JPH04278999A - Signal processor for audio device - Google Patents

Abstract

PURPOSE:To reduce cost and to eliminate a bad influence of high frequency noise upon another circuit part by reducing the minimum storage capacity of a data RAM required for reverberation processing without degrading the performance of a reverberation processing device and building the RAM in one chip of a digital signal processor. CONSTITUTION:In a signal processing device for audio device where delay data as initial reflected sounds SD1 and reverberation sounds SD2 are generated by an operation part 40 and a RAM 45 with respect to n-bit digital audio signals of right R and left L channels and generated delay data are added to audio signals of L and R channels to obtain the reverberation effect, data for product sum operation is read and written between the operation part 40 and the RAM 45 after temporary conversion of n-bit digital input data to m-bit data n>m, and delay data to be added as reverberation sounds SD2 is generated in common to L and R channels.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal processing device for an audio device that can perform reverb processing on L and R audio signals.

0002

[Prior Art] In recent years, in various audio devices,
A digital signal processor (hereinafter referred to as DSP) is installed in the circuit system that processes audio signals converted into digital data, making it possible to add various effects such as echo, reverb, and equalization to audio signals. There is something. By the way, reverb processing means that when the original sound S is input at time t1, as shown in FIG. 7, the early reflected sound SD1 (t2
By adding and outputting the reverberation sound SD2 (at the time t3 to t4), the outputted sound is given a sense of presence as if it were being heard in a theater or hall. Of course, the initial reflection sound SD1 for the original sound S
and reverberant sound SD2, or change the density (output interval) of early reflected sound SD1 and reverberant sound SD2.
It is well known that a variety of reverb effects can be obtained by adjusting the expansion and contraction of delay time.

[0003] Conventionally, delayed signal components have been generated by digital filter means formed within the DSP, and such reverb processing has been performed, but in this case, the delay time of the delayed audio signal components to be added to the original audio signal is In order to obtain a sufficient amount of DSP chip 10 and external data RAM 20
is used. That is, when the original audio signal S in the form of digital data is input, the DSP 10 stores the data in the RAM based on the control signal (arithmetic processing parameters) from the CPU.
20 to generate early reflected sound SD1 and reverberant sound SD2, which are added to the original audio signal S and output.

FIG. 9 shows an example of an arithmetic processing block formed by the DSP 10 and data RAM 20 for reverb processing of actual L and R stereo audio signals. For example, sampling frequency fS = 44.1 KHz, quantization 1
When the 6-bit L and R digital audio signals are input to the DSP 10, the L and R audio signals are first input to the sampling frequency converters 11L and 11R to convert the sampling frequency to 1/2, and then are sampled. Frequency fS
=22.05KHz, quantized into 16-bit digital data. This digital data is the early reflection sound S
D1 and reverberation sound SD2 are input to the reverb processing calculation section 12 as data for generating the reverberation sound SD2.

The reverb processing calculation unit 12 includes early reflected sound calculation units 13L, 13R, and reverberation sound calculation units 14L,
14R are provided, each with 256Kbi, for example.
By temporarily storing 16-bit audio data in the data RAM 20 with a storage capacity of t, reading it out at a predetermined timing to obtain a delayed signal component, and further performing predetermined multiplication and addition on the delayed signal component, a predetermined level and delay amount can be obtained. ,
and early reflected sound SD1 or reverberant sound S with set density.
D2. Generated early reflection sound S
D1 and reverberant sound SD2 are combined by mixing circuits 15L and 15R, and supplied to sampling frequency converters 16L and 16R, where the sampling frequency fS = 44.1KH is again set.
z digital audio signal and further mixed circuit 17L.
, 17R are combined with the original audio signal S. As a result, a so-called reverb-processed audio signal in which early reflected sound SD1 and reverberation sound SD2 are added as delayed audio signal components to the original audio signal S as shown in FIG.
It will be output from 0.

[0006] In addition, early reflected sound SD1 and reverberant sound SD
The data for the generation of 2 is the sampling frequency of 1
By converting to /2, the amount of data is halved, and when considered with a RAM of the same capacity, it is possible to obtain twice the delay time, and the relative processing time is also doubled. On the other hand, since the sampling frequency fS is 22.05 KHz, it is necessary to cut the sound range above 11 KHz to remove aliasing noise, but for human hearing, it is
Even if the sound range of 1 kHz or more is cut, it does not cause much trouble, and it is only for the early reflection sound SD1 and reverberant sound SD2, not for the original audio signal S, so it is not necessary to have a strictly high sound quality. , there is no particular problem.

[0007]

[Problems to be Solved by the Invention] By the way, the initial reflection sound SD1 and the reverberation sound SD2, which can obtain a reverberation effect that is almost satisfactory for audio equipment, are produced using the DSP10.
In order to generate the delayed signal component, a sufficient delay signal component must be obtained by reading and writing between the reverb processing calculation section 12 and the data RAM 20. Here, when the audio data to be processed by the DSP 10 is 16 bits, even if the data amount is reduced by converting the sampling frequency as described above, the storage capacity of the data RAM 20 is at least 256 Kbit (i.e. 16 bits x 16K) is required. However, the storage capacity is 25
It is difficult to incorporate a 6Kbit data RAM into one chip DSP, so the data RAM 20 is
As mentioned above, it is externally attached to the SP10. Therefore, in order to implement a digital reverb processing function in audio equipment, a DSP chip and a large-capacity separate data RAM are required, resulting in a considerable increase in cost.

Furthermore, data is exchanged between the DSP 10 and the data RAM 20 at a fairly high speed.
The signal contains high frequency components of about 4 to 8 MHz. For this reason, the DSP chip and data RAM
When installed separately in audio equipment, there is a major problem in that this high frequency component leaks and causes noise in other circuits, causing adverse effects such as poor tuner reception or malfunction.

[0009]

[Means for Solving the Problems] In view of the above-mentioned problems, the present invention makes it possible to significantly reduce the minimum required data RAM storage capacity in order to obtain a sufficient reverb effect as an audio device. It has at least an arithmetic unit that can perform a sum-of-products operation on n-bit digital input data, and a RAM that stores data for this arithmetic unit to perform the sum-of-products operation. , generates delay data as early reflection sound and delay data as reverberation sound for input L channel and R channel n-bit digital audio signals, and generates delay data as early reflection sound and reverberation sound. In the signal processing device for an audio device that can obtain a reverb effect by adding it to the L channel and R channel audio signals, the arithmetic unit and the RA
When reading and writing data for product-sum calculations between M and M, n-bit digital input data is converted to m-bits (n>m), and delay data to be added as reverberation sound is converted to the L channel. The R channel is configured to be generated in common.

0010

[Operation] The data generated by the arithmetic unit is only a delayed signal component (early reflection sound and reverberation sound) with respect to the original audio signal, and this delayed signal component has the same level of sound quality (quantization bit number) as the original audio signal. At the very least, the deterioration in sound quality is almost unrecognizable to human hearing. Therefore, even if the number of quantization bits is reduced only for the exchange of audio data between the arithmetic unit and RAM for obtaining delayed signal data,
There is not much difference in the sound quality of the SP output audio signal compared to when the quantization bit number is not converted.
This means that the storage capacity of M can be greatly reduced. moreover,
By generating reverberation sound in common for the L channel and the R channel, the minimum required RAM storage capacity can be further reduced. The reduction in storage capacity also makes it possible to incorporate RAM into the DSP chip.

[0011]

[Embodiment] FIG. 1 shows, as an embodiment of the present invention, a signal processing device for an audio device constructed of a DSP chip so as to be able to obtain a reverberation effect on L and R audio signals through digital data calculation. Yes, 30 is DS
P is shown as a whole, 40 is a reverse processing calculation unit, 41
Reference numerals L and 41R represent early reflection calculation units, 42 a mixing circuit, 43 a reverberation sound calculation unit, and 44L and 44R a mixing circuit. 45
indicates a data RAM with a data storage capacity of 64 Kbits,
This data RAM 45 is a DSP 30 using a 1-chip IC.
Built-in. Note that the same reference numerals as in FIG. 9 indicate the same parts.

In the DSP 30 of this embodiment, when L and R digital audio signals with a sampling frequency fS = 44.1 KHz and quantized 16 bits are input, the sampling frequency converters 11L and 11R convert the L and R audio signals. The sampling frequency of is converted to 22.05 KHz and inputted to the reverb processing calculation unit 40 as data for generating early reflected sound SD1 and reverberant sound SD2.

In the reverb processing calculation section 40, early reflection sound calculation sections 41L and 41R are provided in order to obtain early reflection sound SD1 for the L and R audio signals, but data for generating reverberation sound SD2 is provided. As, L
, R are synthesized in the mixing circuit 42,
The signal is supplied to the reverberation sound calculation section 43. And reverberation sound calculation section 4
3, the reverberation sound SD2 is commonly generated for the L and R audio data, and the generated reverberation sound SD2 is sent to the mixing circuit 44.
At L and 44R, the L and R audio signals are synthesized with the separately generated early reflected sounds SD1 and SD1, respectively, and are supplied to the sampling frequency converters 16L and 16R, where the sampling frequency fS = 44.1 is again
It is returned to a KHz digital audio signal, and then sent to the mixing circuit 1.
By being synthesized with the original audio signal S at 7L and 17R,
The reverb-processed L and R audio signals will be output.

FIG. 2 shows a specific example of the configuration of the reverb processing calculation unit 40 that generates the early reflected sound SD1 and the reverberant sound SD2. In the figure, D1 to D6 are delay processing units that write input data into the data RAM 45 and read it out at predetermined timings to form delayed signal components. The delay is obtained by adding . Also, M is a multiplier whose predetermined gain is individually set according to the reverb processing parameter, P is an adder, and each delay processing unit D1 to D6 is a cyclic digital Configuring the filter. The data RAM 45 in which the delay processing units D1 to D6 of the reverb processing calculation unit 40 record and read data has a storage area allocated to each of the delay processing units D1 to D6.

In the early reflection calculation units 41L and 41R, the input data (16 bits) is processed by the delay processing units D1 and D2.
Predetermined coefficients are sequentially added to the data RA.
The L audio signal and 16-bit early reflection sound SD1 for each R audio signal
is generated. Here, the delay processing units D1 and D2 perform a process of converting 16-bit data into 8-bit data when storing data in the data RAM 45, and store the data in the data RAM 45.
45 is stored in an 8-bit state. Also, data R
The 8-bit data read from AM45 is converted to 16-bit data again before being output. The bit conversion method will be described later.

On the other hand, the data related to the L and R audio signals, which have been synthesized by the mixing circuit 42, are input to the reverberation sound calculation unit 43 that generates the reverberation sound SD2. In other words, L,
The same reverberation sound SD2 will be generated for the R audio signal. The reverberation sound calculation unit 43 includes a processing unit 43a configured to actually obtain reverberation sound with a predetermined delay time and output level using a recursive digital filter;
A processing section 43b for setting the density of reverberation sound is provided, and the processing section 43a adds a predetermined coefficient to the input data (16 bits), and also adds a predetermined coefficient to the input data (16 bits).
, D4, the data is converted into 8 bits, stored in a predetermined area of the data RAM 45, read out at a predetermined timing, converted again into 16 bits, and output as a delayed signal component. A predetermined gain is given to each output delayed signal component, and the signals are sequentially added to form reverberant sound SD2 data. Furthermore, this reverberation sound SD
2. The processing unit 43b outputs a delayed signal component and adjusts the delay time of the data in order to adjust the density. The delay processing sections D5 and D6 in the processing section 43b also record and read data into and from the data RAM 45 after being converted into 8-bit data.

In this embodiment, the data RAM 45 only has a data storage capacity of 64 Kbits, which allows it to be built into the DSP 30. And 64Kbit data storage capacity R
In order to obtain a reverberation effect that is comparable to the case where AM is used and a RAM with a data storage capacity of 256 Kbit is used, the storage area is shared for L and R for the reverberant sound SD2, In other words, when storing data in the data RAM 45 in order to generate the same reverberant sound SD2 and obtain delayed signal components, 16
This is done after converting bit audio data to 8 bits. In other words, strictly speaking, with regard to the delayed signal component added to the original audio signal S as a reverb effect, sound quality deterioration occurs during bit conversion, and the reverberant sound SD
2 cannot be said to be a strictly accurate reverberant sound SD2 for the L and R original voices, respectively, due to the shared use of L and R, but for human hearing, the early reflection sound SD1 and the reverberant sound SD2 are bits. Even after conversion, the deterioration in sound quality is not so clearly recognized, and since the auditory sensation of reverberation depends on the early reflected sound SD1, the shared use of L and R of the reverberation sound SD2 is not a particular problem. In practice, it is possible to obtain an almost sufficient reverb effect as an audio device.

Delay processing units D1 to D6 and data RAM
As specific examples of the 16-bit to 8-bit bit conversion method performed when storing/reading data between 45 and 45 bits, two types of methods will be described below.

The first possible bit conversion method is to truncate (or round) the lower 8 bits when storing and pad the lower 8 bits with 0 when reading, as shown in the conceptual diagram in FIG. When data is processed in 16 bits in the reverb processing calculation unit 40, the lower 8 bits are discarded and stored in the data RAM 45.
The 8-bit data read from the data RAM 45 is supplied with 8-bit storage data and stored therein, and "00000000" is added as the lower 8 bits of the 8-bit least significant bit (LSB). 1
This is 6-bit data.

A second method is to perform bit conversion using μ compression/expansion characteristics. An example of this is shown in Figure 4.
As shown in the conceptual diagram, for the data that is being processed in 16 bits in the reverb processing calculation section 40, first the lower n bits (for example, 2 to 4 bits) are cut off from the 16 bits, and then the (16-n) bits are Fromμ
Data is compressed to 8 bits based on compression characteristics and stored in RAM4.
5, and on the other hand, the 8-bit data read out from the data RAM 45 is expanded to (16-n) bits based on μ expansion characteristics, and the lower n bits are padded with 0 to make 16-bit data. .

For example, when compressing 12 bits to 8 bits after first truncating the lower 4 bits,
), but since it is digital data, it is realized by polygonal line approximation. Therefore, first, the dynamic range of 12-bit data (-2048 to 2048) is divided into Q blocks, and conversion is performed for each block. The conversion formula for 12->8-bit conversion is D
Let t8 = 2-mDt12+C, and each block B1 ~
BQ to Dt8 = 2-m1 Dt12+C1...
・・・・・・Dt8 =2-mQ Dt12+CQ
Try to give. Here, Dt8 represents 8-bit data, Dt12 represents 12-bit data, C represents a constant, and m represents a natural number. That is, the configuration of the 12-8 bit conversion processing unit according to this example is as shown in FIG. The 8-bit data Dt8 is divided into B1 to BQ and converted into 8-bit data Dt8 according to the conversion formula given respectively.
and is outputted via the synthesis processing section 61.

Furthermore, when reading data from the data RAM 45, the 8-bit data to be read is first processed by polygonal line approximation based on the expansion characteristics shown in FIG. Expand to 12 bits. And as the lower bit of the 12 bits “
0000'' is added to make it 16-bit data.

[0023] In the case of audio data, by performing non-linear 16-8 bit conversion using such compression/expansion characteristics, the processing speed is reduced by 3 to 4 times compared to when linear processing is performed.
It is known that there is a bit improvement effect, and it can be said that the conversion method of this example maintains the sound quality equivalent to 11 to 12 bits in the 8-bit data stored in the data RAM 45.

Although two examples of bit conversion methods that can be applied in this embodiment have been described, no matter what bit conversion method is adopted, the signal processing device for audio equipment of this embodiment requires only the RAM required for reverb processing. Needless to say, it is possible to reduce the capacity. Also,
Data held in RAM does not necessarily have to be 8 bits. Furthermore, early reflection sound SD1,
The configuration of the digital filter for generating the reverberant sound SD2 may be either a cyclic type or an acyclic type.

002
5]

As explained above, in the signal processing device for an audio device of the present invention, reading and writing of data for the product-sum calculation between the calculation unit and the RAM is performed by converting n-bit digital data into m-bit (n>m ), and the reverberation sound is generated in common for the L channel and R channel, thereby reducing the minimum amount of RAM required for reverb processing without substantially degrading the performance of the reverb processing device. significantly reduces the storage capacity of
It is possible to reduce costs and reduce RAM
Since it can be incorporated into the SP and integrated into a single chip, high frequency noise will not leak out of the IC, which has the effect of eliminating the negative effects of high frequency noise.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.

FIG. 2 is a circuit diagram showing a specific example of a reverb processing calculation section according to the present embodiment.

FIG. 3 is an explanatory diagram of an example of a bit conversion method that can be adopted in this embodiment.

FIG. 4 is an explanatory diagram of another example of a bit conversion method that can be adopted in this embodiment.

FIG. 5 is a compression/expansion characteristic diagram used for bit conversion in this embodiment.

FIG. 6 is a block diagram of a polygonal line approximation process during bit compression used for bit conversion in this embodiment.

FIG. 7 is an explanatory diagram of reverb processing.

FIG. 8 is an explanatory diagram of a conventional DSP capable of reverb processing.

FIG. 9 is a block diagram of a conventional signal processing device for an audio device that is capable of reverb processing.

[Explanation of symbols]

30 DSP 40 Reverb processing calculation section 41L, 41R Early reflection calculation section 42 Mixed circuit 43 Reverberation sound calculation section 45 RAM D1 ~ D6 Delay processing section M Multiplier P Adder

Claims (1)

[Claims] 1. A computing unit capable of performing a sum-of-products operation on n-bit digital input data, and a RAM for storing data for the computing unit to perform the sum-of-products operation, the computing unit comprising: and RAM generates delay data as early reflection sound and delay data as reverberation sound for the input L channel and R channel n-bit digital audio signals, and generates delay data as early reflection sound and delay data as reverberation sound. In the signal processing device for an audio device that can obtain a reverb effect by adding delayed data as reverberant sound to the L channel and R channel audio signals, the arithmetic unit and the R channel
Reading and writing of data for product-sum calculations with AM is n.
An audio device characterized in that digital input data of bits is converted into m bits (n>m), and delay data to be added as the reverberation sound is generated in common for the L channel and the R channel. signal processing equipment.