阅读说明：本技术 一种扬声器虚拟声重放的音色均衡方法 (Tone equalization method for virtual sound reproduction of loudspeaker ) 是由 刘路路 谢菠荪 于 2020-04-20 设计创作，主要内容包括：本发明提供的一种扬声器虚拟声重放的音色均衡方法,输入原始的单通路声频信号<I>E</I><Sub>0</Sub>；对单通路声频信号经过一对听觉传输合成滤波器进行处理,得到两通路的信号<I>E</I><Sub><I>L</I></Sub>和<I>E</I><Sub><I>R</I></Sub>,所述一对听觉传输合成滤波器采用包括高频段功率均衡和低频段乘以与频率无关的增益因子的扬声器虚拟声信号处理函数对单通路声频信号进行滤波,将两通路信号<I>E</I><Sub><I>L</I></Sub>和<I>E</I><Sub><I>R</I></Sub>分别馈给对称布置在水平面前方的左扬声器和右扬声器重放。通过引入分频频率,在分频频率以上的高频段对扬声器虚拟声信号进行功率均衡,而保持低频段幅度谱的形状不变,改善了扬声器虚拟声重放的音色,并且可以避免重放的低频不足,且本发明不影响重放的虚拟源定位。(The invention provides a tone equalization method for virtual sound reproduction of a loudspeaker, which inputs an original single-channel audio signal E 0 (ii) a Processing the single-channel audio signal by a pair of auditory transmission synthesis filters to obtain two-channel signals E L And E R the pair of auditory transmission synthesis filters the single-path audio signal using a speaker virtual acoustic signal processing function including high-band power equalization and low-band multiplication by a frequency-independent gain factor, filtering the two-path signal E L And E R respectively fed to the left and right loudspeakers symmetrically arranged in front of the horizontal plane. By passingThe invention introduces the crossover frequency, carries out power equalization on the virtual sound signal of the loudspeaker in a high frequency band above the crossover frequency, keeps the shape of a low frequency band amplitude spectrum unchanged, improves the tone color of the virtual sound replay of the loudspeaker, can avoid the insufficient low frequency of the replay, and does not influence the virtual source positioning of the replay.)

1. A method of tone equalization for virtual sound reproduction by a loudspeaker, comprising the steps of:

inputting an original single-path audio signal E₀；

Processing the single-channel audio signal by a pair of auditory transmission synthesis filters to obtain a two-channel signal E_LAnd E_RThe pair of auditory transmission synthesis filters employs speaker virtual sound including high band power equalization and low band multiplication by a frequency-independent gain factorThe signal processing function filters the single path audio signal,

will two-path signal E_LAnd E_RAre fed to the left and right loudspeakers, respectively, which are arranged in front of the horizontal plane.

2. The method of claim 1, wherein the speaker virtual sound signal processing function comprising high-band power equalization and low-frequency multiplication by a frequency-independent gain factor is as follows:

E_L＝G'_L(θ_S,f)E₀E_R＝G'_R(θ_S,f)E₀

wherein, G'_L(θ_SF) is the left speaker signal processing function, G'_R(θ_SF) is the right speaker signal processing function, α and β represent the transfer functions from the two speakers to the ipsilateral and ipsilateral ears, respectively, H_L(θ_SF) and H_R(θ_SAnd f) respectively represent an azimuth angle theta_SThe transfer function of the target virtual source to the left ear and the right ear; f represents frequency; f. of₀Is a frequency division point which represents high frequency and low frequency and is larger than f₀Is a high frequency band, less than f₀Low frequency band is obtained; g₀Is a constant representing the gain at low frequencies.

3. Method for tone equalization of the virtual sound reproduction of loudspeakers according to claim 2, characterised in that said crossover frequency f₀Obtained by the following method:

in diffuse sound field approximation, the center frequency is f_CIs generated at two sound receiving points which are separated by delta rThe normalized cross-correlation coefficient of (a) is:

where Ψ represents the normalized cross-correlation coefficient, k_cRepresentative wave number, c 343m/s is sound velocity since | sin (k)_CDelta r) | is less than or equal to 1, and the maximum value | Ψ ∞ of the normalized cross-correlation coefficient_max≤(k_CΔr)^-1Given this maximum, the division frequency is obtained:

4. a method as claimed in claim 3, characterized in that said Δ r is the average distance between the ears, the corresponding crossover frequency f₀＝1.5kHz。

5. The method of claim 2, wherein the low frequency gain constant G is a constant that is constant for the timbre equalization of the virtual sound reproduction of the loudspeaker₀At a dividing frequency f according to the power spectrum of the high frequency band and the low frequency band₀And selecting nearby continuous conditions.

Technical Field

The invention relates to the technical field of electroacoustic, in particular to a tone equalization method for virtual sound reproduction of a loudspeaker.

Background

Loudspeaker virtual sound reproduction, also called auditory transmission reproduction, is a spatial sound reproduction technique. It uses head related transfer function signal processing to control binaural sound pressure in speaker playback, thereby using a small number of speakers to generate perceived virtual sources in different spatial directions. One typical application of speaker virtual sound reproduction is virtual surround sound reproduction, which uses a pair of speakers arranged in front of a listener to generate a virtual speaker in a front half-horizontal plane, thereby enabling reproduction of signals of multi-channel surround sound with a small number of speakers. This approach has been applied to virtual playback of 5.1-channel and other multi-channel surround sound, such as Dolby virtual surround sound technology, SRS and Qsurround technology, among others. Another typical application of loudspeaker virtual sound reproduction is sound reproduction in computer games, with a pair of loudspeakers arranged on both sides of the computer display to achieve a spatial sound localization effect.

One problem with virtual sound reproduction from loudspeakers is that the sound quality changes during reproduction, which affects the effect of the reproduction. There is a need to add tone equalization signal processing to improve playback. A full (audible) band power equalization method in the range of 20Hz to 20kHz is disclosed in the national patent of invention grant (ZL 02134416.7). The method can reduce the change of the reproduced tone color to a certain extent. In the national patent of invention grant (ZL200610037495.0), the head-related transfer function filters for signal processing are simplified and the change in the reproduced timbre is further reduced. However, the timbre equalization signal processing involved in the two national patent applications still suffers from the low frequency inadequacy of reproduced sound.

Disclosure of Invention

The invention further provides a tone color equalization method for virtual replay of the loudspeaker on the basis of the prior art. The method introduces a tone color equalization method of a frequency division band in the virtual sound signal processing of the loudspeaker, can further reduce the change of tone color in the playback process, particularly reduce low-frequency loss, and can not sense the virtual source direction.

The invention discloses a tone equalization method for virtual sound reproduction of a loudspeaker, which comprises the following steps:

inputting an original single-path audio signal E₀；

Processing the single-channel audio signal by a pair of auditory transmission synthesis filters to obtain a two-channel signal E_LAnd E_RSaid pair of auditory transmit synthesis filters filtering the single path audio signal with a loudspeaker virtual acoustic signal processing function comprising a high band power equalization and a low band multiplied by a frequency independent gain factor,

will two-path signal E_LAnd E_RRespectively fed to the left and right loudspeakers symmetrically arranged in front of the horizontal plane.

In one embodiment, the virtual sound signal processing function of the speaker including the high-band power equalization and the low-frequency multiplied by the frequency-independent gain factor is specifically as follows:

E_L＝G'_L(θ_S,f)E₀E_R＝G'_R(θ_S,f)E₀

wherein, G'_L(θ_SF) is the left speaker signal processing function, G'_R(θ_SF) is the right speaker signal processing function, α and β represent the transfer functions from the two speakers to the ipsilateral and ipsilateral ears, respectively, H_L(θ_SF) and H_R(θ_SAnd f) respectively represent an azimuth angle theta_STo the left and right ears of the target virtual source of (1); f represents frequency; f. of₀Is a frequency division point which represents high frequency and low frequency and is larger than f₀Is a high frequency band, less than f₀Low frequency band is obtained; g₀Is representative of the low frequency gainIs constant.

In one embodiment, the frequency division frequency f₀Obtained by the following method:

in diffuse sound field approximation, the center frequency is f_CThe normalized cross-correlation coefficient of the acoustic signal generated at two sound receiving points separated by Δ r is:

where Ψ represents the normalized cross-correlation coefficient, k_cRepresenting wave number, c 343m/s is sound velocity due to | sin (k)_CDelta r) | is less than or equal to 1, and the maximum value | Ψ ∞ of the normalized cross-correlation coefficient_max≤(k_CΔr)^-1Given this maximum, the division frequency is obtained:

in one embodiment, Δ r is an average distance between two ears, and the corresponding frequency f is divided₀＝1.5kHz。

In one embodiment, the low frequency gain constant G₀At a dividing frequency f according to the power spectrum of the high frequency band and the low frequency band₀And selecting nearby continuous conditions.

The principle of the invention is as follows: in loudspeaker virtual sound reproduction, the virtual source position is determined by the relative amplitude and relative phase between the channel signals. The simultaneous multiplication or division of the loudspeaker signals by a common factor related to frequency does not affect the position of the virtual source, but changes the power spectrum of the loudspeaker signals so that the reproduced timbre can be equalized. In an ideal diffusion field, sound pressures generated by a plurality of loudspeakers at sound receiving points are not in correlated superposition, so that the power equalization is carried out on each loudspeaker signal, and the total power spectrum of each loudspeaker signal is not changed along with the frequency change. In practice, however, reproducing the sound field in a room is not an ideal diffuse sound field. Particularly at low frequencies, the sound pressure at the sound receiving point is not an uncorrelated superposition of the sound pressures generated by the loudspeakers. Power equalization throughout the full band range (from 20Hz to 20kHz) can result in excessive low frequency attenuation, causing insufficient low frequency playback. Since the sound field of the actual playback room is relatively close to the diffuse sound field above a certain frequency, the power of the virtual sound signal of the loudspeaker can be equalized in a high frequency band above the certain frequency, and the amplitude spectrum shape of the signal is kept unchanged by multiplying a low frequency band below the certain frequency by a gain factor irrelevant to the frequency. This makes it possible to improve the reproduction of high-frequency timbres while avoiding the disadvantages of low frequencies.

Compared with the prior art, the invention has the following beneficial effects:

1. the invention introduces the frequency division frequency, the frequency band above the frequency division frequency is a high frequency band, the frequency band below the frequency division frequency is a low frequency band, the power of the virtual sound signal of the loudspeaker is balanced in the high frequency band, the amplitude spectrum shape of the signal is kept unchanged in the low frequency band, the tone color of the virtual sound reproduction of the loudspeaker is improved, the insufficient low frequency of the reproduction can be avoided, and the invention does not influence the virtual source positioning of the reproduction.

2. The invention can be used for virtual replay of various multi-channel (such as 5.1 channels) surround sound, and can also be used for computer sound replay and game replay.

Drawings

Fig. 1 is a block diagram of a tone equalization method for virtual sound reproduction of a speaker according to an embodiment of the present invention.

Fig. 2 is a schematic diagram of a virtual playback speaker arrangement and speaker to binaural transmission.

Fig. 3 is a result of a virtual source location experiment demonstrating the present invention.

Fig. 4 is a result of a subjective evaluation test for timbre in accordance with the present invention.

Detailed Description

The present invention will be described in further detail with reference to the following drawings and examples, but the scope of the present invention is not limited thereto.

A block diagram of the system of the present invention is shown in fig. 1. A pair of real horizontal speakers are arranged in front left and front right directions of a listener. Inputting an original single-path audio signal E₀(ii) a For single-channel audio signal through a pair of auditory transmission synthesis filtersProcessing to obtain two-path signal E_LAnd E_RThe pair of auditory transmission synthesis filters the single-path audio signal using a speaker virtual acoustic signal processing function including high-band power equalization and low-band multiplication by a frequency-independent gain factor, filtering the two-path signal E_LAnd E_RRespectively fed to the left and right loudspeakers symmetrically arranged in front of the horizontal plane.

The audio signal is processed by a pair of auditory transmission synthesis filters and then fed to the left front loudspeaker and the right front loudspeaker for reproduction respectively. The response of the acoustic transmission synthesis filter is obtained by equalizing binaural sound pressures reproduced by the two speakers to that of the target sound source and adding a tone equalization process.

The azimuth angle of the horizontal plane is selected to be more than or equal to 0 degree<360 °, where θ is 0 °,90 °,180 °, and 270 ° represent right front, right, rear, and left directions, respectively. For the horizontal angle theta in the free field_SThe sound pressure of the ears can be respectively transmitted from the sound source to the Head Related Transfer Function (HRTF) H of the left ear_L(θ_SF) Head Related Transfer Function (HRTF) H of the right ear_R(θ_SAnd f) obtaining:

P_L(f)＝H_L(θ_S,f)E₀(f) P_R(f)＝H_R(θ_S,f)E₀(f) (2)

where f denotes the frequency, E₀＝E₀(f) Is the amplitude signal of the sound source, H_L(θ_SF) and H_R(θ_SAnd f) respectively represent an azimuth angle theta_STo the left and right ear.

For the virtual reproduction of the two loudspeakers shown in fig. 2, the azimuth angles of the two loudspeakers are theta_LAnd theta_R. Four transfer functions (HRTFs) from two loudspeakers to two ears are respectively H_LL(θ_L，f)，H_RL(θ_L，f)，H_LR(θ_RF) and H_RR(θ_RF) the two loudspeaker signals are respectively E_LAnd E_R. Reproducing the generated binaural sound pressure to produce a stack of binaural sound pressures for two loudspeakersAdding:

considering left-right symmetry, the transfer functions from the two loudspeakers to the ipsilateral ear are equal, and the transfer functions to the different ears are also equal:

H_LL(θ_L,f)＝H_RR(θ_R,f)＝α H_LR(θ_R,f)＝H_RL(θ_L,f)＝β (4)

(3) is formed as

P'_L(f)＝αE_L+βE_RP'_R(f)＝βE_L+αE_R(5)

When equation (2) is equal to equation (5), that is, when binaural sound pressure generated by virtual playback is equal to the target sound source, signals of the two speakers are obtained as follows:

E_L＝G_L(θ_S,f)E₀E_R(f)＝G_R(θ_S,f)E₀(6)

where the response of a pair of auditory transmitters is

Whereby the two loudspeaker signals are transmitted using a pair of acoustic transmission filters G_L(θ_SF) and G_R(θ_SF) to the input signal E₀And (4) filtering to obtain the filter.

High frequency power equalization is the equalization of a pair of auditory transmission responses of equation (7) with a common factor that is frequency dependent. Equalization with a constant power factor will cause the response of the auditory transmission filter to become:

the denominator of the above equation is a common factor of constant power, which is frequency dependent. Keeping the total power of the high-frequency band signal unchanged after equalization:

|G'_L(θ_S,f)|²+|G'_R(θ_S,f)|²＝1 (9)

the embodiment adopts a tone equalization method with sub-bands: at the frequency of division f₀In the high frequency band, power equalization processing is added; for frequency division frequency f₀The following low frequency bands are not increased in frequency-dependent equalization but merely multiplied by a frequency-independent gain factor G₀So that the power spectra of the high and low bands are continuous around the crossover frequency. After the equalization process, a pair of auditory transmission filter responses shown in formula (1) is obtained.

E_L＝G'_L(θ_S,f)E₀E_R＝G'_R(θ_S,f)E₀

Frequency division frequency f₀This is selected as follows. In diffuse sound field approximation, the center frequency is f_CThe normalized cross-correlation coefficient of the acoustic signal generated at two sound receiving points separated by Δ r is:

where c is 343m/s is the speed of sound. Due to | sin (k)_CDelta r) | is less than or equal to 1, and the maximum value | Ψ ∞ of the normalized cross-correlation coefficient_max≤(k_CΔr)^-1Given this maximum, the division frequency can be obtained

If the average distance between ears is selected as 0.175m, the correlation coefficient | Ψ is not zero_maxWhen the ratio is 0.1, 0.2 and 0.3, the results are divided intoFrequency f₀3.1kHz, 1.6kHz and 1.0kHz respectively. Considering that the low-frequency binaural time difference of 1.5kHz is the main localization factor of virtual sound reproduction of the loudspeaker, in order to make this factor not affected by equalization, the crossover frequency in the embodiment of the present invention is taken as f₀This also allows the correlation coefficient to approach 0.2 at 1.5 kHz.

The principle of the embodiment of the invention is as follows:

in loudspeaker virtual sound reproduction, the virtual source position is determined by the relative amplitude and relative phase between the channel signals. The simultaneous multiplication or division of the loudspeaker signals by a common factor related to frequency does not affect the position of the virtual source, but changes the power spectrum of the loudspeaker signals so that the reproduced timbre can be equalized. In an ideal diffusion field, sound pressures generated by a plurality of loudspeakers at sound receiving points are not in correlated superposition, so that the power equalization is carried out on each loudspeaker signal, and the total power spectrum of each loudspeaker signal is not changed along with the frequency change. In practice, however, reproducing the sound field in a room is not an ideal diffuse sound field. Particularly at low frequencies, the sound pressure at the sound receiving point is not an uncorrelated superposition of the sound pressures generated by the loudspeakers. Power equalization throughout the full band range (from 20Hz to 20kHz) can result in excessive low frequency attenuation, causing insufficient low frequency playback. Since the sound field of the actual playback room is relatively close to the diffuse sound field above a certain frequency, the present embodiment performs power equalization on the virtual sound signal of the speaker in the high frequency band above the certain frequency while maintaining the amplitude spectrum shape of the low frequency band signal, thereby improving the playback high frequency tone color and avoiding the deficiency of the low frequency.

The embodiment can be realized by a general DSP hardware circuit or a special integrated circuit, and can also be realized on a multimedia computer by software programmed by an algorithm language (such as VC + +).

The embodiment can be designed into special hardware or general software for sound reproduction in aspects of digital televisions, intelligent loudspeaker systems and the like, and can also be used as hardware or software for sound reproduction of multimedia computers.

For ease of understanding, the following describes the application of the inventive arrangements to a multimedia computer.

Reading the single-pass signal stored in the hard disk by the computer or obtaining the single-pass signal through a digital transmission medium such as the internet, and then using the computer software to execute the virtual signal processing of fig. 1 (or using a dedicated hardware circuit on the sound card of the computer) to obtain the two-pass signal E_LAnd E_RThen the two full-frequency band loudspeakers are respectively fed to an external connection or a computer for reproduction.

The signal processing in the embodiment of the invention uses head-related transfer function data of a KEMAR artificial head, and the length of the head-related transfer function data is 512 points, and the sampling frequency is 44.1 kHz. Virtual signal processing is implemented using Finite Impulse Response (FIR) filters.