A method for decoding sound field playback space based on sparse plane wave decomposition

Abstract

The present invention belongs to the technical field of spatial sound field playback, and specifically relates to a sound field playback spatial decoding method of high-order Ambisonics based on sparse plane wave decomposition. The method realizes sound field playback through two stages. The first stage pre-acquires the driving function of the loudspeaker system to play back plane waves in different directions, which is called the pre-calculation stage; the second stage performs sparse plane wave decomposition on the target sound field to calculate the playback driving function, which is called the playback decoding stage. The present invention uses the spherical harmonic domain sparse plane wave decomposition method to play back the sound field, and effectively improves the sound field playback accuracy when the target sound field is generated by a small number of sound sources. The method of the present invention is simple and low in cost, and can use the spherical harmonic coefficients of the target sound field with lower orders to achieve higher-precision playback; it has a good computing speed, and the plane wave generator of the playback system can be measured and calculated in advance, and the loudspeaker system drive signal can be quickly obtained in the playback stage; at the same time, a larger optimal listening range is guaranteed.

Description

Technical Field

[0001] The invention belongs to the technical field of spatial sound field playback, and in particular relates to a sound field playback spatial decoding method of high-order Ambisonics based on sparse plane wave decomposition. Background Art

[0002] Using speaker arrays to reproduce sound fields is an important way to achieve acoustic virtual reality and has been widely studied in recent decades. The purpose of sound field reproduction is to provide listeners with different auditory experiences, such as listening to music, watching movies or playing video games. High Order Ambisonics (HOA) is a common sound field reproduction method. HOA is divided into encoding and decoding stages. In the encoding stage, the target sound field is decomposed in the spherical harmonic domain to obtain a set of spherical harmonic coefficients representing the original sound field. In the decoding stage, the speaker array driving function is obtained so that the spherical harmonic coefficients of the reproduced sound field are the same as those obtained in the encoding stage.

[0003] The existing HOA decoding method is usually the modal matching method (Poletti M A. Three-dimensional surround sound systems based on spherical harmonics [J]. J. Audio Eng. Soc, 2005, 53 (11): 1004–1025.), which decomposes the sound field into a superposition of a set of spherical harmonic functions (basis functions) in the spherical coordinate system, and uses the spherical harmonic function modes between the primary sound field and the playback sound field to match and solve the driving function of the secondary sound source (speaker array). However, in the sound field playback method using the modal matching method, the order of the recorded sound field of the target sound field has a great influence on the measurement accuracy and the size of the optimal point. The size range of the optimal point is inversely proportional to the frequency and proportional to the mode order. When the constant modal order is 4, at a frequency of 2kHz, the size of the optimal point is reduced to less than the size of a human head.

[0004] In order to improve the playback performance and ensure that the optimal point size is within a certain range, when the target sound field is controlled by a small number of sound sources, the present invention proposes a high-order Ambisonics decoding method based on sparse plane wave decomposition. Summary of the invention

[0005] In order to avoid the shortcomings of the prior art, the present invention proposes a high-order Ambisonics decoding method based on sparse plane wave decomposition.

[0006] Technical Solution

[0007] The present invention adopts a plane wave sparse decomposition method in the spherical harmonic function domain to decompose the target sound field into a limited number of plane waves, and reproduces the target sound field according to a pre-acquired plane wave driving function of the reproduction system.

[0008] The present invention realizes the sound field playback through two stages. The first stage is to obtain the driving function of the speaker system to play back plane waves in different directions in advance, which is called the pre-calculation stage; the second stage is to perform sparse plane wave decomposition on the target sound field to calculate the playback driving function, which is called the playback decoding stage. The following are the specific implementation steps of the two stages:

[0009] Pre-calculation phase:

[0010] Step 1: Determine the sound pressure of the target sound field, which can be the sound pressure of the sound field environment recorded by using a high-order Ambisonics microphone array, or the sound pressure of a synthesized desired acoustic scene.

[0011] Step 2: Measure the spherical harmonic coefficients of the target sound field. The measured sound pressure of the target sound field can be effectively collected by a spherical microphone array and decomposed in the spherical harmonic function domain to obtain a set of spherical harmonic coefficients of the sound field. Commonly used spherical microphone arrays include hollow spherical arrays, cardioid microphone spherical arrays, double-radius hollow spherical arrays, rigid spherical arrays, etc.

[0012] Step 3: Decompose the measured or synthesized target sound field sound pressure into a plane wave in the spherical harmonic domain. The spherical harmonic coefficients of the target sound field can be expressed as a weighted combination of a set of plane wave spherical harmonic coefficients.

[0013] Step 4: The spherical harmonic coefficients of the target sound field are expressed as a weighted combination of a set of plane wave spherical harmonic coefficients. The plane wave weights are sparse, and the sparse solution of the plane wave weights can be obtained using the 1-norm.

[0014] Replay decoding stage

[0015] Step 1: For a fixed loudspeaker system and reproduction, measure the loudspeaker to reproduction area transfer function.

[0016] Step 2: Calculate the weight matrix of the plane wave generator. In the entire playback area, the plane wave generator can be obtained by calculating the minimum error between the playback sound field and the plane wave sound field. The loudspeaker drive signals that generate plane waves in different directions are calculated.

[0017] Step 3: Combine the calculated plane wave weight with the plane wave generator to obtain the driving function of the speaker in the high-order Ambisonics decoding method of the present invention.

[0018] Beneficial Effects

[0019] The present invention proposes to use the spherical harmonic domain sparse plane wave decomposition method to replay the sound field, which effectively improves the accuracy of sound field reproduction when the target sound field is generated by a small number of sound sources. The method of the present invention is simple and has a lower cost than the traditional method, and can use lower-order spherical harmonic coefficients of the target sound field to achieve higher-precision reproduction; the present invention has a good computing speed, and the plane wave generator of the reproduction system can be measured and calculated in advance, and the speaker system driving signal can be quickly obtained during the reproduction stage; at the same time, compared with the traditional method, the present invention ensures a larger optimal listening range. In addition, the method proposed in this article can also be applied to the reproduction system in a reverberant environment. Through actual testing, it is verified that the present invention has higher reproduction accuracy than the traditional method, and can therefore be applied to the target sound field reproduction in a reverberant environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 Schematic diagram of the distribution of loudspeakers in a certain space in an embodiment

[0021] Figure 2 The comparison diagram of the playback errors of the two methods in different optimal listening ranges in the reverberation environment in the embodiment is shown in FIG.

[0022] Figure 3 Flow chart of the sound field playback method of the present invention in an embodiment DETAILED DESCRIPTION

[0023] The sound field playback method of the present invention is now described in detail with reference to examples and drawings, but the embodiments of the present invention are not limited thereto.

[0024] The sound field radiated by the sound source at a distance of (3,0,0) is reproduced. The reproduction speaker array consists of 144 speakers with a radius of 1m and uniform spherical distribution. The target sound field for reproduction is a square area of ​​0.5m×0.5m, with its center and the center of the speaker as the coordinate origin.

[0025] Preprocessing stage:

[0026] Step 1: Determine the sound pressure p of the target sound field, which is the sound pressure of the sound field environment recorded by the high-order Ambisonics microphone array, or the synthesized expected sound field scene sound pressure. The sound pressure of the target sound field is expressed in the spherical harmonic domain as:

[0027]

[0028] Among them, ω=2πf is the angular velocity, f is the frequency, r is the distance, and θ is the azimuth. is the pitch angle, k=ω / c is the wave number, c is the speed of sound, n is the order, and m is the angle. are the frequency-dependent spherical harmonic coefficients, j n (·) is the nth-order spherical Bessel function. are spherical harmonics:

[0029]

[0030] Among them, P n is the Legendre polynomial of order n.

[0031] Step 2: Plane wave decomposition of spherical harmonics. The spherical harmonic coefficients of the desired sound field can be expressed as a weighted combination of a set of plane wave spherical harmonic coefficients:

[0032]

[0033] Where L w is the number of plane wave bases, w l is the weight of the lth plane wave, and “*” is the conjugate symbol.

[0034] Step 3: Consider all orders and all angles and express formula (3) in matrix form:

[0035] Pw=A (4)

[0036] Where P is (N+1) 2 ×L w dimensional matrix, containing the plane wave spherical harmonic coefficients of different orders and angles. w is L w ×1 vector, representing the plane wave weights in different directions.

[0037] Step 4: Find the 1-norm sparse solution of formula (4):

[0038]

[0039] Decoding stage:

[0040] Step 1: Assume that the speaker weight vector d is generated by L w The acoustic field controlled by a plane wave is expressed as

[0041]

[0042] Among them, the first w A plane wave is represented as X is the generator weight matrix.

[0043] Step 2: When the speaker array and the playback acoustic environment are fixed, X can be pre-calculated. w The plane wave is:

[0044]

[0045] Where G is the transfer function from the loudspeaker to the control point, For the first wThe sound pressure generated by the plane wave at the control point. The solution is:

[0046]

[0047] Where λ0 is the regularization parameter and H is the conjugate transpose.

[0048] Step 3: The solution of formula (5) is expressed as w=f(P,A), then in the high-order Ambisonics decoding method of the present invention, the driving function of the speaker is:

[0049] d=Xf(P,A) (9)

Claims

1. A method for decoding a sound field playback space based on sparse plane wave decomposition, characterized in that: The sound field playback is achieved through two stages. The first stage is to obtain the driving function of the speaker system to play back plane waves in different directions in advance, which is called the pre-calculation stage. The second stage is to perform sparse plane wave decomposition on the target sound field to calculate the playback driving function, which is called the playback decoding stage. The specific process of the pre-calculation stage is as follows: Step 1: Determine the sound pressure of the target sound field, which is the sound pressure of the sound field environment recorded by a high-order Ambisonics microphone array, or the sound pressure of the synthesized expected acoustic scene; Step 2: Spherical harmonic coefficient measurement of the target sound field; the measured sound pressure of the target sound field can be effectively collected by a spherical microphone array and decomposed in the spherical harmonic function domain to obtain a set of spherical harmonic coefficients of the sound field; Step 3: Decompose the measured or synthesized target sound field sound pressure in the spherical harmonic domain by plane wave; the spherical harmonic coefficients of the target sound field can be expressed as a weighted combination of a set of plane wave spherical harmonic coefficients; Step 4: The spherical harmonic coefficients of the target sound field are expressed as a weighted combination of a set of plane wave spherical harmonic coefficients; the plane wave weights are sparse, and the sparse solution of the plane wave weights can be obtained using the 1 norm; The specific process of the playback decoding stage is as follows: Step 1: For a fixed loudspeaker system and playback, measure the transfer function from the loudspeaker to the playback area; Step 2: Calculate the plane wave generator weight matrix; In the whole playback area, the plane wave generator can be obtained by calculating to minimize the error between the playback sound field and the plane wave sound field; Calculate and obtain loudspeaker driving signals that generate plane waves in different directions; Step 3: Combine the calculated plane wave weights with the plane wave generator to obtain the driving function of the speaker in the high-order Ambisonics decoding method.

Images