JP5310064B2 - Impulse response processing device, reverberation imparting device and program - Google Patents

Description

  The present invention relates to a technique for processing an impulse response used for imparting reverberation.

  In a device that adds reverberation to an acoustic signal by convolution calculation of an impulse response, a technique for changing the length of time during which reverberation continues (hereinafter referred to as “reverberation time”) has been proposed. For example, Patent Document 1 discloses a technique for generating a new impulse response having a desired reverberation time by multiplying each of two types of impulse responses by an exponential function and then adding (linear combination).

JP 2004-294712 A

  However, in the technique of Patent Document 1, by multiplying the exponential function to increase the intensity of the impulse response, the intensity of noise (background noise) superimposed on the impulse response is also amplified. Therefore, there is a problem that the quality of the reverberant sound is degraded. In view of the above circumstances, an object of the present invention is to change the reverberation time while maintaining the sound quality of the reverberant sound.

  In order to solve the above-described problems, the impulse response machining apparatus according to the first aspect of the present invention is different from each other in waveform classification means for dividing the first impulse response into a plurality of basic blocks on the time axis. An interpolation length setting means for setting an interpolation length, a time adjusting means for expanding a time difference between each of a plurality of basic blocks and a basic block immediately after the basic block to an interpolation length set by the interpolation length setting means, and an interpolation block Interpolating processing means for generating the second impulse response by arranging the interpolating block between the basic blocks processed by the time adjusting means.

  In the above configuration, since the reverberation time is extended by increasing the time difference of each basic block that divides the first impulse response, compared with the configuration in which the reverberation time is extended by increasing the amplitude of the first impulse response. The second impulse response of high-quality reverberation sound with suppressed noise can be generated. In addition, since an interpolation block is arranged between each basic block, the natural reverberation sound of the natural audible sound is compared with the case where the interval between the basic blocks is simply expanded to generate the second impulse response. It is possible to generate a two impulse response. Furthermore, since the time difference (interpolation length) of each basic block after processing by the time adjusting means is different, the frequency component resulting from the period of the arrangement of the basic block and the interpolation block is dispersed. Therefore, as compared with a configuration in which the time difference between the basic blocks is increased by a common interpolation length, there is an advantage that the sense of pitch of the reverberant sound due to the period of the arrangement of the basic blocks and the interpolation blocks is reduced.

  The impulse response processing apparatus according to the second aspect of the present invention includes waveform dividing means for dividing the first impulse response into a plurality of basic blocks on the time axis, and different first and second interpolation lengths. Interpolation length setting means for setting a plurality of interpolation lengths and means for expanding the time difference between two adjacent basic blocks, wherein the time difference between the first basic block and the immediately following basic block among the plurality of basic blocks is calculated. A time adjustment unit that expands to the first interpolation length and expands the time difference between the second basic block different from the first basic block and the next basic block among the plurality of basic blocks to the second interpolation length, and generates an interpolation block Interpolating processing means, and waveform synthesizing means for generating a second impulse response by disposing an interpolating block between each basic block processed by the time adjusting means. Also in the above configuration, there is an advantage that a natural reverberation sound in which noise is suppressed can be generated as in the impulse response processing apparatus according to the first aspect. Further, the first interpolation length after adjusting the time difference between the first basic block and the immediately following basic block is different from the second interpolation length after adjusting the time difference between the second basic block and the immediately following basic block. Compared with the configuration in which the time differences of the basic blocks after adjustment by the time adjusting means are common, the frequency components resulting from the arrangement of the basic blocks and the interpolation blocks are dispersed as in the first mode. Therefore, there is an advantage that the sense of pitch of reverberant sound due to the period of the basic block and the interpolation block is reduced.

  In a preferred aspect of the present invention (first aspect and second aspect), the interpolation processing means calculates an interpolation block having a time length equal to the interpolation length of the two basic blocks for each of two successive basic blocks. Generate. According to the above aspect, since the time difference of each basic block after adjustment by the time adjusting means matches the section length of the interpolation block, the basic block and the interpolation block are continuously arranged. Therefore, there is an advantage that it is possible to generate a second impulse response of a reverberant sound that is natural on hearing.

  In a preferred aspect of the present invention, the interpolation processing means generates an interpolation block from the basic block after the division by the waveform dividing means. In the above aspect, since an interpolation block (typically an interpolation block whose acoustic characteristics are similar to the basic block) based on the basic block is generated, a second impulse response of reverberant sound that is natural on hearing can be generated. There is an advantage.

In a preferred aspect of the present invention, the interpolation length setting means includes a random number generation means for generating a random number, and an arithmetic means for sequentially setting each of the plurality of interpolation lengths from the random number. In the above aspect, it is possible to disperse a plurality of interpolation lengths simply and appropriately. In a further preferred aspect, the interpolation length setting means includes range setting means for setting a random number distribution range in accordance with an expansion rate of the second impulse response with respect to the first impulse response. According to the above aspect, it is possible to appropriately limit the range in which each interpolation length is distributed.

  In a preferred aspect of the present invention, the interpolation length setting means sets the plurality of interpolation lengths so that the distribution range of the plurality of interpolation lengths becomes wider as the expansion rate of the second impulse response with respect to the first impulse response is higher. According to the above aspect, while suppressing the difference in acoustic characteristics between the first impulse response and the second impulse response when the expansion rate is low, the sense of pitch of the reverberant sound that is particularly problematic when the expansion rate R is high is obtained. It can be reduced.

  The reverberation imparting apparatus according to the present invention includes the impulse response processing apparatus according to each of the above aspects, and reverberation imparting means for performing a convolution operation between the second impulse response generated by the impulse response processing apparatus and the acoustic signal. . According to the reverberation imparting device of the present invention, the same operations and effects as those of the impulse response machining device according to each of the above aspects are realized.

The impulse response processing apparatus according to each of the above aspects is realized by hardware (electronic circuit) such as a DSP (Digital Signal Processor) dedicated to impulse response processing, as well as a CPU (Central Processing Unit) and the like. It is also realized by cooperation between a general-purpose arithmetic processing unit and a program . The program of the present invention is provided to a user in a form stored in a computer-readable recording medium and installed in the computer, or provided from a server device in a form of distribution via a communication network and installed in the computer. Is done.

It is a block diagram of the reverberation provision apparatus which concerns on 1st Embodiment of this invention. It is a block diagram of an impulse response processing unit. It is a conceptual diagram for demonstrating operation | movement of an impulse response processing part. It is a block diagram of an interpolation processing unit. It is a conceptual diagram for demonstrating the processing of the impulse response in contrast 1. It is a block diagram of an interpolation length setting unit. It is a conceptual diagram which shows the range where interpolation length is distributed. It is a block diagram of the interpolation length setting part in 2nd Embodiment. It is a block diagram of the interpolation process part in 3rd Embodiment. It is a conceptual diagram for demonstrating operation | movement of a waveform process part. It is a conceptual diagram for demonstrating operation | movement of 4th Embodiment.

<A: First Embodiment>
FIG. 1 is a block diagram of a reverberation imparting apparatus according to the first embodiment of the present invention. As shown in FIG. 1, a signal supply device 92 and a sound emitting device 94 are connected to the reverberation imparting device 100. The signal supply device 92 supplies an acoustic signal SIN representing a sound (musical tone or voice) waveform to the reverberation applying device 100. For example, a sound collection device that generates an acoustic signal SIN corresponding to the surrounding sound and a playback device that acquires and outputs the acoustic signal SIN from a recording medium are employed as the signal supply device 92. The reverberation imparting device 100 generates and outputs a reverberation sound signal SR obtained by adding reverberation to the acoustic signal SIN. The reverberant sound signal SR is reproduced as a sound wave by being supplied to a sound emitting device 94 (for example, a speaker or headphones).

  As illustrated in FIG. 1, the reverberation imparting device 100 is a computer system including an arithmetic processing device 12 and a storage device 14. The storage device 14 stores a program executed by the arithmetic processing device 12 and data used by the arithmetic processing device 12. For example, a sample series (coefficient calculation coefficient sequence) representing the waveform of the initial impulse response H 1 is stored in the storage device 14. The initial impulse response H1 is generated in advance by, for example, collecting a reverberation sound when an impulse sound is generated in a specific acoustic space. A known recording medium such as a semiconductor storage device or a magnetic storage device is arbitrarily adopted as the storage device 14.

  The arithmetic processing unit 12 functions as a plurality of elements (impulse response processing unit 22 and reverberation applying unit 24) by executing a program stored in the storage device 14. In addition, a configuration in which each element of the arithmetic processing unit 12 is mounted in a distributed manner on a plurality of devices (integrated circuits), or a configuration in which an electronic circuit (DSP) dedicated to processing of the acoustic signal SIN realizes each element is also adopted. The

  The impulse response processing unit 22 processes a new impulse response (hereinafter referred to as “new impulse response”) H2 having a reverberation time different from that of the initial impulse response H1 by processing the initial impulse response H1 stored in the storage device 14. Generate. Specifically, the new impulse response H2 is a sample sequence of a waveform obtained by extending the reverberation time to R times the initial impulse response H1 (1 <R ≦ 2). The reverberation imparting unit 24 generates a reverberation sound signal SR by performing a filtering process (convolution operation) using the new impulse response H2 generated by the impulse response processing unit 22 on the acoustic signal SIN. A known technique is arbitrarily employed for the generation of the reverberation sound signal SR by the reverberation imparting unit 24.

  An input device 16 is connected to the reverberation imparting device 100. The input device 16 is configured by an operator operated by a user for inputting an instruction to the reverberation imparting device 100. The user can variably specify the expansion rate R of the reverberation time according to the operation on the input device 16.

  FIG. 2 is a block diagram of the impulse response processing unit 22, and FIG. 3 is a conceptual diagram for explaining specific processing by the impulse response processing unit 22. As shown in FIG. 2, the impulse response processing unit 22 includes a waveform division unit 32, a windowing unit 34, a time adjustment unit 36, an interpolation length setting unit 38, an interpolation processing unit 42, and a waveform synthesis unit 44. The

  As shown in part (A) of FIG. 3, the waveform classification unit 32 converts the initial impulse response H1 stored in the storage device 14 into (M + 1) basic blocks Ba [1] to Ba [ M + 1] (M is a natural number). The section length (number of samples) of each of the basic blocks Ba [1] to Ba [M + 1] corresponds to 2N samples (for example, N = 64) of the initial impulse response H1. In addition, the time difference between the basic block Ba [i] (i = 1 to M) and the basic block Ba [i + 1] that follow each other on the time axis corresponds to N samples. Therefore, the basic block Ba [i] and the basic block Ba [i + 1] partially overlap. That is, the latter half N samples in each basic block Ba [i] and the first half N samples in the basic block Ba [i + 1] are common.

  The windowing unit 34 in FIG. 2 multiplies each basic block Ba [1] to Ba [M + 1] divided by the waveform dividing unit 32 by a window function wA as shown in part (B) of FIG. (M + 1) basic blocks Bb [1] to Bb [M + 1] are generated. The section length of each of the basic blocks Bb [1] to Bb [M + 1] corresponds to 2N samples. A function (for example, a Hanning window) whose function value decreases as it approaches the both ends is suitable as the window function wA. In part (B) of FIG. 3, the basic blocks Bb [1] to Bb [M + 1] after multiplication by the window function wA are alternatively illustrated by a graphic representing the shape of the window function wA. Further, even-numbered basic blocks Bb (Bb [2], Bb [4],...) And odd-numbered basic blocks Bb (Bb [1], Bb [3],. For the sake of distinction, the broken lines are shown for convenience.

  The interpolation length setting unit 38 in FIG. 2 sets M different interpolation lengths L [1] to L [M]. The interpolation length L [i] set by the interpolation length setting unit 38 is used for adjusting the position of each basic block Bb [i] by the time adjustment unit 36 and for generating the interpolation block P [i] by the interpolation processing unit 42. Is done. A specific configuration of the interpolation length setting unit 38 and a method for setting the interpolation lengths L [1] to L [M] will be described later.

  The time adjusting unit 36 in FIG. 2 adjusts the positions of the basic blocks Bb [1] to Bb [M + 1] on the time axis. Specifically, as shown in part (C) of FIG. 3, the time difference between the basic block Bb [i] and the basic block Bb [i + 1] (the midpoint C [i] of the basic block Bb [i] The interval between the midpoint C [i + 1] of the basic block Bb [i + 1] is set by the interpolation length setting unit 38 based on the section length N (for N samples) after processing by the windowing unit 34. The time adjustment unit 36 adjusts the positions of the basic blocks Bb [1] to Bb [M + 1] on the time axis so that the interpolation length L [i] is expanded. For example, as shown in part (C) of FIG. 3, the time difference between the basic block Bb [1] and the basic block Bb [2] is expanded to the interpolation length L [1]. Further, the time difference between the basic block Bb [2] and the basic block Bb [3] is expanded to the interpolation length L [2] (L [2] <L [1]), and the basic block Bb [3] and the basic block Bb. The time difference from [4] is expanded to the interpolation length L [3] (L [3]> L [1]).

  The interpolation processing unit 42 of FIG. 2 generates M interpolation blocks P [1] to P [M]. The interpolation block P [i] is a sample series (a set of L [i] samples) in which the section length is adjusted to the interpolation length L [i], as shown in part (D) of FIG. That is, the interpolation processing unit 42 has an interval length equal to the time difference (interpolation length L [i]) between the basic block Bb [i] adjusted by the time adjusting unit 36 and the basic block Bb [i + 1]. Generate [i]. For example, as shown in part (D) of FIG. 3, the interpolation block P [1] is adjusted to the interpolation length L [1] between the basic block Bb [1] and the basic block Bb [2], and the interpolation block P [2] is adjusted to an interpolation length L [2] between the basic block Bb [2] and the basic block Bb [3]. The interpolation block P [i] is generated from the basic block Ba [i] and the basic block Ba [i + 1] after the division by the waveform classification unit 32.

  As shown in FIG. 4, the interpolation processing unit 42 includes a waveform merging unit 52 and a waveform adjusting unit 54. The waveform merging unit 52 generates the interpolation block Pa [i] by merging the basic block Ba [i] and the basic block Ba [i + 1] that are in succession on the time axis. Specifically, the interpolation block Pa [i] is an average value of the samples of the basic block Ba [i] and the basic block Ba [i + 1] (that is, the basic block Ba [i] and the basic block Ba [i]. i + 1] and 2N samples corresponding to the average value of the time samples corresponding to each other.

  The waveform adjustment unit 54 generates the interpolation block P [i] by adjusting the section length (time length corresponding to 2N samples) of the interpolation block Pa [i] to the interpolation length L [i]. Specifically, the waveform adjustment unit 54 multiplies the interpolation block Pa [i] by the window function wB (for example, Hanning window) whose window width is set to the interpolation length L [i], thereby interpolating the block P [i]. Is generated.

  The waveform synthesizer 44 shown in FIG. 2 is arranged between each of the basic blocks Bb [1] to Bb [M + 1] after being adjusted by the time adjusting unit 36, as shown in part (E) and part (F) of FIG. In addition, a new impulse response H2 is generated by arranging each of the interpolation blocks P [1] to P [M] generated by the interpolation processing unit 42. The interpolation block P [i] is arranged between the basic block Bb [i] and the basic block Bb [i + 1]. Since the time difference between the basic block Bb [i] and the basic block Bb [i + 1] and the section length of the interpolation block P [i] are both the interpolation length L [i], they are shown in part (E) of FIG. Thus, the starting point of the interpolation block P [i] matches the midpoint C [i] of the basic block Bb [i], and the end point of the interpolating block P [i] is within the basic block Bb [i + 1]. The position on the time axis of the interpolation block P [i] is selected so as to coincide with the point C [i + 1].

  The waveform synthesizer 44 at the same time for each of the basic blocks Bb [1] to Bb [M + 1] and each of the interpolation blocks P [1] to P [M] arranged on the time axis as described above. Add the values for each corresponding sample. The new impulse response H2 shown in part (F) of FIG. 3 is the sum of the samples of the basic blocks Bb [1] to Bb [M + 1] and the samples of the interpolation blocks P [1] to P [M]. It is a time series.

  Since the new impulse response H2 is generated by adding the basic blocks Bb [1] to Bb [M + 1] and the interpolation blocks P [1] to P [M] as described above, the waveform adjustment in FIG. If the section 54 simply adjusts the section length of the interpolation block Pa [i] to the interpolation length L [i], the amplitude of the section corresponding to the interpolation block P [i] in the new impulse response H2 may be excessive. There is sex. Furthermore, the longer the interval in which the basic block Bb [i] and the basic block Bb [i + 1] overlap on the time axis (that is, the smaller the interpolation length L [i]), the greater the amplitude of the new impulse response H2. The problem of growth becomes obvious. Therefore, the window function wB used by the waveform adjustment unit 54 in FIG. 4 is selected so that the amplitude (the numerical value of each sample) of the interpolation block Pa [i] is reduced. Specifically, as shown in part (D) of FIG. 3, the window function wB is set so that the amplitude of the interpolation block P [i] after multiplication by the window function wB decreases as the interpolation length L [i] decreases. The selected configuration is preferred.

  According to the first embodiment in which the time difference between each of the basic blocks Bb [1] to Bb [M + 1] is expanded, the reverberation time is obtained by multiplying the initial impulse response H1 by an exponential function as described below. There is an advantage that the deterioration of the sound quality of the reverberant sound after extension is suppressed as compared with the extension configuration (hereinafter referred to as “comparative 1”).

A new impulse response H2 ′ having a proportionality of 1 is generated by multiplying the initial impulse response H1 by an exponential function a (t), for example, as expressed by Equation (1). In contrast 1, as shown in FIG. 5, the amplification factor of the amplitude of the new impulse response H2 ′ with respect to the amplitude (intensity) of the initial impulse response H1 increases exponentially as it reaches the rear of the initial impulse response H1. The intensity of noise (background noise) superimposed on the rear part of the initial impulse response H1 becomes apparent in the new impulse response H2 ′. Therefore, there is a problem that the sound quality of the reverberant sound is deteriorated as the reverberation time of the new impulse response H2 ′ is extended as compared with the initial impulse response H1.

  On the other hand, in the first embodiment, the intensity of the initial impulse response H1 is not amplified, and the time difference between each of the plurality of basic blocks Bb [1] to Bb [M + 1] dividing the initial impulse response H1 is expanded. Since the new impulse response H2 is generated by performing (extending the initial impulse response H1 in the direction of the time axis), the problem of the proportional 1 that the noise of the initial impulse response H1 is manifested in the new impulse response H2 is Is prevented. Therefore, it is possible to extend the reverberation time while maintaining the sound quality of the reverberant sound. Moreover, since the interpolation block P [i] is arranged between the basic block Bb [i] and the basic block Bb [i + 1], the basic block Bb [i] and the basic block Bb [i + 1 are simply set. Compared with the case of generating a new impulse response H2 by expanding the time difference from the above, it is possible to generate a natural reverberant sound.

  Next, a specific configuration of the interpolation length setting unit 38 will be described with reference to FIG. As shown in FIG. 6, the interpolation length setting unit 38 includes a random number generation unit 62 and a calculation unit 64. The random number generator 62 generates M random numbers λ [1] to λ [M]. The random numbers λ [1] to λ [M] are random number sequences generated within a range from the lower limit value λ_min to the upper limit value λ_max so that the M average values become zero. Specifically, uniform random numbers in which each numerical value in the range from the lower limit value λ_min to the upper limit value λ_max is generated with the same probability, or the probability of occurrence of each numerical value in the range from the lower limit value λ_min to the upper limit value λ_max is approximately a normal distribution. The normal random numbers are suitably adopted as M random numbers λ [1] to λ [M]. The lower limit value λ_min and the upper limit value λ_max are set to predetermined values having the same absolute value (λ_min = −λ_max).

The calculation unit 64 sets the interpolation length L [i] (L [1] to L [M]) variably according to the random number λ [i]. As defined by the following equation (2), the interpolation length L [i] is an addition of the section length N (N samples) and the section length δ [i] corresponding to the random number λ [i]. Equivalent to. That is, the section length δ [i] in Equation (2) corresponds to an increase in the time difference between the basic block Bb [i] and the basic block Bb [i + 1]. The section length δ [i] is variably set according to the random number λ [i] generated by the random number generator 62 as defined by the following formula (3).
L [i] = N + δ [i] (2)
δ [i] = (R−1) · N · (1 + λ [i]) (3)

  The calculation unit 64 calculates the interpolation length L [i] from the random number λ [i] by executing the calculations of the formulas (2) and (3). The interpolation length L [i] calculated by the calculation unit 64 is applied to the adjustment of the position of each basic block Bb [i] by the time adjustment unit 36 and the generation of the interpolation block P [i] by the interpolation processing unit 42.

Since the average value (total value) of the M random numbers λ [1] to λ [M] is zero, the average value of the section lengths δ [1] to δ [M] is (R−1) · N. . Therefore, as understood from the equation (2), the average value of the M interpolation lengths L [1] to L [M] is N · R. Further, the minimum value L_min and the maximum value L_max of the interpolation length L [i] are expressed by the following equations (4a) and (4b). That is, as shown in FIG. 7, the interpolation lengths L [1] to L [M] range from L_min in the formula (4a) to L_max in the formula (4b) centering on the average value N · R (hereinafter “distribution”). Distributed within A) according to the random number λ [i].
L_min = N · {1+ (R−1) · (1 + λ_min)} (4a)
L_max = N · {1+ (R−1) · (1 + λ_max)} (4b)

  As understood from the equation (3) and FIG. 7, the higher the expansion rate R, the distribution range A of the interpolation lengths L [1] to L [M] (the distribution lengths δ [1] to δ [M] Range) becomes wider. For example, it is assumed that the lower limit value λ_min of the random numbers λ [1] to λ [M] is −0.5 and the upper limit value λ_max is +0.5. The distribution range A when the expansion rate R is set to 1.1 is a range (distribution width 0.1 N) from 1.05 N (L_min) to 1.15 N (L_max). On the other hand, the distribution range A when the expansion rate R is set to 1.5 is a range (distribution width 0.5 N) from 1.25 N (L_min) to 1.75 N (L_max). As illustrated above, the higher the expansion rate R, the higher the degree of dispersion of the interpolation lengths L [1] to L [M] (the distribution range A becomes wider).

Further, the time difference between the basic block Bb [i] and the basic block Bb [i + 1] is expanded from the section length N to the interpolation length L [i] by the processing by the time adjustment unit 36, and therefore the basic block Bb [i]. The local expansion rate R [i] at is expressed by the following equation (5). That is, the local expansion rate R [i] changes for each basic block Bb [i] in accordance with the random number λ [i].
R [i] = L [i] / N
= (N + δ [i]) / N
= {N + (R-1) .N. (1 + λ [i])} / N
= 1 + (R-1). (1 + λ [i]) (5)
However, since the average value of the M interpolation lengths L [1] to L [M] is N · R, the expansion rate R [i] changes for each basic block Bb [i]. Focusing on the entire impulse response H2, the reverberation time is expanded to R times the initial impulse response H1. Therefore, the user can arbitrarily adjust the reverberation time of the reverberation sound signal SR by operating the input device 16 and appropriately indicating the expansion rate R.

  Next, a configuration in which the interpolation lengths L [1] to L [M] are set to a common numerical value (for example, a time length (N · R) obtained by multiplying a time length corresponding to N samples by an expansion rate R) ( (Hereinafter referred to as “Comparison 2”) will be considered for comparison with this embodiment. In contrast 2, since the basic blocks Bb [1] to Bb [M + 1] and the interpolation blocks P [1] to P [M] are alternately arranged on the time axis at equal intervals, the new impulse response H2 The intensity of the frequency component corresponding to the period of the arrangement of the basic blocks Bb [1] to Bb [M + 1] and the interpolation blocks P [1] to P [M] becomes obvious. Therefore, the pitch feeling according to the period of the arrangement of the basic blocks Bb [1] to Bb [M + 1] and the interpolation blocks P [1] to P [M] is the reverberant sound SR. It may be perceived by the listener.

  In the first embodiment, the period (interpolation length L [i]) of the arrangement of the basic blocks Bb [1] to Bb [M + 1] and the interpolation blocks P [1] to P [M] changes every moment. Therefore, concentration of intensity for a specific frequency is relaxed (dispersed). Therefore, there is an advantage that the sense of pitch in the reverberant sound of the reverberant sound signal SR is reduced as compared with the proportional 2.

  Note that the sense of pitch due to the cycles of the basic blocks Bb [1] to Bb [M + 1] and the interpolation blocks P [1] to P [M] tends to become more apparent as the expansion rate R increases. In the first embodiment, the higher the expansion rate R, the wider the distribution range A of the interpolation lengths L [1] to L [M] (the variation in the interpolation lengths L [1] to L [M] increases). Also, there is an advantage that the sense of pitch can be effectively reduced even when the expansion rate R is high. On the other hand, as the distribution range A of the interpolation length L [i] is wider, the possibility that the acoustic characteristics of the new impulse response H2 deviate from the initial impulse response H1 increases. The difference in characteristics between the initial impulse response H1 and the new impulse response H2 is particularly easily perceived by the listener when the expansion rate R is low. In the first embodiment, the distribution range A of the interpolation length L [i] becomes narrower as the expansion rate R is lower. Therefore, when the expansion rate R is low, a new impulse response H2 that maintains the acoustic characteristics of the initial impulse response H1 is obtained. There is an advantage that it can be generated. That is, according to the first embodiment in which the distribution range A of the interpolation length L [i] is variably controlled according to the expansion rate R, the difference in acoustic characteristics between the initial impulse response H1 and the new impulse response H2 is determined as the expansion rate. While suppressing when R is low, it is possible to reduce the sense of pitch that is particularly problematic when the expansion rate R is high.

<B: Second Embodiment>
Next, a second embodiment of the present invention will be described. In addition, about the element in which an effect | action and a function are equivalent to 1st Embodiment in each following form, the same code | symbol as the above is attached | subjected and each detailed description is abbreviate | omitted suitably.

  FIG. 8 is a block diagram of the interpolation length setting unit 38. As shown in FIG. 8, the interpolation length setting unit 38 of the second embodiment has a configuration in which a range setting unit 66 is added to the first embodiment. The range setting unit 66 variably sets the distribution range (upper limit value λ_max, lower limit value λ_min) of the random numbers λ [1] to λ [M] according to the expansion rate R.

  Assume that the maximum value L_max of the interpolation length L [i] needs to be limited within a predetermined range. For example, since the difference in characteristics between the initial impulse response H1 and the new impulse response H2 increases as the interpolation length L [i] increases, in order to suppress the difference in characteristics between the initial impulse response H1 and the new impulse response H2. It is necessary to limit the maximum value L_max of the interpolation length L [i] to a predetermined range. Therefore, the range setting unit 66 variably sets the upper limit value λ_max of the random number λ [i] according to the expansion rate R so that the distribution range A of the interpolation length L [i] is limited to a predetermined range.

Assume that the maximum value L_max of the interpolation length L [i] is limited to 2N. The following equation (6b) is derived from the condition of the equation (6a) that the maximum value L_max of the equation (4b) is equal to 2N.
L_max = N · {1+ (R−1) · (1 + λ_max)} = 2N (6a)
λ_max = 1 / (R−1) −1 (6b)

  The range setting unit 66 sets the maximum value λ_max by executing the calculation of Equation (6b), and sets the minimum value λ_min by inverting the sign of the maximum value λ_max. That is, the distribution range (upper limit value λ_max, lower limit value λ_min) of the random number λ [i] is variably set according to the expansion rate R. The random number generation unit 62 generates a random number λ [i] within the range from the lower limit value λ_min set by the range setting unit 66 to the upper limit value λ_max.

  According to the second embodiment, there is an advantage that the interpolation length L [i] can be limited within a predetermined range. In the above, the maximum value λ_max is calculated from the expansion rate R. However, a configuration in which the minimum value λ_min is variably set according to the expansion rate R and the sign of the minimum value λ_min is inverted to calculate the maximum value λ_max is also suitable. It is.

<C: Third Embodiment>
In the first embodiment, the waveform adjustment unit 54 generates the interpolation block P [i] from the interpolation block Pa [i] obtained by merging the basic block Ba [i] and the basic block Ba [i + 1]. However, since the waveform of the interpolation block Pa [i] is similar to the waveform of the basic block Ba [i] and the basic block Ba [i + 1], the new impulse response H2 repeats the similar waveform on the time axis. Therefore, there is a possibility that a sense of incongruity (growing feeling) may be perceived in the reverberant sound of the reverberant sound signal SR.

  In order to solve the above problem, the interpolation processing unit 42 of the third embodiment includes a waveform processing unit 56 shown in FIG. The waveform processing unit 56 deforms the waveform of the interpolation block Pa [i] generated by the waveform merging unit 52 to generate the interpolation block Pb [i]. The waveform adjustment unit 54 generates an interpolation block P [i] from the interpolation block Pb [i] generated by the waveform processing unit 56 by the same method as in the first embodiment.

  As shown in FIG. 10, the waveform processing unit 56 generates the interpolation block Pb [i] by reversing the front and back of the waveform of the interpolation block Pa [i] on the time axis. That is, the interpolation block Pb [i] is a series of 2N samples obtained by reversing the order of 2N samples constituting the interpolation block Pa [i]. As described above, if the interpolation block P [i] generated through the processing by the waveform processing unit 56 is used, a waveform that is appropriately different in the new impulse response H2 is repeated. Therefore, the reverberation sound of the reverberation sound signal SR is It becomes natural sound in terms of hearing.

  In addition, the content of the process by the waveform processing part 56 is not limited to the above illustration. For example, a configuration in which the waveform processing unit 56 generates the interpolation block Pb [i] by rotating the phase of the interpolation block Pa [i] is also suitable. That is, the waveform processing unit 56 generates the interpolation block Pb [i] by rotating the phase of the frequency spectrum specified from the interpolation block Pa [i] by a predetermined angle and then converting it to the time domain. The above configuration also has an advantage that a reverberation signal SR of natural reverberation can be generated.

<D: Fourth Embodiment>
Next, a fourth embodiment of the present invention will be described. In 1st Embodiment, the case where the expansion rate R was 2 times or less was assumed. This form is a form for extending the reverberation time at an extension rate R exceeding twice. In this embodiment, when the expansion rate R is less than twice, the reverberation time is extended by the same processing as in the first to third embodiments.

  FIG. 11 is a conceptual diagram for explaining the operation of the fourth embodiment. When the expansion rate R exceeds 2 (for example, when R = 2.5), the interpolation length L [i] exceeds 2N samples of the initial impulse response H1. Therefore, if only one interpolation block P [i] composed of 2N samples is placed between the basic block Bb [i] and the basic block Bb [i + 1], the new impulse response H2 The intensity of the section corresponding to between the basic block Bb [i] and the basic block Bb [i + 1] is insufficient. Therefore, as shown in FIG. 11, the waveform synthesizer 44 includes a plurality of interpolation blocks P [i] (P [i] _1, P) between the basic block Bb [i] and the basic block Bb [i + 1]. [i] _2) is arranged to generate a new impulse response H2.

  The interpolation block P [i] _1 in FIG. 11 is a series of 2N samples generated from the basic block Ba [i] and the basic block Ba [i + 1]. The interpolation block P [i] _2 is a sequence of (L [i] −N) samples generated from the basic block Ba [i] and the basic block Ba [i + 1]. The waveform synthesis unit 44 arranges the interpolation block P [i] _1 and the interpolation block P [i] _2 between the basic block Bb [i] and the basic block Bb [i + 1].

  The waveform processing unit 56 of the interpolation processing unit 42 generates a plurality of interpolation blocks Pb [i] (Pb [i] _1, Pb [i] _2) having different frequency characteristics from the interpolation block Pa [ i]. For example, the waveform processing unit 56 generates two interpolation blocks Pb [i] (Pb [i] _1, Pb [i] _2) by rotating the phase of the interpolation block Pa [i] by different angles. . That is, the waveform processing unit 56 generates the interpolation block Pb [i] _1 by rotating the phase of the interpolation block Pa [i] by the angle θ1, and sets the phase of the interpolation block Pa [i] to the angle θ2 (θ2 ≠ θ1). ) To generate the interpolation block Pb [i] _2.

  The waveform adjustment unit 54 converts a plurality of interpolation blocks P [i] (P [i] _1) from a plurality of interpolation blocks Pb [i] (Pb [i] _1, Pb [i] _2) processed by the waveform processing unit 56. , P [i] _2). The interpolation block P [i] _1 is calculated by multiplying the window function wB_1 whose window width is set to the section length 2N and the interpolation block Pb [i] _1, and the interpolation block P [i] _2 has a window width of the section. It is calculated by multiplying the window function wB_2 set to the length (L [i] -N) and the interpolation block Pb [i] _2. Similar to the generation of the interpolation block P [i] in the first embodiment, the longer the interval in which the interpolation block Pb [i] _1 and the basic block Bb [i + 1] overlap on the time axis, the longer the interpolation block P [i]. The window function wB_2 is selected so that the amplitude of] _2 decreases.

  In the above configuration, a plurality of interpolation blocks P [i] (P [i] _1, P [i] _2) are arranged between the basic block Bb [i] and the basic block Bb [i + 1]. Therefore, it is possible to set the expansion rate R to 2 times or more. The number of interpolation blocks P [i] arranged between the basic block Bb [i] and the basic block Bb [i + 1] is appropriately changed according to the expansion rate R. For example, if the expansion rate R is 3 or more, three or more interpolation blocks P [i] are arranged between the basic block Bb [i] and the basic block Bb [i + 1].

<E: Modification>
Various modifications are added to the above embodiments. Specific modifications are exemplified below. Two or more aspects may be arbitrarily selected from the following examples and combined.

(1) Modification 1
In each of the above embodiments, the interpolation length L [i] is calculated from the random number λ [i]. However, the configuration using the random number λ [i] for calculating the interpolation length L [i] is not essential in the present invention. For example, a numerical sequence in which a plurality of different (M) numerical values are arranged in advance is prepared in advance, and the interpolation length setting unit 38 selects each numerical value in the numerical sequence in the order of arrangement and uses the numerical value to determine the interpolation length L. A configuration for calculating [i] is also suitable. That is, the interpolation length setting unit 38 is included as means for setting a plurality of different interpolation lengths (L [1] to L [M]), and what is a specific method for calculating the interpolation length L [i]? It is unquestionable.

(2) Modification 2
In the above embodiment, the case where each of the M interpolation lengths L [1] to L [M] is different is illustrated, but two or more interpolations among the plurality of interpolation lengths L [1] to L [M] are illustrated. A configuration having a common length L [i] is also employed. That is, among the basic blocks Bb [1] to Bb [M + 1], the interpolation length L [u] after the time difference between the basic block Bb [u] (u = 1 to M) and the basic block Bb [u + 1] is increased. ] Is different from the interpolation length L [v] after expansion of the time difference between the basic block Bb [v] (v ≠ u) and the basic block Bb [v + 1], the interpolation length L [u] Regardless of the difference in interpolation length L other than L and v [v], when all of the interpolation lengths L [1] to L [M] are compared with the common proportional 2, the intensity concentration for a specific frequency is The expected effect of mitigating (and reducing the sense of pitch of reverberant sound) is certainly realized. However, from the viewpoint of maximizing the effect of reducing the sense of pitch of reverberant sound, a configuration in which none of the M interpolation lengths L [1] to L [M] are common is particularly suitable.

(3) Modification 3
The main cause of the sense of pitch perceived by reverberant sound in contrast 2 is the periodic (regular) arrangement of basic blocks Bb [1] to Bb [M + 1] after adjustment by the time adjustment unit 36. Therefore, if the time length after adjustment (interpolation length L [i]) between the basic block Bb [i] and the basic block Bb [i + 1] is changed, the interpolation blocks P [1] to P [ Even when the section length of M] is common, the desired effect of reducing the sense of pitch of reverberant sound is realized. Therefore, a configuration in which the section lengths of the interpolation blocks P [1] to P [M] are common is also employed. However, if the interpolation length L [i] of the basic block Bb [i] and the basic block Bb [i + 1] and the section length of the interpolation block P [i] do not match, the waveform of the new impulse response H2 is the interpolation block. There is a possibility of discontinuity in the vicinity of both ends of P [i], and the sound quality of the reverberant sound may deteriorate. Therefore, from the viewpoint of reducing the sense of pitch of the reverberant sound while suppressing the deterioration of sound quality due to the discontinuity of the waveform of the new impulse response H2, the section length of the interpolation block P [i] is the basic block Bb [i]. A configuration that matches the interpolation length L [i] of the basic block Bb [i + 1] is particularly suitable.

(4) Modification 4
The types of random numbers λ [1] to λ [M] are arbitrary. For example, a configuration in which the random number generator 62 generates a natural random number is also employed. A configuration in which the types of random numbers λ [1] to λ [M] generated by the random number generator 62 are variably controlled is also suitable. For example, according to the configuration in which the random number generator 62 generates a random number selected by an operation on the input device 16 among a plurality of types of random numbers (for example, uniform random number, normal random number, natural random number), the use of the reverberation imparting device 100 is used. It is possible to generate an appropriate new impulse response H2 according to the purpose and the required effect (for example, the degree to which the pitch feeling of reverberant sound is required to be reduced).

  Further, the criterion for determining the lower limit value λ_min and the upper limit value λ_max of the random numbers λ [1] to λ [M] by the range setting unit 66 is not limited to the expansion rate R. For example, according to the configuration in which the range setting unit 66 variably sets the lower limit value λ_min and the upper limit value λ_max according to the operation on the input device 16, the purpose of use of the reverberation imparting device 100 and the necessary effect (for example, the pitch of the reverberation sound) It is possible to generate an appropriate new impulse response H2 according to the degree to which the feeling is required to be reduced. Note that the lower limit value λ_min and the upper limit value λ_max do not have to be equal in absolute value. That is, a configuration in which the lower limit value λ_min and the upper limit value λ_max are set independently is also employed.

(5) Modification 5
The stage of multiplying the window function wA is changed as appropriate. For example, a configuration in which the windowing unit 34 multiplies each basic block Bb processed by the time adjusting unit 36 by the window function wA is also employed. However, the multiplication of the window function wA is not an essential requirement of the present invention. For example, the time adjusting unit 36 expands each interval between the basic blocks Ba [1] to Ba [M + 1] divided by the waveform dividing unit 32, and the expanded basic blocks Ba [1] to Ba [M + 1]. A configuration is also employed in which the waveform synthesis unit 44 generates a new impulse response H2 by inserting the interpolation blocks P [1] to P [M] at intervals of. Therefore, partial overlap between the basic block Ba [i] and the basic block Ba [i + 1] is not essential in the present invention. However, in a configuration that does not use the window function wA, the reverberant sound may become discontinuous at the boundary between the basic block Ba [i] and the interpolation block P [i], and the sound quality may deteriorate. From the viewpoint of generating natural reverberation sound, there is a configuration in which basic blocks Ba [1] to Ba [M + 1] are generated so as to partially overlap on the time axis and then multiplied by the window function wA. Is preferred.

(6) Modification 6
A method of generating the interpolation blocks P [1] to P [M] is arbitrary in the present invention. In the above embodiment, the basic block Ba [i] and the basic block Ba [i + 1] are averaged to generate the initial interpolation block Pa [i]. However, the basic block Ba [i] and the basic block Ba are generated. A configuration in which the waveform merging unit 52 generates an addition or weighted sum with [i + 1] as the interpolation block Pa [i] is also employed. Further, the basis of calculation of the interpolation block Pa [i] is not limited to the basic block Ba [i] and the basic block Ba [i + 1]. For example, a configuration in which one basic block Ba [i] (or basic block Ba [i + 1]) is used as the interpolation block Pa [i], or three or more consecutive basic blocks (for example, basic block Ba [ A configuration is also adopted in which the interpolation block Pa [i] is calculated by adding or averaging i−1], Ba [i], Ba [i + 1]).

  However, a configuration in which the basic blocks Ba [1] to Ba [M + 1] extracted from the initial impulse response H1 are used for generating the interpolation blocks P [1] to P [M] is not essential to the present invention. For example, from interpolated blocks Pa [1] to Pa [M] (for example, a block in which another impulse response having a characteristic similar to the impulse response H is segmented) irrespective of the initial impulse response H1 (basic block Ba). A configuration for generating the interpolation blocks P [1] to P [M] is also employed. However, in the configuration in which the interpolation block P [i] is generated regardless of the initial impulse response H1, the acoustics of the basic blocks Bb [1] to Bb [M + 1] and the interpolation blocks P [1] to P [M] Due to the difference in characteristics (frequency characteristics), a sense of discomfort in the sense of hearing may be perceived in the reverberant sound of the reverberant sound signal SR. Therefore, from the viewpoint of improving the sound quality of the reverberant sound, the basic block Ba [1] to Ba [M + 1] of the initial impulse response H1 is used to generate the interpolation blocks P [1] to P [M]. Is preferred.

(7) Modification 7
The method for making the frequency characteristics of the plurality of interpolation blocks P [i] (P [i] _1, P [i] _2,...) Generated by the interpolation processing unit 42 in the fourth embodiment different is the rotation of the phase. Is not limited to the method of changing the angle θ of each interpolation block P [i]. For example, as illustrated in FIG. 10, by inverting the waveform of the odd-numbered interpolation block Pa (Pa [1], Pa [3],...), The odd-numbered interpolation block Pb (Pb [1], Pb [ 3],... Are generated, and the even-numbered interpolation blocks Pa (Pa [2], Pa [4],...) Generated by the waveform merging unit 52 are converted into even-numbered interpolation blocks Pb (Pb [2],. A configuration used as Pb [4],. However, a configuration in which the frequency characteristics of the plurality of interpolation blocks P [i] (P [i] _1, P [i] _2,...) In the fourth embodiment are common is also employed.

(8) Modification 8
In the above embodiment, the reverberation imparting device 100 including the impulse response processing unit 22 and the reverberation imparting unit 24 is illustrated, but the impulse response processing device (impulse response) in which the reverberation imparting unit 24 is omitted from the reverberation imparting device 100 of FIG. The present invention is also established as the processing portion 22). The new impulse response H2 generated by the impulse response processing apparatus is provided to a separate reverberation imparting apparatus 100 (reverberation imparting unit 24) via, for example, a portable recording medium or communication network, and used for generating reverberation sound. .

DESCRIPTION OF SYMBOLS 100 ... Reverberation imparting device, 12 ... Arithmetic processing device, 14 ... Memory | storage device, 16 ... Input device, 22 ... Impulse response processing part, 24 ... Reverberation imparting part, 32 ... Waveform classification part, 34 ... ... Windowing part 36... Time adjusting part 38... Interpolation length setting part 42... Interpolation processing part 44 44 Waveform synthesizing part 52 52 Waveform merging part 54 54 Waveform adjusting part 56. ... Waveform processing part, SIN ... Acoustic signal, SR ... Reverberation signal, Ba [1] to Ba [M + 1], Bb [1] to Bb [M + 1] ... Basic block, H1 ... Initial Impulse response, H2 ... New impulse response, P [1] to P [M], Pa [1] to Pa [M], Pb [1] to Pb [M] ... Interpolation block, R ... Expansion rate.

Claims (7)

Waveform dividing means for dividing the first impulse response into a plurality of basic blocks on the time axis;
Interpolation length setting means for setting different interpolation lengths for each basic block;
Time adjustment means for expanding the time difference between each of the plurality of basic blocks and the basic block immediately after the basic block to the interpolation length set by the interpolation length setting means;
Interpolation processing means for generating an interpolation block;
An impulse response processing apparatus comprising: a waveform synthesizing unit that generates a second impulse response by disposing the interpolation block between the basic blocks processed by the time adjusting unit. Waveform dividing means for dividing the first impulse response into a plurality of basic blocks on the time axis;
Interpolation length setting means for setting a plurality of interpolation lengths including different first interpolation length and second interpolation length;
Means for enlarging the time difference between two adjacent basic blocks, wherein the time difference between the first basic block and the immediately following basic block among the plurality of basic blocks is expanded to the first interpolation length; Time adjustment means for expanding a time difference between a second basic block different from the first basic block and the immediately succeeding basic block to the second interpolation length among the basic blocks;
Interpolation processing means for generating an interpolation block;
An impulse response processing apparatus comprising: a waveform synthesizing unit that generates a second impulse response by disposing the interpolation block between the basic blocks processed by the time adjusting unit. The impulse response processing apparatus according to claim 1, wherein the interpolation processing unit generates an interpolation block having a time length equal to the interpolation length of the two basic blocks for every two basic blocks preceding and following each other. The interpolation length setting means includes
Range setting means for setting a range of random number distribution in accordance with an expansion rate of the second impulse response with respect to the first impulse response;
Random number generating means for generating random numbers within the range set by the range setting means;
The impulse response processing apparatus according to any one of claims 1 to 3, further comprising a calculation unit that sequentially sets each of the plurality of interpolation lengths from the random number. The impulse response processing apparatus according to any one of claims 1 to 4,
A reverberation imparting device comprising: a reverberation imparting unit that performs a convolution operation between the second impulse response generated by the impulse response processing device and an acoustic signal. Waveform segmentation processing for segmenting the first impulse response into a plurality of basic blocks on the time axis;
Interpolation length setting processing for setting different interpolation lengths for each basic block;
A time adjustment process for expanding the time difference between each of the plurality of basic blocks and the basic block immediately after the basic block to the interpolation length set in the interpolation length setting process;
An interpolation process for generating an interpolation block;
A program that causes a computer to execute waveform synthesis processing for generating a second impulse response by disposing the interpolation block between the basic blocks after execution of the time adjustment processing.   Waveform segmentation processing for segmenting the first impulse response into a plurality of basic blocks on the time axis;
  An interpolation length setting process for setting a plurality of interpolation lengths including different first interpolation lengths and second interpolation lengths;
  A process of expanding a time difference between two adjacent basic blocks, wherein a time difference between a first basic block and a next basic block among the plurality of basic blocks is expanded to the first interpolation length, A time adjustment process for expanding a time difference between a second basic block different from the first basic block and the immediately following basic block to the second interpolation length among the basic blocks;
  An interpolation process for generating an interpolation block;
  A waveform synthesis process for generating a second impulse response by arranging the interpolation block between the basic blocks after the execution of the time adjustment process;
  A program that causes a computer to execute.

Images