of describing particular embodiments only and is not intended to be limiting of the method of modeling the acoustic effect of the human head. As used in the method description and the appended claims of a method of modeling the acoustic effects of a human head, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

Unless otherwise indicated, all numbers expressing quantities of dimensions such as length, width, height, and so forth used in the specification and claims are to be understood as being modified in all instances by the term "about". Accordingly, unless indicated otherwise, the numerical properties set forth in the specification and claims are approximations that may vary depending upon the desired properties sought to be obtained in an embodiment of the method of modeling the acoustic effects of a human head. Although the numerical ranges and parameters setting forth the broad scope of methods of modeling the acoustic effects of a human head are approximations, the numerical values set forth in the specific examples are reported as precisely as possible. Any numerical value, however, inherently contains certain errors necessarily resulting from the errors found in their respective measurements.

Referring now to fig. 1, the description and drawings disclose a novel method 2 of modeling the acoustic effect of a human head for transmitting a binaural signal to a human ear. In general, novel method 2, which models the acoustic effects of a human head, contains three discrete elements: 1) a high frequency head model 4 based on ray tracing and a semi-planar portion (hereinafter referred to as "high frequency model"), 2) a far field shading filter 6 based on acoustic measurements of a three dimensional (hereinafter referred to as "3D") head model or numerical simulations of a 3D head model (hereinafter referred to as "acoustic reference of the head model"), and 3) a near field compensation filter 8 based on acoustic reference of the head model and configured to compensate for acoustic variations between far field and near field regions. As used herein, the term "far field region" is defined to mean a source located at a distance of one meter or more from the center of the head model. As used herein, the term "near field region" is defined to mean a source located at a distance of less than one meter from the center of the head model.

Referring again to fig. 1 and in the case of the high frequency head model 4, conventional ray tracing methods assume that the geometric path to the ear is equivalent to the time at which the sound wave reaches the same ear. Conventional head modeling methods based on ray tracing may employ simple shapes such as spheres or ellipsoids. It has been determined that this method is effective for human head sized subjects above about 2kHz and can be used to predict high frequency Interaural Time Differences (ITDs).

Referring again to fig. 1, in other conventional head modeling approaches, the geometry of spherical and elliptical head models is optimized to fit the high frequency Interaural Time Difference (ITD) of acoustic measurements. It is shown that the spherical model is most suitable for acoustic measurements when displaced axially backwards and upwards in the ear. This can lead to differences between the measurement reference (interaural axis) and the origin of the model. Similar performance was obtained from an elliptical head model. Although the elliptical head model is defined only as a high frequency model, the spherical head model contains the three discrete elements described above. However, spherical head models do not adequately approximate the shape of a human head. Thus, the spherical head model cannot accurately model the ITD of a human head, and the resulting size cannot be easily correlated with an anthropomorphic human head size.

Referring now to fig. 2, a geometric model of a human head 10 is shown. A geometric model of the human head 10 may be used to predict high frequency arrival times to the monaural. The geometric model of the human head 10 includes left and right entrances 12a, 12b to the ear canal, each located on opposite left and right sides 14a, 14b of the head. The geometric model of the human head 10 includes an interaural axis A-A. As used herein, the term "interaural axis" is defined as the axis extending between the left and right entrances 12a, 12b of the ear canal.

Referring now to fig. 3A, a geometric model of a human head 10, ear canal entrance 12a, and interaural axis a-a is illustrated. In contrast to conventional spherical or elliptical models, the geometric model of the human head 10 is segmented into a desired plurality of half-planes 16a to 16_∞Wherein the half planes 16a to 16_∞Extend in a radial direction from the interaural axis a-a. Half planes 16a to 16_∞Defines the shape of the human head.

Referring now to FIG. 3B, the half planes 16a to 16_∞Each forming a different polar angle α a to α_∞. As used herein, the term "polar angle" is defined as the forehead half-plane 16 oriented in the half-plane and along the Y-axis and extending radially from the interaural axis A-A_fpThe angle formed therebetween.

Referring now to FIG. 3C, there are no half-planes 16a through 16_∞And the resulting geometric model of the human head 10 is illustrated after the interpolation is considered. The resulting geometric model of the human head has a generally oval shape when viewed from the side.

Referring now to fig. 4A, a first embodiment of a half plane 16a is illustrated. The arrival time of the half-plane 16a is modeled using a rectangular shape 20 in combination with a semicircular shape 22. The rectangular shape has a width a and a height b. The semi-circular shape 22 has a radius c, where the radius c is equal to half the width a of the rectangle 20 (1/2). In some examples, radius c may be greater than half of width a (1/2), in which case rectangular shape 20 may have other shapes, such as non-limiting examples of isosceles trapezoidal shapes, as shown in fig. 4B and described below. The combination of the dimensions of width a, height b and radius c is intended to closely approximate an anthropomorphic human head measurement. In the illustrated embodiment, the width a is in the range of about 12.0cm to about 17.0cm, the height b is in the range of about 0.0cm to about 13.0cm, and the radius c is in the range of about 6.0cm to about 10 cm. However, in other embodiments, the width a may be less than about 12.0cm or greater than about 17.0cm, the height b may be greater than about 13.0cm, and the radius c may be less than about 6.0cm or greater than about 10.0cm, sufficient for the resulting shape to closely approximate a anthropomorphic human head measurement.

Referring now to fig. 4B, a second embodiment of the half plane 116a is illustrated. The shape of the half plane 116a is different from the half plane 16a shown in fig. 4A. Without being bound by theory, it is believed that the half plane 116a can accommodate a particular head size and can better match the shape of a human head above and behind the ear. In this embodiment, half-plane 116a is modeled using an isosceles trapezoid shape 120 in combination with a semicircular shape 122. The isosceles trapezoid shape 120 has a base width a, a height b, and a waist (leg) height b'. The base width a of isosceles trapezoid shape 120 is equal to base width a of rectangular shape 20, as shown in FIG. 4A and described above. The semi-circular shape 122 has a radius c, wherein the radius c is greater than half of the base width a of the isosceles trapezoid shape 120 (1/2). The combination of the dimensions of base width a, height b, waist height b' and radius c is intended to closely approximate an anthropomorphic human head measurement. In the illustrated embodiment, the waist height b' is in the range of about 0.0cm to about 13.0 cm. However, in other embodiments, the waist height b' may be greater than about 13.0cm, sufficient to make the resulting shape closely approximate a anthropomorphic human head measurement.

Referring again to fig. 2, the anthropomorphic head width hw defines the width of the geometric model of the human head 10 as the length of the interaural axis between the left and right entrances 12a, 12b of the ear canal. It should be understood that the anthropomorphic head width hw does not include pinna.

Referring again to FIG. 3A, all of the half-planes 16a through 16a_∞All having the same width but different lengths. As used herein, the term "length of the half-plane" is defined to mean the sum of the rectangular or trapezoidal height b and the radius c of the half-circle 22, 122. The half planes 16a to 16 in consideration of interpolation_∞The shape of the human head model is defined. The aspect ratio of the human head model may be modified independently of the width of the human head model. As used hereinThe term "aspect ratio" is defined as the ratio of the length of the half plane to half the width of the human head model. Modifying the aspect ratio of the human head model allows for a configurable geometric model of human head 10 from a nearly spherical to a very narrow implementation. The origin of the geometric model of the human head 10 is the intermediate point between the ear canal inlets 12a, 12b on the interaural axis a-a, resulting in a reference that is consistent with the acoustic measurements.

Particular attention is paid to defining a geometric model of the human head 10 corresponding to a physical shape. Ideally, this physical shape would correspond to a human head, and advantageously allows acoustic measurements or numerical simulations of ray tracing unpredictable phenomena. Non-limiting examples of such phenomena are low frequency phase characteristics, acoustic shadow and near field shadow representation.

Similar to the conventional model, the present head model geometry is optimized to fit the high frequency ITD of the acoustic measurements. To better conform to the measurements and ensure that the resulting optimal shape resembles the human head, modifications to the predicted high frequency arrival time are introduced to account for three components that affect the propagation path: 1) additional propagation paths due to the presence of the pinna, 2) multiple propagation paths around the head, and 3) additional human head width above and behind the ear.

Once the head model geometry is defined, acoustic measurements and/or numerical simulations of the head model may be performed for a plurality of source locations. The acoustic measurements and/or numerical simulations may be repeated for a plurality of head model sizes, head model shapes, head model widths, and head model aspect ratios in a manner that may substantially approximate a head continuum. It is further contemplated that the head model may be adapted to a particular human head based on anthropomorphic human head measurements.

Referring again to fig. 1 and 3A and discussed above, high frequency head model 4 is based on semi-planes 16a through 16_∞And a ray tracing formula that predicts the arrival time at the ear relative to the center of the head. The ray tracing formula uses the following equation:

If it is notThen

Or

If it is notThen

Where θ is the angle between the source and the ear relative to the center of the head, a is the width of the rectangular or trapezoidal shape, b is the height of the rectangular or trapezoidal shape, b' is the height of the waist of the isosceles trapezoid, c is the radius of the semicircle, and Co is the speed of sound. In some examples where the half-plane consists of a rectangle and a semicircle, then b' is equal to b.

Next, referring again to fig. 1 and with respect to the far-field shadow filter 6 (hereinafter "shadow filter") based on acoustic referencing of a head model, it has been found that conventional ray tracing equations for rigid spheres provide accurate pure delays above about 2 kHz. In order to model the sound shadow of the human head and increase the low frequency binaural time difference, the conventional method defines a 1-pole 1-zero shadow filter to enhance the conventional ray tracing formula. The 1 pole 1 zero shading filter is described by the following conventional equation:

wherein

Where r is an amount that varies with head size.

Subsequent methods modify this work by comparing the filter characteristics to those known in the art (hereinafter referred to as "spherical rayleigh approximation") and match the shadowing characteristics, low frequency interaural time differences and spatial variations. In the proposed approach, the same 1-pole 1-zero filter design approach is incorporated; however, the variability of α with respect to the sound source position is redefined to account for the shape of the novel model of human head 10.

Next, with respect to the near-field compensation filter 8 (hereinafter referred to as "near-field filter") based on acoustic measurement of the head model, the conventional method defines a near-field model that compensates for the measured far-field HRTF. In later correlation work, methods involving 1-pole 1-zero approximations of these methods were developed again using the spherical rayleigh approximation. In these prior art examples, a spherical rayleigh approximation is used to predict the near-field response when the virtual source is close to a rigid sphere. By comparing these spectral changes with the spectrum at the reference distance (1m), a "difference filter" can be composed. These differential filters can be used to compensate the measured HRTF under the assumption that human head variations are similar.

The proposed near-field filter 8 is configured to take into account large variations in interaural level differences as the virtual source approaches the human head, especially at low frequencies and single-sided sources. In a similar way to the above far-field filter discussed above, the proposed near-field filter 8 is based on an acoustic reference of the human head 10, as discussed above. The near field filter 8 may be of any desired and suitable configuration known in the art.

The novel method 2 of modeling 2 the acoustic effect of a human head for transmitting a binaural signal to a human ear aims at accurately estimating the amount of Interaural Time Difference (ITD) and Interaural Level Difference (ILD). Referring now to fig. 5A through 5E, the results of method 2 are illustrated. Referring first to fig. 5A, a graph depicting Interaural Time Differences (ITDs) at different frequencies is presented at 200. The graph 200 of fig. 5A has a vertical axis 222 of interaural time difference (in μ s) and a horizontal axis 224 of frequency (in Hz). For a source position (az, el) — 80.0, the model interaural time difference 230(ITD) is compared to the measured interaural time difference 232 (ITD). Similarly, for a source position (az, el) — 45.0, the model interaural time difference 234(ITD) is compared to the measured interaural time difference 236 (ITD). In a similar manner, for a source position (az, el) — (0.0), the model interaural time difference 238(ITD) is compared to the measured interaural time difference 240 (ITD). Further, for a source position (az, el) — (45.0), the model interaural time difference 242(ITD) is compared to the measured interaural time difference 244 (ITD). Finally, for the source position (az, el) — (80.0), the model interaural time difference 246(ITD) is compared to the measured interaural time difference 248 (ITD). As clearly shown in fig. 5A, the novel method of modeling a human head provides an accurate prediction of interaural time differences in frequency, as compared to acoustic measurements of a human model head (without pinna and torso).

Referring now to fig. 5B, a graph depicting the measured amplitude of a Head Related Transfer Function (HRTF) for a mannequin head having a torso for different source orientations in the horizontal plane (0.0 degrees in elevation) is presented in 300.

Referring now to fig. 5C, a graph depicting the magnitude of the Head Related Transfer Function (HRTF) for acoustic referencing of a head model for different source orientations in the horizontal plane (elevation 0.0 degrees) is presented in 400.

Referring now to fig. 5D, a graph depicting model amplitudes of Head Related Transfer Functions (HRTFs) for different source orientations in the horizontal plane (elevation 0.0 degrees) is presented in 500.

Referring now to fig. 5B, 5C, and 5D, graphs 300, 400, and 500 have vertical axes 322, 422, and 522 (in degrees), respectively, of active azimuth angles and horizontal axes 324, 424, and 524 (in Hz), respectively, of frequencies. Head Related Transfer Function (HRTF) magnitudes (in dB) corresponding to the colors illustrated by the graphs 300, 400, and 500 are specified in the vertical bars 330, 430, and 530 located on the right side of the graphs 300, 400, and 500.

As clearly shown in fig. 5B, 5C and 5D, the novel method of modeling a human head advantageously provides accurate prediction of head-related transfer function (HRTF) amplitude in frequency, as compared to acoustic measurements of a phantom head without pinna but with torso. Without being bound by theory, it is believed that the difference between the measurement illustrated by graph 300 and the acoustic reference of the model head illustrated by graph 400 can be traced back to the ripples observed in the measurement due to the presence of the torso and shoulder reflections. Further, without being bound by theory, it is believed that the difference between the acoustic references of the head model shown in graph 500 and graph 400 can be traced back to an approximation by using a 1 pole 1 zero shadow filter.

Referring now to fig. 5E, a graph depicting an alpha distribution of the shape of a novel model that considers the human head, in terms of sound source position (in azimuth and elevation), is presented in 600. Graph 600 has a vertical axis 622 (in degrees) for source elevation and a horizontal axis 624 (in degrees) for source azimuth. Alpha values (unitless) corresponding to the graphic 600 rendered in color are specified in the vertical bar 630 to the right of the graphic 600.

A departure from conventional approaches is directed to advantageously capturing the non-spherical nature of the human head. Thus, the novel method models the human head more accurately and can be adapted to a specific human head based on anthropomorphic head measurements.

In accordance with the provisions of the patent statutes, the principles and modes of operation of methods for modeling the acoustic effects of a human head have been explained and illustrated in certain embodiments. It must be understood, however, that the method of modeling the acoustic effects of a human head may be practiced otherwise than as specifically explained and illustrated without departing from its spirit or scope.