JP6695738B2 - Signal conversion coefficient calculation device, signal conversion device, and program - Google Patents

Description

  The present invention relates to a signal conversion coefficient calculation device, a signal conversion device, and a program.

  There are multi-channel audio systems, such as 22.2 ch audio, with more channels than the conventional 5.1 ch. In the production of these contents, the audio engineer performs mixing so as to achieve the optimum balance in the environment where the speaker is arranged at the specified position. On the other hand, when viewing content in a synthetic sound field generated by sound waves emitted from a plurality of sound sources, it is known that the sound image is localized in the normal direction of the wavefront at the listening position in the synthetic sound field (for example, Non-Patent Document 1). reference).

Ando Akio, "Sound Reproduction", Corona Publishing, 2014, p106-121

  In the production of audio content, the content is produced according to the number of channels and the speaker arrangement defined in a certain audio format. Therefore, even when viewing the content produced with the number of channels and the speaker arrangement defined by a certain audio format, it is desirable that the content is reproduced with the number of speakers and the speaker arrangement according to the regulation. However, it is difficult to install a speaker having a number of channels exceeding 5.1 ch such as 22.2 ch sound in a general home. Therefore, even in an environment in which at least one of the number of channels and speaker arrangement specified by the audio format of the playback content is different, the spatial impression should be changed as much as possible when viewed in the environment of the specified number of channels and speaker arrangement. It is desired to convert the audio signal of the reproduced content so that the audio signal can be viewed without any need. In addition, when listening to the content, it is possible that a plurality of persons listen at the same time or listen to the content while moving. Therefore, it is desirable to be able to listen to the content uniformly at a plurality of listening positions.

  The present invention has been made in view of the above circumstances, and is intended to enable near-homogeneous listening at a plurality of listening positions when audio content is reproduced in an environment different from that when the audio content is produced. Provided are a signal conversion coefficient calculation device, a signal conversion device, and a program capable of calculating a signal conversion coefficient.

  One aspect of the present invention is that a plurality of speakers are arranged so that normal vectors of wavefronts formed at a plurality of listening positions in a sound field of a reproduction environment synthesized by sound waves emitted from a plurality of speakers are close to each other. A calculation unit that calculates a complex weighting coefficient used for conversion of an audio signal to be output for each of the speakers, and the calculation unit reduces the variance of the normal vector of the wavefront formed at each of the plurality of listening positions. The signal conversion coefficient calculation device calculates the complex weight coefficient as described above.

  One aspect of the present invention is the above-described signal conversion coefficient calculation device, wherein the calculation unit arbitrarily sets a difference between a normal vector of a wavefront formed at each of the plurality of listening positions and a normal vector of a desired wavefront. The complex weighting factor is calculated under the constraint that the norm of is within a predetermined range.

  One aspect of the present invention is the signal conversion coefficient calculation device described above, wherein the calculation unit forms an angle between a normal vector of a wavefront formed at each of the plurality of listening positions and a normal vector of a desired wavefront. , The complex weighting factor is calculated under the constraint that it falls within a predetermined range.

  One aspect of the present invention is the signal conversion coefficient calculation device described above, wherein the calculation unit calculates a normal vector of the desired wavefront based on a speaker position and a listening position in an environment different from the reproduction environment. .

  According to an aspect of the present invention, any one of the above-described signal conversion coefficient calculation devices and the signal conversion coefficient calculation device causes each of the plurality of speakers to output based on a complex weighting coefficient calculated for each of the plurality of speakers. And a signal conversion unit that converts an audio signal.

  One embodiment of the present invention is a program that causes a computer to function as any of the above-described signal conversion coefficient calculation devices.

  According to the present invention, it is possible to calculate a signal conversion coefficient for enabling nearly homogeneous listening at a plurality of listening positions when reproducing audio content in an environment different from that when producing audio content.

FIG. 1 is a block diagram showing a configuration of an audio system according to an embodiment of the present invention. It is a functional block diagram which shows the structure of the signal conversion coefficient calculation apparatus by the same embodiment. It is a flowchart which shows the process of the signal conversion coefficient calculation apparatus by the same embodiment.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Consider playing audio content produced in a certain audio format in a conversion destination environment that is different from the conversion source environment. The conversion source environment is an environment in which the number of channels and the speaker arrangement are defined by the audio format of the audio content. The conversion destination environment is a reproduction environment in which at least one of the number of channels and the speaker arrangement is different from the conversion source environment. For example, the conversion source environment is 22.2 ch, the conversion destination environment is 7.1 ch or 9.1 ch, etc., but the conversion source environment and the conversion destination environment can be arbitrary.

  FIG. 1 is a configuration diagram of an audio system 100 according to the present embodiment. The audio system 100 reproduces the acoustic content created in the conversion source environment in the conversion destination environment. The conversion source environment is an environment in which M (M is an integer of 1 or more) speakers 3 are present. Hereinafter, the M speakers 3 will be described as speakers 3-1 to 3-M, respectively. The audio content includes audio signals A of M channels. The audio signal A of the m-th channel (m is an integer of 1 or more and M or less) is referred to as an audio signal A-m. The audio signal A-m is the audio signal A reproduced at the position of the speaker 3-m.

  The audio system 100 includes a signal conversion coefficient calculation device 1, a signal converter 5 (a signal conversion device), and L (L is an integer of 1 or more) speakers 7. Hereinafter, the L speakers 7 are described as speakers 7-1 to 7-L, respectively. In the figure, the audio system 100 is provided with the signal conversion coefficient calculation device 1 outside the signal converter 5, but may be provided inside the signal converter 5.

  Since the speaker 7 is not located at the same position as the speaker 3 in the conversion destination environment, the audio signal produced in the conversion source environment may be reproduced by the speaker 7 at a position where the speaker 3 does not exist. Even in such a case, the signal converter 5 converts the audio signal in order to reproduce the audio content in the conversion destination environment so that the sound image is localized in the same direction as when reproduced by the speaker arrangement of the conversion source environment. In addition, at this time, it is desirable to be able to listen with a uniform quality at a plurality of listening positions or a certain area.

  The signal converter 5 inputs the audio signals A-1 to A-M. The signal converter 5 includes M FIR (Finite Impulse Response) filters 51 (signal converter) and an adder 52 corresponding to each speaker 7. The FIR filter 51-1-m corresponding to the speaker 7-l (1 is an integer of 1 or more and L or less) and to which the audio signal A-m is input is described as FIR filter 51-l-m. Further, the adder 52 corresponding to the speaker 7-l will be referred to as an adder 52-l.

  The FIR filter 51 is a digital filter. The FIR filter 51-1m multiplies each frequency component of the audio signal A-m by a complex weighting coefficient in the frequency domain (performs a convolution operation in the time domain), and adds the audio signal of the multiplication (convolution) result to an adder. 52-l. By using the complex weight coefficient, both the amplitude and the phase of the voice signal can be controlled. The adder 52-1 adds the audio signals output from each of the FIR filters 51-1-1 to 51-I-M and outputs the audio signal of the addition result to the speaker 7-1. The speaker 7-l reproduces the audio signal input from the adder 52-l.

  FIG. 2 is a functional block diagram showing the configuration of the signal conversion coefficient calculation device 1, and only the functional blocks related to this embodiment are extracted and shown. The signal conversion coefficient calculation device 1 shown in the figure includes an input unit 11, a calculation unit 12, and an output unit 13.

  The input unit 11 inputs various kinds of information to the calculation unit 12. The sound field of the conversion destination environment is formed by the sound waves emitted from the speakers 7-1 to 7-L. For each of the audio signals A-1 to AM, the calculation unit 12 determines that the wavefront normal vectors formed at each of the plurality of listening positions in the sound field of the conversion destination environment are close to each other, and the normal vectors are A complex weighting coefficient used for conversion of the audio signal input to each speaker 7 is calculated so as to be close to the normal vector of the desired wavefront. Therefore, the calculation unit 12 sets the variance of the normal vector of the composite wavefront formed at each of the plurality of listening positions in the sound field of the conversion destination environment as an evaluation function, and reduces the value calculated by this evaluation function. A complex weighting coefficient corresponding to each speaker 7 is calculated. At this time, the calculation unit 12 determines the difference between the normal vector of the wavefront formed at each of the plurality of listening positions and the desired normal vector, or the angle formed between each normal vector and the desired normal vector. Is a constraint condition that is within a predetermined range. The calculation unit 12 calculates the normal vector of the wavefront formed at the listening position in the sound field of the conversion destination environment using the complex weighting coefficient corresponding to each speaker. The listening position is the position of the listening point. The desired wavefront is the wavefront at the listening position when the speaker 3 in the conversion source environment reproduces the audio signal, but the user may arbitrarily determine the desired wavefront. The output unit 13 outputs the complex weighting coefficient calculated by the calculation unit 12 to the signal converter 5. The complex weighting coefficient is calculated for each angular frequency bin, and the inverse DFT (Discrete Fourier Transform) of the complex weighting coefficient calculated for each angular frequency bin is applied to each FIR filter 51 of the signal converter 5. It is implemented as a filter coefficient. The angular frequency bin is the angular frequency discretized by the DFT.

  The calculation unit 12 includes a normal vector calculation unit 121, an evaluation function calculation unit 122, a determination unit 123, and a complex weight coefficient calculation unit 124. The details of each unit will be described below.

  According to the theory described in Non-Patent Document 1, when content is viewed in a synthesized sound field of sound waves emitted from a plurality of sound sources, a sound image is localized in the direction of the wavefront normal at the listening position in the synthesized sound field. The calculation unit 12 keeps the variance of the normal vector of the composite wavefront formed at each of the plurality of listening positions in the sound field of the conversion destination environment as small as possible, and then performs conversion based on the theory described in Non-Patent Document 1. A normal vector of the wavefront at the listening position when the audio signal is reproduced by the speaker 3 having the original environment, and a plurality of listening positions or a certain region when the audio signal is reproduced at each position of the plurality of speakers 7 of the conversion destination environment. The complex weighting coefficient for the audio signal input to each speaker 7 in the conversion destination environment is determined so that the normal vector of the wave front in the inside of the inside becomes as close as possible. That is, the calculation unit 12 outputs the audio signal A-m to the speakers 7-1 to 7-L in the frequency domain set in each of the FIR filters 51-1-m to 51-L-m. A complex weighting coefficient to be multiplied with the audio signal is calculated. The audio system 100 reproduces an audio signal in which the FIR filter 51 corresponding to each speaker 7 multiplies a complex weighting coefficient in the frequency domain (convolution operation in the time domain is performed). As a result, the audio system 100 maintains the quality at a plurality of positions as much as possible even within a plurality of listening positions or in a certain area of the conversion destination environment, and reproduces it in the speaker arrangement of the conversion source environment and the localization direction of the sound image. The audio content can be played back without changing.

Of the L loudspeakers 7 in the conversion destination environment, the position vector of the n-th (n is an integer of 1 or more and L or less) speaker 7-n is set to r _sn = (x _n , y _n , z _n ) and the n-th position _Let a _n (ω) be the complex weighting coefficient corresponding to the speaker 7. ω is the angular frequency. Sum of sound pressures p ^ (r, ω) at the position of position vector r = (x, y, z) obtained by sound waves emitted from N (1 ≦ N ≦ L) speakers 7 in the conversion destination environment Can be expressed by the following formula (1). Here, it is assumed that all sound sources can be regarded as point sound sources.

  j is an imaginary number, c is the speed of sound, and t is time. By modifying the equation (1), the following equation (2) is obtained.

Note that Re [•] indicates the real part of ・, and Im [•] indicates the imaginary part of ・. Φ (r, ω) represents the phase at a position at a certain time of the synthetic sound field reproduced from the N speakers 7. At this time, a plurality of listening points (listening positions) in which the position vector in the conversion destination environment is r _p ′ = (x _p ′, y _p ′, z _p ′) (p = 1, 2, 3, ..., P) The normal vector G (r _p ′, ω) of the wavefront of the synthetic sound field in can be expressed by the following equation (3).

Note that | _{r = rp ′} represents that the value of r _p ′ is substituted for r.

The normal vector calculation unit 121 calculates the normal vector G (r _p ′, ω) of the wavefront of the synthetic sound field at each listening position of the conversion destination environment by the expression (3), and the calculation result is the evaluation function calculation unit 122. Output to. The position vector of each speaker 7 and the position vector of each listening position used for the calculation of Expression (3) are obtained from the information of the position of each speaker 7 and each listening position input from the input unit 11.

Considering to listen uniformly at multiple points, under the constraint that the wavefront normal vector at each of the multiple points in the synthesized sound field of the conversion destination environment faces a direction close to the desired normal vector, those multiple points It is desirable that the variance of the normal vector at is as small as possible. Therefore, a desired normal vector at a plurality of listening positions of the position vector r _p ′ = (x _p ′, y _p ′, z _p ′) (p = 1, 2, 3, ..., P) is defined as F _p . At this time, the complex weighting coefficient calculation unit 124 performs complex optimization by performing constraint optimization that minimizes the evaluation value calculated by the evaluation function J of Expression (4) using the following Expression (5) as a restriction condition. determining a weighting factor _{a n.}

The evaluation function J of Expression (4) is an expression for calculating the variance of the normal vector of the wavefront at each of the plurality of listening positions in the conversion destination environment. The left side of the constraint condition shown in Expression (5) represents the square norm of the difference vector between the desired normal vector F _p and the normal vector G (r _p ′, ω) at the listening position. The threshold value k _p corresponding to the p-th (p = 1, 2, 3, ..., P) listening position may be set in advance or may be arbitrarily determined by the user.

  Further, the constraint condition may be expressed by the following expression (6).

Here, [·] ^T represents transposition of a vector or a matrix. The left side of Expression (5) represents the cosine of the angle formed by the desired normal vector F _p and the normal vector G (r _p ′, ω) at the p-th listening position. The threshold value cos θ _p corresponding to the p-th (p = 1, 2, 3, ..., P) listening position may be set in advance or may be arbitrarily determined by the user.

Evaluation function calculation unit 122, based on the respective listening positions the vector generator 121 is calculated using the complex weight coefficients sequentially calculated by the complex weight coefficient calculator _{124 G (r p ', ω} ), of the The value of the evaluation function J is calculated.

Further, in this case, the calculation unit 12 gives the initial value a _n0 (ω) of the complex weighting coefficient a _n (ω), and recursively minimizes the evaluation function J to obtain the optimum a _n (ω). To explore. The user can arbitrarily set the initial value a _n0 (ω) of the complex weighting coefficient a _n (ω).

The determination unit 123 calculates the difference between the current evaluation value and the previous evaluation value when the calculation unit 12 recursively calculates the complex weighting coefficient a _n (ω). Current evaluation value is based on the G of each listening position calculated by using the current complex weight coefficient _{a n (ω) (r p} ', ω), the evaluation function calculating unit 122, is calculated by the evaluation function J It is an evaluation value. The previous evaluation value is an evaluation function calculated based on G (r _p ′, ω) calculated using the complex weight coefficient a _n (ω) immediately before being updated to the current complex weight coefficient a _n (ω). The part 122 is the evaluation value calculated by the evaluation function J. When the determination unit 123 determines that the difference is less than the predetermined threshold value, it determines the current complex weighting coefficient a _n (ω) as the final complex weighting coefficient. The threshold may be given in advance or may be arbitrarily determined by the user.

The complex weight coefficient calculation unit 124 sequentially calculates the optimum complex weight coefficient a _n (ω) (n = 1, 2, 3, ..., L) that minimizes the evaluation function J. This makes it possible to realize signal conversion for reproducing the audio content in a homogeneous listening environment at a plurality of listening positions. When sound waves are emitted from the N speakers set by the user out of the L speakers 7, the complex weighting coefficient calculation unit 124 determines the complex weighting coefficient a for the (L−N) speakers 7 not selected. _{Let n} = 0. Thus, for example, even when the user selects N (<L) speakers 7, L complex weight coefficients a _n (ω) are calculated, and the evaluation function does not reach the global optimum value. However, no sound is played from the apparently unnecessary speaker.

When the desired normal vector is the normal vector of the wavefront at the listening position in the conversion source environment, the normal vector calculation unit 121 calculates the normal of the wavefront in the conversion source environment from the speaker position and the listening position of the conversion source environment. The vector is also calculated. When the position vector of the i (i = 1, 2, 3, ..., M) -th speaker 3 in the speaker arrangement of the conversion source environment is r _si = (x _si , y _si , z _si ), r = (x , y, z), the sound pressure p _i (r, ω) is expressed by the following equation (7).

j is an imaginary unit, c is the speed of sound, and t is time. Here, Φ _0i (r, ω) is represented by the following expression (8).

Φ _0i (r, ω) represents the phase at a certain position at a certain time. At this time, the normal vector F _i (r ₁ , ω) of the wavefront at r ₁ = (x ₁ , y ₁ , z ₁ ) can be expressed as the following Expression (9).

  The desired normal vector may be a normal vector of the wavefront at any listening position in any environment different from the conversion source environment and the conversion destination environment. Information on the speaker position and the listening position of the conversion source environment or an arbitrary environment for calculating the desired normal vector by the expression (9) is input by the input unit 11. Further, the desired normal vector may be determined in advance or may be arbitrarily determined by the user.

  When the complex weighting coefficient is calculated by the above method, the amplitude is not compensated. Therefore, it goes without saying that the normalization is performed so that the amplitudes of the audio channels are the same.

  FIG. 3 is a flowchart showing the complex weight coefficient calculation process executed by the signal conversion coefficient calculation device 1. The signal conversion coefficient calculation device 1 performs the complex weighting coefficient calculation process shown in the figure for each of the audio signals A-1 to AM.

  First, the input unit 11 inputs the position of each speaker 7 in the conversion destination environment, the plurality of listening positions in the conversion destination environment, the initial value of the complex weighting factor, and the information about the determination threshold (step S110). The evaluation function calculation unit 122 acquires the desired normal vector (step S120). The desired normal vector may be set in the calculation unit 12 in advance or may be input by the input unit 11. Alternatively, the normal vector calculation unit 121 may calculate a desired normal vector by Expression (9) based on the position of each speaker and the listening position in the conversion source environment or any environment input by the input unit 11. ..

The normal vector calculation unit 121 uses the position of each speaker 7 in the conversion destination environment, the listening position in the conversion destination environment, and the current complex weighting coefficient to synthesize at each listening position in the conversion destination environment according to Expression (3). A normal vector G (r _p ′, ω) of the wavefront of the sound field is calculated (step S130). The initial value of the complex weighting coefficient is used to calculate the first normal vector G (r _p ′, ω).

The evaluation function calculation unit 122 uses the normal vector G (r _p ′, ω) calculated by the normal vector calculation unit 121 and the desired normal vector acquired in step S120 to calculate the evaluation function shown in Expression (4). An evaluation value is calculated by J (step S140).

  The determination unit 123 determines whether or not the difference between the evaluation value calculated using the current complex weighting coefficient and the evaluation value calculated using the previous complex weighting coefficient is equal to or more than a threshold value (step S150). ). When the determination unit 123 determines that the difference is equal to or greater than the threshold value (step S150: YES), the determination unit 123 instructs the complex weight coefficient calculation unit 124 to calculate the complex weight coefficient.

  The complex weight coefficient calculation unit 124 updates the value of the complex weight coefficient (step S160). For example, the complex weighting factor calculation unit 124 changes a part or all of the values of the complex weighting factor and outputs it to the normal vector calculation unit 121. The normal vector calculation unit 121 calculates a normal vector by the complex weight coefficient changed by the complex weight coefficient calculation unit 124, and outputs the calculation result to the complex weight coefficient calculation unit 124. When the calculated normal vector satisfies the constraint condition shown in Expression (5) or Expression (6), the complex weight coefficient calculation unit 124 sets the changed value as a new complex weight coefficient, and the normal vector calculation unit 121 Output to. The calculation unit 12 repeats the processing from step S130.

  When the determination unit 123 determines that the difference between the evaluation value calculated using the current complex weighting coefficient and the evaluation value calculated using the previous complex weighting coefficient is less than the threshold value (step S150). : NO), the current complex weighting coefficient is output (step S170). The output unit 13 outputs the complex weighting coefficient output by the determination unit 123 to the outside of the signal conversion coefficient calculation device 1.

Note that the calculation unit 12 may calculate a filter array to be mounted on the FIR filter 51 and output it to the output unit 13 as follows. That is, the calculation unit 12 calculates the complex weighting coefficient for each of the N speakers 7-n for each angular frequency bin for the audio signal A-i of the i-th channel (i is an integer of 1 or more and M or less). The angular frequency bin is the angular frequency discretized by the DFT. The calculation unit 12 generates a complex weight coefficient matrix a _ni (ω) in which the calculated complex weight coefficients are arranged as elements. The calculation unit 12 calculates a filter string to be mounted on the FIR filter 51-n-i by performing an inverse DFT on the complex weighting coefficient matrix a _ni (ω). In this way, the calculation unit 12 calculates the filter coefficient for the audio signal input to each of the plurality of speakers 7. The calculation unit 12 calculates a complex weighting coefficient matrix for all channels i (i = 1, 2, 3, ..., M) in the conversion source environment, and mounts the calculated result in the signal converter 5 as an FIR filter matrix. be able to.

  According to the above-described embodiment, the signal conversion coefficient calculation device 1 enables near-homogeneous listening at a plurality of listening positions or a certain region when reproducing the audio content in an environment different from the environment at the time of producing the audio content. It is possible to calculate a complex weighting coefficient which is a signal conversion coefficient for At this time, for example, the signal conversion coefficient calculation device 1 reproduces the audio content so that the plurality of listening positions or a certain area in the environment different from the environment when the audio content is produced have an impression close to the environment at the time of production. The signal conversion coefficient can be calculated.

  The signal conversion coefficient calculation device 1 described above has a computer system inside. The process of the operation of the signal conversion coefficient calculation device 1 is stored in a computer readable recording medium in the form of a program, and the above process is performed by the computer system reading and executing this program. The computer system mentioned here includes a CPU, various memories, an OS, and hardware such as peripheral devices.

  Further, the “computer system” also includes a homepage providing environment (or display environment) when using a WWW system.

  Further, the “computer-readable recording medium” refers to a portable medium such as a flexible disk, a magneto-optical disk, a ROM, a CD-ROM, or a storage device such as a hard disk built in a computer system. Further, the "computer-readable recording medium" means to hold a program dynamically for a short time like a communication line when transmitting the program through a network such as the Internet or a communication line such as a telephone line. In this case, a volatile memory inside a computer system that serves as a server or a client in that case holds a program for a certain period of time. Further, the program may be for realizing a part of the functions described above, or may be a program for realizing the functions described above in combination with a program already recorded in the computer system.

DESCRIPTION OF SYMBOLS 1 ... Signal conversion coefficient calculation device 3-1 to 3-M, 7-1 to 7-L ... Speaker 5 ... Signal converter 11 ... Input part 12 ... Calculation part 13 ... Output part 51-1-1 to 51-L -M ... FIR filters 52-1 to 51-L ... Adder 121 ... Normal vector calculation unit 122 ... Evaluation function calculation unit 123 ... Judgment unit 124 ... Complex weight coefficient calculation unit

Claims (6)

Conversion of audio signals output from the plurality of speakers so that normal vectors of wavefronts formed at a plurality of listening positions in a sound field of a reproduction environment synthesized by sound waves emitted from the plurality of speakers are close to each other. A calculation unit that calculates a complex weighting coefficient to be used for each speaker,
The calculation unit calculates the complex weighting coefficient such that the variance of the normal vector of the wavefront formed at each of the plurality of listening positions is small.
An apparatus for calculating a signal conversion coefficient, comprising: The calculating unit, the arbitrary norm of the difference between the normal vector of the wavefront and the normal vector of the desired wavefront formed at each of the plurality of listening positions, the complex with the constraint that it is within a predetermined range. Calculate the weighting factor,
The signal conversion coefficient calculation device according to claim 1, wherein The calculation unit, the angle formed by the normal vector of the wavefront and the normal vector of the desired wavefront formed at each of the plurality of listening positions, the complex weighting coefficient under the constraint that the angle is within a predetermined range. To calculate,
The signal conversion coefficient calculation device according to claim 1, wherein The calculation unit calculates a normal vector of the desired wavefront based on a speaker position and a listening position in an environment different from the reproduction environment,
The signal conversion coefficient calculation device according to claim 2 or 3, characterized in that. A signal conversion coefficient calculation device according to any one of claims 1 to 4,
A signal conversion unit that converts an audio signal to be output from each of the plurality of speakers, based on a complex weighting coefficient calculated by the signal conversion coefficient calculation device for each of the plurality of speakers;
A signal conversion device comprising:   A program that causes a computer to function as the signal conversion coefficient calculation device according to any one of claims 1 to 4.

Images