JP2000099062A - Digital signal processing device and sound effect adding device - Google Patents

Abstract

(57) [Summary] PROBLEM TO BE SOLVED: To add a signal processing unit of an option and expand the function so as to be able to cope with different numbers of channels and different sampling frequencies. SOLUTION: An optional FIR processing unit 405 having the same configuration as a normally used FIR processing unit 402 is added. The input selected by the input selection unit 401 is input to the FIR processing units 402 and 405, and reverberation data of four channels is generated by convolution. The reverberation sound data of each channel is mixed with the original sound data in the output mixing section 403, and an output of four channels to which reverberation is added can be obtained. Further, by doubling the sampling frequency generated by the PLL unit 404, reverberation can be added to an input signal having a double sampling frequency. Further, FIR processing sections 402 and 405
By inputting the L and R channels separately, reverberation can be added while maintaining stereo localization.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital signal processing device for performing a convolution process and a sound effect adding device for adding a sound effect such as a reverberation sound.

[0002]

2. Description of the Related Art One of devices for adding a sound effect to an audio signal is a reverberation adding device (reverberator). This reverberation adding apparatus is often used, for example, in a recording studio to add a reverberant sound to an audio signal to give the sound a spaciousness or depth. By adding reverberation to a sound recorded in a studio or the like, it is possible to give an effect as if it were actually played in a hall or a more special effect.

[0003] As one of digital reverberation adding devices, reverberation sound is actually generated in a hall or an iron plate echo, and an impulse response is collected based on the generated reverberation sound. Convolution methods have already been proposed. With such a digital reverberation adding device, a more natural reverberation sound based on an impulse response can be obtained.

[0004] As a specific configuration example of such a digital reverberation adding apparatus, there is one that uses an FIR (Finite Impulse Response) filter to convolve input audio data and an impulse response in the time axis direction. The impulse response is necessary corresponding to the sample of the input digital audio signal. Therefore, 2 ¹⁹ points (52
4,288 points: the fraction is omitted and described as 512 k points), for example, if the sampling frequency of the digital audio signal is 4
At 8 kHz, a reverberation time of about 10 seconds can be obtained.

[0005] However, since multiplication processing of the 512k-point audio sample and the impulse response is required, there is a problem that the multiplication processing becomes enormous. In order to solve this problem, a method of converting an input digital audio signal and an impulse response into frequency element data by respectively performing Fourier transform, multiplying the audio sample and the impulse response converted into the frequency element, and adding the multiplication results Has been proposed. This method has an advantage that the scale of hardware can be smaller than the above-described method of convolution on the time axis.

However, it is necessary to temporarily store the input data corresponding to the required reverberation time in a memory and then output the operation result after the end of the operation processing, so that the reverberation sound for the input is output. There is a problem that the delay until the time is increased. In order to solve this problem, a method has been proposed in which impulse response data is divided on a time axis and input data is convolved by each of the divided impulse responses (Japanese Patent Laid-Open No. 8-501667). ).

[0007]

The above-mentioned conventional digital reverberation adding device usually has a function of inputting a monaural signal and outputting a stereo signal. However, there are different sampling frequencies for digital audio signals, such as 48 kHz and 96 kHz. The number of output channels is not limited to two, but may be four. The conventional digital reverberation adding device cannot easily cope with the difference in the sampling frequency and the difference in the number of channels, and has a problem of lack of expandability.

Accordingly, it is an object of the present invention to provide a digital signal processing device and a sound effect adding device having expandability in a time direction such as a sampling frequency and a spatial direction such as the number of channels.

[0009]

According to a first aspect of the present invention, there is provided a first signal processing unit for performing digital signal processing on an input digital signal, the first signal processing unit comprising:
A second signal processing unit having the same function as that of the first signal processing unit, and a clock generating a sampling frequency clock supplied to the first and second signal processing units. A generator, an input selector for selectively supplying a plurality of channels of input digital signals to the first and second signal processors, and an output mixer connected to the first and second signal processors. , A first and second signal processing unit, a clock generation unit, an input connection unit, and a control unit for controlling an output mixing unit.

According to a second aspect of the present invention, in a sound effect adding apparatus for generating a sound effect by digital signal processing, a first signal processing section for performing convolution processing of a digital audio signal and an impulse response, Of the second signal processing unit, which has the same function as the signal processing unit of
A signal processing unit, a clock generation unit that generates a clock of a sampling frequency supplied to the first and second signal processing units, and a first and second signal that selectively input digital signals of a plurality of channels. An input selection unit for supplying to the processing unit, an output mixing unit connected to the first and second signal processing units, a first and second signal processing unit,
This is a sound effect adding device including a clock generation unit, an input connection unit, and a control unit that controls an output mix unit.

By optionally adding a signal processing unit, the function of signal processing can be expanded by switching the number of channels of the input digital signal and the sampling frequency.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment in which the present invention is applied to a digital reverberation adding apparatus will be described below. FIG. 1 shows a basic configuration of an embodiment. An input selection unit 401, an FIR processing unit 402, an output mixing unit 403, a PLL unit 404 for generating a sampling clock required for processing of each unit, and a CPU (not shown) incorporating software for controlling these components (Cen
The processing device is configured by the tral processing unit.

An L (left) and R (right) channel digital audio signal is input to an input selection unit 401, and a digital audio signal of one of the channels, ie, a monaural audio signal, is supplied to an FIR processing unit 402. You. The input digital audio signal is obtained by linearly or non-linearly quantizing one sample into 24 bits at a sampling frequency FS.

The FIR processing section 402 includes an input selection section 4
Each sample in the time series of the audio signal of the L channel or the R channel selected in 01 is multiplied by the impulse response data, and the multiplication result is added. The FIR processing unit 402 has, as impulse response data, data obtained by collecting and processing a mechanical oscillator such as an iron plate and an impulse response of a space such as a concert hall. In particular,
Impulse response data is stored in a memory in the FIR processing unit 402. Thereby, it has a reverberation sound adding function. The FIR processing unit 402 includes a DSP (Digital Signal
Processor).

The present invention can be applied to a case where a signal block having a digital signal processing function such as a digital mixer other than the reverberation sound addition is used. Further, the present invention is applicable not only to processing of digital audio signals but also to processing of digital information signals such as digital video signals.

The FIR processing section 402 includes an input selection section 4
01, two FIR filters to which the monaural audio data selected is supplied, and an L-channel impulse response and an R-channel impulse response to each FIR filter, whereby stereo reverberation data is supplied. It has a configuration to generate. FIR processing unit 40
2 outputs the generated two-channel reverberation sound data to the output mixing unit 403.

The output mixing section 403 includes level controllers (faders) 411, 412, and 423 and an adder 4
14, 415. The output mixing unit 403 can expand the number of input channels to be mixed and the number of output channels after mixing to those other than those shown in FIG. 1, and the number of channels and the manner of mixing are controlled by the CPU. The output mixing unit 403
For example, it is configured by a DSP.

The level controller 411 controls the level of the reverberation sound data of the L channel, the level controller 412 controls the level of the reverberation sound data of the R channel, and the level controller 413 controls the input selection unit 4.
01 controls the level of the input audio data (original sound) of one of the channels selected. The outputs of the level controllers 411 and 413 are added by an adder 414 to generate L-channel output audio data in which reverberation is mixed. Level controller 4
Outputs 12 and 413 are added by an adder 415 to generate R-channel output audio data in which reverberation is mixed.

A CPU (not shown) includes an input selection unit 40
1, the processing operation of the FIR processing unit 402, the mixing operation of the output mixing unit 403, and the frequency (FS / 2F) of the sampling clock generated by the PLL unit 404.
S) is controlled by software. In this embodiment, the FIR processing unit 405 configured as a DSP board can be added as an optional (optionally selectable) component. 2, 3 and 4 by adding an optional FIR processing unit 405.
The digital reverberation adding apparatus having the configuration shown in FIG.

FIG. 2 shows a configuration in which a monaural signal is input, a four-channel signal is output, and the configuration is extended to be applicable to a multi-channel audio system. FIG.
In the configuration, the input selection unit 401 selects one of the two input channels, and audio data of the selected channel is input to the FIR processing unit 402 and the optional FIR processing unit 405. PLL section 4
04 generates a sampling clock having a frequency FS synchronized with the input audio data, and outputs the sampling clock to the input selection unit 401, the FIR processing units 402 and 4
05, to the output mix unit 403.

The FIR processing unit 402 generates reverberation data of the first and second channels by convolution,
The FIR processing unit 405 generates reverberation data of the third and fourth channels by convolution. The output mixing unit 403 mixes the reverberation sound data of the first and second channels with the audio data (original sound) to generate level controllers 411, 412, and 413.
And level adders 414 and 415 for mixing the reverberation sound data of the third and fourth channels with the audio data (original sound).
1, 422, 423 and adders 424, 425.

FIG. 3 shows a configuration in which a stereo signal is generated, reverberation sound is added to each channel of the stereo signal, and the stereo signal is output. In the configuration of FIG. 3, while maintaining the stereo localization,
Reverberation can be added. The input selection unit 401 outputs each channel of the stereo input as it is, the L channel of the input audio data is input to the FIR processing unit 402, and the R channel is input to the FIR processing unit 405.

The output mixing section 403 includes an FIR processing section 4
02 and the output of the FIR processing unit 405, adders 431 and 432 that mix the same channels (that is, LL channel and RL channel, LR channel and RR channel), and add Level controllers 411 and 413 and an adder 414 for adjusting the level of the output of the adder 431 and the original sound of the L channel, and mixing the output of the adder 432 and the original sound of the R channel after adjusting the levels. 422, 423 and an adder 425. The adder 414 outputs L-channel audio data in which reverberation is mixed, and the adder 425 outputs R-channel audio data in which reverberation is mixed.

FIG. 4 shows that the frequency of the sampling clock generated by the PLL unit 404 is doubled from FS by 2 times.
This is an extended example that can be applied when the sampling frequency of the input audio data is 2FS by controlling to switch to FS. By switching the frequency division ratio of the PLL unit 404, the clock frequency can be changed.
In the input selection unit 401, the FIR processing units 402 and 405, and the output mixing unit 403, since the frequency of the input signal and the output signal is doubled, the time available for the processing is halved. That is, since the sampling period is reduced by half, the processing time from the processing of a certain sample to the arrival of the next sample is halved.

As in the case of FIGS. 1 and 2 for a monaural input and a stereo output, the input data of one channel to be processed by the input selector 401 is selected. The selected audio data is supplied to the FIR processing unit 402 and the optional FIR processing unit 405. The FIR processing units 402 and 405 output reverberation sound data of stereo in the case of the sampling frequency FS, but output the reverberation sound data of one channel because the sampling frequency is doubled (processing time is halved). . FIR processing unit 40
2 outputs the reverberation data of the L channel,
The R processing unit 405 outputs R channel reverberation data. In the output mixing unit 403, the levels are adjusted by the level controllers 411, 412, and 413, and the reverberation sound is mixed with the original sound in the adders 414 and 415, and stereo signals are output from the adders 414 and 415.

In one embodiment of the present invention described above,
The control operation of the CPU for controlling each component is shown in the flowchart shown in FIG. In step S1, an optional DSP (ie, FIR
It is determined whether or not the processing unit 405) is mounted.
If not, in step S2, PL
The sampling frequency generated by L section 404 is set to FS. Then, it is determined whether or not the input selection unit 401 has selected the L channel. in that case,
The L channel is selected (step S4), and if not, the R channel is selected (step S5). Then, the output mix unit 403 outputs the L channel (= A × Rvl).
+ C × input) and R channel (= B × Rvr + C ×
Output).

The control from step S1 to step S6 corresponds to the configuration of FIG. Rvl and Rvr are the reverberation data of the L channel generated by the FIR processing unit 402, A and B are coefficients by the level controllers 411 and 412, respectively, and C is a coefficient by the level controller 413.

In step S1, an optional DS
If it is determined that P (that is, the FIR processing unit 405) is mounted, it is determined in step S7 whether the mode setting is monaural input / stereo output (2FS). Otherwise, in step S8, P
The sampling frequency generated by the LL section 404 is set to FS.

Next, in step S9, it is determined whether the mode setting is monaural input / 4 channel output. If so, in step S10, it is determined whether or not the input selection unit 401 has selected the L channel. If so, the L channel is selected (step S11); otherwise, the R channel is selected (step S12). And the output mixing unit 4
03 produces the following output:

The first channel = A × Rv1 + E × input. The second channel = B × Rv2 + E × input. The third channel = C × Rv3 + F × input. The fourth channel = D × Rv4 + F × input. , Corresponds to the configuration of FIG. Rv1 and Rv2 are reverberation data of the first and second channels generated by the FIR processing unit 402, A, B, C, and D are coefficients by the level controllers 411, 412, 421, and 422, respectively, and E and F Are coefficients by the level controllers 413 and 423, respectively.

If it is determined in step S9 that the mode setting is not monaural input / 4 channel output, in step S14, the input selection unit 401 supplies the L channel to the FIR processing unit 402, and an optional The R channel is supplied to the FIR processing unit 405. Then, the output mixing unit 403 generates the following two-channel outputs.

L channel = A × (Rv1 + Rv3) + C × input R channel = B × (Rv2 + Rv4) + C × input The control from step S7 to step S15 described above corresponds to the configuration of FIG. Rv1 and Rv2 are reverberation sound data of the L and R channels generated by the FIR processing unit 402, Rv3 and Rv4 are reverberation sound data of the L and R channels generated by the FIR processing unit 405, and A
And B are level controllers 411 and 4 respectively.
Where C is the level controller 413
And 423.

In step S1, an optional DS
P (that is, the FIR processing unit 405) is mounted, and if it is determined in step S7 that the input is monaural input / stereo output (2FS), the frequency of the sampling clock generated by the PLL unit 404 becomes 2F.
S is switched to (S16). Then, it is determined whether or not the input selection unit 401 has selected the L channel. If so, the L channel is selected (step S4); otherwise, the R channel is selected (step S5). And the output mixing unit 4
03 outputs the L channel (= A × Rvl + C × input) and the R channel (= B × Rvr + C × input).

The control that finally reaches step S6 after steps S1 to S7 and S16 corresponds to the configuration of FIG. 4 applied when the frequency of the sampling clock is twice (2FS) the normal frequency. Rvl and Rvr are FIR
L channel reverberation data generated by the processing unit 402, where A and B are level controllers 411, respectively.
And 412, and C is a coefficient by the level controller 413.

A specific example of the embodiment of the present invention will be described. First, a description will be given of a process of obtaining an impulse response by collecting reverberation of an actual hall or the like. FIG. 6 shows an impulse response collecting device 9 according to the present invention.
7 shows an example of the configuration of FIG. In this example, the iron plate echo device 9
Measure the impulse response of No. 2. The impulse response collection device 97 can be constituted by, for example, a personal computer. This device 97 generates an impulse response measurement signal, outputs the signal to a measurement target, collects measurement results, and converts the measurement results into impulse response data. The impulse response data is stored, for example, as a file.

The measurement signal generator 90 generates a TSP (time stretch pulse) signal for measuring an impulse response. The TSP signal is a kind of sweep signal, and an impulse signal can be obtained by dividing the signal by a signal having an inverse characteristic. In order to measure the impulse response, it is more preferable to directly generate an impulse signal. However, since measurement is difficult, such a method is used.
The TSP signal generated by the measurement signal generator 90 is D /
The signal is converted into an analog signal via the A converter 91 and input to the iron plate echo device 92.

In the iron plate echo device 92, the input TS
A reverberation sound is generated by the P signal. This reverberation is L
These are output as analog audio signals of the (left) and R (right) channels. These outputs are converted into digital audio signals for the L and R channels by the A / D converter 93. The A / D converter 93 performs sampling at, for example, a sampling frequency of 48 kHz or 96 kHz and a quantization bit number of 24 bits.
As for the output of the A / D converter 93, each of the L and R channels is input to the impulse response collection device 97. The input signal is stored in, for example, a hard disk drive or a memory (not shown).

The reverberation time is defined as the time from when the sound stops until the sound pressure level attenuates by 60 dB. In this example, 6 dB is allocated to 1 bit in the quantization bit number of 24 bits.

The generation of the TSP signal by the measurement signal generator 90 is performed N times. The output signals of the N times are synchronized in the synchronous adder 94 at the timing of signal generation, and are synchronously added. By synchronously adding the N signals, only the reproducible signal is added and the noise component generated at random is not added, so that the S / N ratio can be improved. The S / N ratio is (10 logN) d
B is improved. For example, the S / N ratio is 12d at N = 16.
B is improved.

Each of the L and R channels of the synchronously added signal is supplied to an impulse response converter 95. The impulse response converter 95 divides the supplied signal by a signal having the inverse characteristic of the TSP signal. As a result, the TSP signal is converted into an impulse signal, and the measurement result is converted into an impulse response based on the reverberation generated by the impulse signal. The impulse response data is
It is a peak value obtained at intervals corresponding to the sampling frequency. The signal sampled by the A / D converter 93 with a 24-bit quantization bit number has a 32-bit quantization bit number after conversion.

L output from the impulse response conversion unit
The channel impulse response data 96L and the R channel impulse response data 96R are stored in a CD-ROM.
And MO are recorded on an appropriate recording medium such as an MO. The impulse response collection device 97 may be provided with an interface such as Ethernet, and supplied to the outside via a network.

FIG. 7 shows an example in which an impulse response is collected in a hall. The hall 101 has a stage 101
A and a seat 101B. Stage unit 101A
The sound source 102 is set at a predetermined position. Sound source 102
Are, for example, dodecahedral speakers provided with speakers in 12 different directions on a spherical surface. Microphones 103L and 103R respectively corresponding to the L and R channels are installed at predetermined positions in the customer seat 101B.

The TSP signal output from the impulse response collection device 97 is converted into an analog signal by the D / A converter 91, amplified by the amplifier 100, and reproduced by the sound source 102 as sound. This reproduced sound is transmitted to the microphone 103L.
And 103R. The outputs of the microphones 103L and 103R are sampled by the A / D converter 93 at a predetermined sampling frequency and a predetermined number of quantization bits, respectively, and are converted into L and R channel digital audio signals, which are supplied to the impulse response collection device 97. The processing in the impulse response collection device 97 is exactly the same as the processing in the iron plate echo device 92 described above.

In this case, the position of the sound source 102 is variously changed, and the impulse response is collected. Also, the sound source 10
The loudspeaker used as 2 is collected by changing its brand and the like in various ways. Similarly, the microphone 103
L and 103R are also recorded in various positions and brands. Thus, in one hole 101, a plurality of data are collected. These may be selectable as variations of the reverberation sound, for example, when adding the reverberation sound.

In FIG. 7, one sound source 102 is used, but two sound sources are provided corresponding to two channels, and a TSP signal is first supplied to the left sound source as described above, and the reproduced sound is reproduced. With microphones 103L and 103R
To obtain the impulse response for the left channel. Next, a TSP signal is supplied to the sound source on the right side, the reproduced sound is recorded by the microphones 103L and 103R, and an impulse response for the right channel is obtained. Then, processing is performed by the impulse response collection device to obtain a stereo impulse response. This impulse response is the same as that of FIG.
Are used in the FIR processing units 402 and 405, respectively.

In FIG. 8, reference numeral 1 generally indicates a specific configuration of the reverberation adding apparatus. This reverberation adding device 1
Digital audio signals for two channels (1ch / 2ch) are converted to AES / EBU (Audio Engineering Socie).
ty / European Broadcasting Union) is input from the digital audio input terminal 10 based on the standard. A digital audio signal supplied from an input terminal 10 is input to an input switcher 12 via a digital input unit 11.
Supplied to

The input digital audio signal is
For example, the sampling frequency is 48 kHz and the number of quantization bits is 24 bits. The processing function can be expanded by attaching an option board 50 described later to the device 1. The option board 50 is shown in FIGS. 2, 3 and 4.
This corresponds to the FIR processing unit 405 in the middle. Further, it is possible to correspond to a digital audio signal having a sampling frequency of 44.1 kHz. In this case, when the option board 50 is installed, the sampling frequency is 8
It is possible to handle 8.2 kHz signals.

When an analog audio signal is input to the reverberation adding device 1, the analog audio input terminals 13L and 13R are used. L (left) and R
Each of the (right) channel audio signals is input from the corresponding side of the input terminals 13L and 13R, and A
The number of quantization bits is sampled at a sampling frequency of 48 kHz by the / D converter 14 at 24 bits, for example.
It is converted to a digital audio signal. The output of the A / D converter 14 is supplied to the input switcher 12.

The input switcher 12 switches the system of the input audio signal under the control of a controller 40 described later or by a manual switch. The input switcher 12 corresponds to the input selector 401 in FIGS. 1, 2, 3, and 4. The output of the input switcher 12 passes through a path 31 to a DSP (Digital
Signal Processor) 30.

The DSP 30 has a DRAM (Dynamic Random).
Access Memory), and performs various controls on input / output digital audio signals based on a program supplied from a controller 40 described later. The DSP 30
The supplied digital audio signal is supplied to DSPs 32A to 32K for performing a convolution operation of an impulse response based on a predetermined program. Also, DSP
At 30, an initial reflected sound is generated based on the input signal. Further, the DSP 30 is supplied with a convolution operation result of the impulse response from a DSP 34 described later.

Each of the DSPs 32A to 32K cuts out the digital audio signal supplied from the DSP 30 into blocks each having a predetermined size, and performs a convolution operation using previously supplied impulse response data. DSP3
Each of 2A to 32K has a DRAM having a capacity corresponding to the number of samples to be processed. In this example, DSP32A ~
32H is one each, DSP32I is two, DS
Each of P32J and 32K has a capacity of 16M bits.
Has RAM.

The result of the convolution operation of the impulse response for each block performed by the DSPs 32A to 32K is as follows:
The signals are added by the adder 33, and the DSP 30
Supplied to In the DSP 34, an overflow of the addition result is detected, and for example, the data in which the overflow has occurred is fixed to a predetermined value. The DSPs 32A to 32K and the adder 33 correspond to the FIR processing unit 402 in FIG. 1, FIG. 2, FIG. 3, and FIG. Option board 50 is FIR
Corresponds to the processing unit 405.

The DSP 30 mixes the input digital audio signal, the above-mentioned initial reflection sound, and the result of the convolution operation of the impulse response supplied via the DSP 34 to generate a reverberation sound for the input digital audio signal. Add and output. The output 35 of the DSP 30 is supplied to the output switcher 18. The DSP 30 and the output switcher 18 correspond to the output mixing unit 403 in FIGS. 1, 2, 3, and 4.

The formed reverberation sound and the unprocessed input digital audio signal are also referred to as “wet component” and “dry component”, respectively. DS
In P30, the mixing ratio of the wet component and the dry component can be freely changed for each of the L and R channels. At the same time, the DSP 30 also adjusts the level of the output signal.

A clock FS or 2FS having a frequency corresponding to the sampling frequency of the digital audio signal to be handled is supplied to the DSP 30. D
The signal processing in SP30 is performed based on this clock.

The output switcher 18 switches the output signal system under the control of a controller 40 described later or by a manual switch. The output can be output as digital and analog audio signals. An output terminal 20 according to the AES / EBU standard from the output switcher 18 via the digital output unit 19
, A digital audio signal for two channels is derived. The digital audio signal output from the output switcher 18 is converted by the D / A converter 21 into analog audio signals of the L and R channels. The analog audio signals of the L and R channels are supplied to analog output terminals 22L and 22L, respectively.
Derived to 2R.

In this example, the input terminal 10, the input terminals 13L and 13R, the output terminal 20, and the output terminal 22L
And 22R each use a cannon type having three signal lines of hot, cold and independent ground lines.

Further, the output switcher 18 can be selected so as to bypass the reverberation adding process in the apparatus 1 for the input audio signal. When the bypass is selected, the input digital audio signal is supplied directly from the input switcher 12 to the output switcher 18 through the bypass path 17.

On the other hand, the entire reverberation adding device 1 is controlled by the controller 40. Controller 40
Is, for example, a CPU (Central Processing Unit) or RAM
(Random Access Memory), ROM (Read Only Memory),
It comprises a predetermined input / output interface. ROM
For example, an initial program for activating the system and a serial number are stored in advance. RAM is CPU
Is a work memory for operating, and a program is externally loaded, for example.

The controller 40 is connected to the bus 41 in 8-bit parallel, for example. The bus 41 is connected to the DS
P30, 32A to 32H, 34 respectively.
Via the bus 41, the controller 40 and each DSP 30,
Communication is performed with 32A to 32H and 34. With this communication, the controller 40 sends each DSP 30, 32A
To 32H and 34H, the controller 40 and each of the DSPs 30 and 32A
Data and commands are exchanged between 32H and Ｈ32H.

As described above, the input switcher 12 and the output switcher 18 are connected to, for example, a bus 41 (not shown), and a controller 40.
Is controlled by

The controller 40 is connected to a display device 42 composed of, for example, a full-dot LCD (Liquid Crystal Display). A predetermined display is performed on the display device 42 based on the display data generated by the controller 40.

Although not shown, the input section 43 has a plurality of input means, for example, a rotary encoder adapted to input data corresponding to a rotation angle, and a plurality of push switches. By operating these input units, corresponding control signals are supplied from the input unit 43 to the controller 40. Based on this control signal, predetermined programs, parameters, and the like are supplied from the controller 40 to the DSPs 30, 32A to 32H, and 34.

The reverberation adding device 1 includes a CD-ROM (C
An ompact Disc-ROM) drive 44 is provided. CD-R
A CD-ROM 45 is inserted into the OM drive 44, and data and programs are read from the CD-ROM 45. The read data and program are CD-R
The data is supplied from the OM drive 44 to the controller 40.

For example, the CD-ROM 45 records impulse response data. The impulse response data is read from the CD-ROM 45 and supplied to the controller 40. Then, from the controller 40, DS
This data is supplied to each of P32A to 32K. The DSPs 32A to 32K perform a convolution operation of the impulse response based on the supplied impulse response data.

By recording a large number of impulse response data collected in various environments on the CD-ROM 45, the same reverberation effect as in the environment corresponding to the impulse response to be used can be obtained. Further, a plurality of impulse response data can be used in combination. It is possible to create a space that does not actually exist. Further, the impulse response data can be processed by the reverberation adding device 1. For example, the reverberation time is adjusted by processing the read impulse response data and performing a fade-out process.

As another example, a CD-ROM 45
Alternatively, data obtained by converting impulse response data into frequency element data by Fourier transform may be recorded. The processing in the reverberation adding device 1 can be reduced.

Further, the CD-ROM 45 also stores display data used for display on the display section 42 described above.

The reverberation adding apparatus 1 has a MIDI (Musical Instrument Digital Interface) as an external interface.
rface). The MIDI signal supplied from the MIDI input terminal 46 is supplied to the controller 40. Based on the supplied MIDI signal, a predetermined function of the device 1 can be controlled. Further, the controller 40 can generate and output a MIDI signal.
The MIDI signal supplied from the MIDI input terminal 46 can be processed and output. The MIDI signal output from the controller 40 is supplied from a MIDI output terminal 47 to an external device. In addition, the MIDI through terminal 48 is connected to the MIDI input terminal 46.
The signal is output as it is.

The reverberation adding apparatus 1 is provided with the option board 50, and is described above with reference to FIGS. 2, 3 and 4.
The functions can be extended as shown in FIG. The option board 50 and the device 1 are connected to terminals 51
To 56 and the terminals 15 and 24 are connected to each other. FIG.
Shows an example of the configuration of the option board 50. The option board 50 is configured to extend and execute the convolution operation of the impulse response by the DSPs 32A to 32K and the adder 33. Therefore, the option board 50 includes the above-mentioned DSP3
DSPs 32L, 32M, and D similar to 2A-32K
SPs 60A to 60L are provided, and an adder 61 and a DSP 62 corresponding to the above-described DSP 34 are provided.

The bus 41 ′ on the board 50 is connected to the bus 41 of the device 1 via the terminal 56. Each DSP 32L, 32M, and DSPs 60A-L on board 50 are
Communication with the controller 40 can be performed via the bus 41 '.

Each of the DSPs 32L and 32M has eight 16-Mbit DRAMs and performs a convolution operation together with the DSPs 32A to 32K. An input digital audio signal is output from the DSP 30,
32L and 32M, respectively. DS
The results of the convolution operation by P32L and 32M are supplied to the adder 33 via terminals 54 and 55, respectively, and are added together with the operation results of the other DSPs 32A to 32K.

On the other hand, the DSPs 60A to 60M perform processing in parallel with, for example, the above-mentioned DSPs 32A to 32M. An input digital audio signal is output from the DSP 30,
The signals are distributed to the DSPs 60A to 60M via the terminal 51.

For example, when the processing of four channels from 1ch to 4ch is performed by mounting the option board 50, the convolution operation of 1ch and 2ch is performed by the DSPs 32A to 32M, and the DSPs 60A to 6M are executed.
The convolution operation of 3ch and 4ch is performed by 0M. When a digital audio signal having a sampling frequency of 96 kHz is handled, for example, DSPs to which blocks having the same number of samples are supplied, that is, DSPs 32A and 60A, DSPs 32B and 60B,. By performing the convolution operation in parallel, it is possible to cope with the processing at double speed.

The convolution operation results in the DSPs 60A to 60M are supplied to the adders 61 and added. The addition result is supplied to the DSP 62, subjected to overflow processing similarly to the above-described DSP 34, and
It is supplied to SP30. Then, in the DSP 30,
If necessary, adjust the ratio of the dry component and the wet component, and adjust the mixing ratio with the signals of other channels,
The output is supplied to the output switcher 18.

The option board 50 includes AES
An input terminal 63 and an output terminal 64 for a digital audio signal based on the / EBU standard are provided. A signal for two channels (3 ch / 4 ch) is input to the input terminal 63, and the input signal is supplied to the input switcher 12 via the terminal 15. Similarly, the two channels (3c) output from the output switcher 18
The output signal for (h / 4ch) is supplied to the board 50 via the terminal 24 and is led out to the output terminal 64. In this example, the terminals 63 and 64 are of a cannon type.

Next, the DSPs 32A to 32M and the DSP 60
The convolution operation of the impulse response performed in A to 60M will be described. Here, in order to avoid complexity, the DSP 32A is used without using the option board 50.
An operation performed only at ~ 32K will be described.

FIG. 10 schematically shows processing in each of the DSPs 32A to 32K. The impulse response data is
Under the control of the controller 40, for example, a CD-ROM
45, are supplied to the DSPs 32A to 32K in advance, and are provided to the DSPs 32A to 32K.
Stored in RAM. And each DSP32A-32K
, The impulse response data is divided at predetermined intervals on the time axis, corresponding to the processing block size determined for each.

Here, each of the DSPs 32A to 32K is
32, and the unit of the impulse response processed by the DSP 32 is N. For example, in this example, DSP3
In 2A, since convolution operation of 128-point impulse response data is performed, N = 128. In the following description, one word corresponds to one sampling data of a digital audio signal. Therefore, one word has a time interval of (1 / sampling frequency) on the time axis, and has a quantization bit number (24 bits) as digital data.

The input data supplied to the DSP 32 is N
It is cut out into block data consisting of words. Therefore, the first N words of time are spent on data input. The input N-word data is stored in the DSP32
Is stored in the DRAM. Then, the convolution operation of the impulse response to the stored N words of input data is performed in the next N words. When all the calculations are completed, the calculation results for N words are output.
Therefore, in the operation of N words, a delay of 2N words occurs for the input / output of data.

FIG. 11 shows the processing in the DSP 32 in more detail. DSP32 is a well-known technology.
The convolution operation of the impulse response is performed using the overlap save method in the cyclic convolution.

That is, as shown in FIG. 11, the n-th block 80B and the preceding (n-1) -th block 80A supplied every N words according to the time axis.
And a DFT (Discrete Fourier Transform) is performed on the data and the data on the time axis is converted to the real part 8 of (N + 1) words.
1A and an imaginary part 81B of (N-1) words.

On the other hand, the impulse response data 82 is subjected to DFT in advance for the real data 82A and the zero data 82B of N words, respectively, and the real part 83 of (N + 1) words is obtained.
It is converted into frequency element data 83 comprising A and an imaginary part 83B of (N-1) words.

Frequency element data 81 based on input data
And corresponding frequency elements of the frequency element data 83 based on the impulse response are multiplied by each other, and a filter process (convolution) of adding equal frequency components to each other is performed on the multiplication result. As a result of this operation, (N +
1) Frequency element data 84 consisting of the real part 84A of the word and the imaginary part 84B of the (N-1) word is obtained. The frequency element data 84 is subjected to IDFT, which is the inverse processing of DFT, to obtain data 86 on the time axis consisting of 2N words.

The result of the IDFT is shown in FIG.
As shown at 86 and 87, 2N words are obtained at N word intervals. In each of the data 85, 86, and 87, the first-half N-word data 85A, 86A, and 87
A is discarded, and output data is obtained, such as the (n-1) th block, the nth block, the (n + 1) th block, and so on. The n-th output data is
Two blocks for the corresponding n-th input data,
I'm late.

A large reverberation time can be obtained by increasing the block size and performing a convolution operation on more impulse response data in one process. However, as described above, there is a delay of two blocks before the input block is output. Therefore, if one block is increased, the delay time until the reverberation processing component is output increases, Not practical. Therefore, in this embodiment, processing for obtaining a desired reverberation time is performed by:
This is performed in parallel for each of a plurality of blocks divided into a predetermined number of points (number of words).

FIG. 12 and FIG. 13 show the convolution operation processing divided into a plurality of blocks. For example, 2
Consider a case in which a convolution operation of ¹⁸ words (256 k words) is performed. In this case, the digital audio signal is 2
It is convolved by 56k words (256k points) of impulse response data. Sampling frequency is 4
A reverberation time of about 5.3 sec is obtained at 8 kHz and a reverberation time of about 5.9 sec at a sampling frequency of 44.1 kHz.

As shown in FIG.
The word is divided into two parts, and the front part on the time axis is further divided into two parts. As described above, the side located forward on the time axis is sequentially divided into two. Then, of the two divisions, each of the sides located on the rear side on the time axis is further divided into two to form two blocks of the same size.

FIG. 13 shows an enlarged portion A of the first 8 k words in FIG. This part A is also divided into two parts in the same manner, but for the first 256 words,
Two blocks of 128 words are formed.
The impulse response is convolved for the block. Therefore, the reverberation component is output after being delayed by the first 256 words. However, for example, when the sampling frequency is 48 kHz, this is a delay of only 5 msec, and there is no problem in terms of adding reverberation.

Thus, in this example having a total of 256 k words, 128 words, 256 words, 512 words,
1k words, 2k words, 4k words, 8k words, 1
Two blocks each having a size of 6 k words, 32 k words and 64 k words are formed.

In the DSPs 32A to 32K, processing is performed for each set having the same block size. That is, as shown in FIG. 12 and FIG.
Input data supplied to A to 32K is DSP3
In each of 2A to 32K, 12 for DSP32A
8 words, 256 words for DSP32B, DSP32C
512 words for DSP, 1k words for DSP32D, DSP3
2k words for 2E, 4k words for DSP32F, DSP
8k words for 32G, 16k words for DSP32H, DS
32k words for P32I, 64 for DSP32J and 32K
Each is cut out into k words.

In each of the processes from 128 words to 32k words, the convolution process for two blocks having the same block size is performed by one DSP in a time-division manner.

That is, in each of the DSPs 32A to 32K, a convolution operation is performed on the cut-out block data using the corresponding impulse response data. The latter half of the set of the same block size is output after being delayed by one block after processing.
As a result, in each of the DSPs 32A to 32K, two blocks of the same size are continuously output. D
The reverberation data 88 is generated by adding the outputs of the SPs 32A to 32K by the adder 33.

The data continuously supplied to each of the DSPs 32A to 32K is
It is well known that reverberation can be added to continuously supplied data by performing processing in each cycle of A to 32K and adding the results.

FIG. 14 is a block diagram showing the functions of the parallel processing of the DSPs 32A to 32K. Input data is supplied in parallel to each of the DSPs 32A to 32K. Each of the DSPs 32A to 32K has N = 12.
8, N = 256, N = 512, N = 1k, N = 2k, N
= 4k, N = 8k, N = 16k, N = 32k and N =
Convolve 64k points. The operation result is delayed by 2N words from each of the DSPs 32A to 32K and supplied to the adder 22.

For example, the input data supplied to the DSP 32A is cut into blocks each having N = 128 words, a convolution process is performed on the cut blocks, and the result of the operation is delayed by 2N words with respect to the input timing. Is output. Then, the next block of N words is fetched, and the same processing is repeated. DSP32B
A similar process is performed in each of ~ 32K.

[0097]

As described above, according to the present invention, the hardware is the same as that of a normally provided signal processing unit. The function can be extended in case. Specifically, the number of input or output channels can be doubled, or the sampling frequency of the signal to be processed can be doubled.
Can be doubled.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration in a case where an optional signal processing unit according to an embodiment of the present invention is not provided.

FIG. 2 is a block diagram showing a configuration in a case where an optional signal processing unit according to an embodiment of the present invention is provided, showing a first example of extended functions.

FIG. 3 is a block diagram showing a configuration having an optional signal processing unit according to an embodiment of the present invention, showing a second example of extended functions.

FIG. 4 is a block diagram showing a configuration having an optional signal processing unit according to an embodiment of the present invention, showing a third example of extended functions.

FIG. 5 is a flowchart illustrating an operation of controlling a signal processing function according to an embodiment of the present invention.

FIG. 6 is a block diagram illustrating an example of a collecting device that collects an impulse response using an iron plate.

FIG. 7 is a block diagram illustrating an example in which an impulse response is collected in a hall.

FIG. 8 is a block diagram more specifically showing an example of the configuration of a reverberation adding apparatus to which the present invention is applied.

FIG. 9 is a block diagram illustrating an example of a configuration of an option board of the reverberation adding device.

FIG. 10 is a schematic diagram schematically showing processing in each DSP that performs a convolution operation.

FIG. 11 is a schematic diagram illustrating processing in each DSP in more detail;

FIG. 12 is a schematic diagram illustrating a parallel convolution operation process divided into a plurality of blocks.

FIG. 13 is a schematic diagram illustrating a part of a parallel convolution operation process divided into a plurality of blocks.

FIG. 14 is a schematic diagram illustrating an example in which processing of different N words is performed in parallel.

[Explanation of symbols]

1 ... reverberation adding device, 30 ... DSP, 32A-3
2M ... DSP, 33 ... Adder, 34 ... DS
P, 40: controller, 42: LCD display unit, 43: input unit, 44: CD-ROM drive, 45: CD-ROM, 50: option board, 60A- 60M DSP, 61 adder, 62 DSP, 401 input select unit, 402 FIR processing unit, 403 output mix unit, 404 PLL unit, 405: Optional FIR processing unit

Claims (5)

[Claims] 1. A first signal processing section for performing digital signal processing on an input digital signal, and a second signal optionally having the same function as the first signal processing section and optionally added thereto. A processing unit; a clock generation unit that generates a clock having a sampling frequency supplied to the first and second signal processing units; and a first and second signal that selectively input digital signals of a plurality of channels. An input selector for supplying to a processor, an output mixer connected to the first and second signal processors, the first and second signal processors, the clock generator, and the input connection A digital signal processing device comprising a control unit for controlling the output mixing unit. 2. A sound effect adding apparatus which generates a sound effect by digital signal processing, wherein: a first signal processing unit for performing convolution processing of a digital audio signal and an impulse response; and the first signal processing unit A second signal processing unit having the same function as that of the first and second signal processing units, and a clock generation unit for generating a clock of a sampling frequency supplied to the first and second signal processing units. An input selector for selectively supplying input digital signals of a plurality of channels to the first and second signal processors; an output mixer connected to the first and second signal processors; A sound effect adding device comprising: a control unit that controls the first and second signal processing units, the clock generation unit, the input connection unit, and the output mix unit. 3. The sound effect according to claim 2, wherein the control section controls the input selection section so that an input audio signal to the first and second signal processing sections is switched between monaural and stereo. Additional equipment. 4. The sound effect adding device according to claim 2, wherein the control unit controls the clock generation unit so that the sampling frequency is switched between FS and 2FS in accordance with the input audio signal. 5. The sound effect adding apparatus according to claim 2, wherein the control section controls the output mixing section so as to switch a mode of mixing the original sound component and the reverberant sound component.