JP2005341204A - Sound field correction method and sound field compensation apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce a signal processing load by decreasing the number of filter stages without damaging sound performance. <P>SOLUTION: In a sound field correction method in which an input sound signal is divided into a plurality of frequency bands, and sound field correction is performed through filtering processing on a basis of each frequency band, the division into the frequency bands is performed by setting the bandwidth in the low frequency band side narrower than the bandwidth in the high frequency band side. Also, downsampling of the signal in each frequency band after the division is performed using a sampling frequency corresponding to the bandwidth. Thus, the sound correction processing is performed against each downsampled frequency band signal. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a sound field correction method and a sound field correction apparatus, and more particularly to a sound field correction method and a sound field correction apparatus that perform sound field correction by performing filtering processing for each frequency band.

2. Description of the Related Art Conventionally, a sound field correction device that performs sound field correction by performing filtering processing for each frequency band is known (see, for example, Patent Document 1).
FIG. 4 is a conceptual diagram of a conventional audio reproduction system.
In this case, if the sound field from the speaker to both ears of the listener is expressed by z-transform, equation (1) is obtained.

スピーカの前段に逆の特性を有するフィルタを設けることにより、再生システム全体のインバルス応答を理想特性に近づけることが可能である。 By the way, if the ideal acoustic characteristics are expressed by equation (2),

By providing a filter having the reverse characteristic in the front stage of the speaker, the impulse response of the entire reproduction system can be brought close to the ideal characteristic.

例えば、（４）式で表されるようにすることにより、Ａ（ｚ）Ｑ（ｚ）が理想とする音響特性Ｄに近づくようにＱを求めることが可能となる。

That is, if the characteristic of the correction filter is expressed by equation (3),

For example, by being expressed by equation (4), Q can be obtained so that A (z) Q (z) approaches the ideal acoustic characteristic D.

The above concept can be similarly applied even when the number of speech abilities, the number of listening points (the number of ears and microphones), or the number of channels is arbitrary.
JP-A-8-37698

By the way, the filter processing required for sound field correction requires a convolution operation with an integration time comparable to the length of the impulse response of the sound field. Therefore, for example, when the filter processing is realized by using the FIR filter, a plurality of filters having a very large number of stages are required, and the signal processing load is large, which is one factor hindering practical use. It was.
Accordingly, an object of the present invention is to provide a sound field correction apparatus capable of reducing the number of filter stages and reducing a signal processing load without impairing acoustic performance.

  In order to solve the above problems, the sound field correction method for correcting the sound field by dividing the input acoustic signal into a plurality of frequency bands and performing filtering processing for each frequency band is performed by reducing the bandwidth of the low frequency band. A band division process for dividing the frequency band into narrower frequency bands than the high frequency band, and down-sampling of the divided frequency band signals at the sampling frequency corresponding to the bandwidth. A downsampling process, and a sound field correction processing process for performing a sound field correction process on the signals of each frequency band of the downsampling.

In this case, an up-sampling process of up-sampling the signals in each frequency band after the sound field correction processing at a sampling frequency corresponding to the sampling frequency of the down-sampling, and an output acoustic signal by combining the signals after the up-sampling And a synthesizing process for outputting as a feature.
In the band division process, the bandwidth may be set based on a reverberation time of the input acoustic signal in each frequency band.

In addition, the sound field correction apparatus that divides an input acoustic signal into a plurality of frequency bands and performs sound field correction by performing filtering processing for each frequency band is performed by changing the frequency band of the low frequency side to the frequency of the high frequency side. A band dividing unit that narrows the bandwidth of the band and divides it into the frequency bands;
A downsampling unit for downsampling the signal of each frequency band after division at a sampling frequency corresponding to the bandwidth; and a sound field for performing sound field correction processing on the signal of each frequency band of the downsampling And a correction unit.

According to the above configuration, the band dividing unit divides the frequency band on the low frequency side narrower than the frequency band on the high frequency side, divides the frequency band, and outputs the frequency band to the downsampling unit. .
The down-sampling unit down-samples the signal in each frequency band after division at the sampling frequency associated with the bandwidth, and outputs the signal to the sound field correction unit.
The sound field correction unit performs a sound field correction process on the signals in each frequency band of downsampling.

In this case, the input acoustic signal includes a plurality of filters that pass a predetermined frequency band, and at least a part of the plurality of filters functions as a decimation filter for avoiding interference due to aliasing during the downsampling. You may make it do.
In addition, an upsampling unit that upsamples a signal in each frequency band after the sound field correction processing at a sampling frequency corresponding to the sampling frequency of the downsampling, and outputs the output acoustic signal by combining the signals after the upsampling A synthesis unit to
You may make it provide.

Further, the upsampling unit includes a plurality of filters that pass a predetermined frequency band among signals in each frequency band after the sound field correction processing, and at least a part of the plurality of filters is at the time of the upsampling. It may be configured to function as an interpolation filter for avoiding interference due to aliasing.
Furthermore, the filter may have a bandwidth set based on the reverberation time of the input acoustic signal in each frequency band.

  According to the present invention, the number of filter stages can be reduced and the signal processing load can be reduced without impairing the acoustic performance.

Next, preferred embodiments of the present invention will be described with reference to the drawings.
[1] Principle Description First, the principle of the present invention will be described prior to the description of a specific embodiment.
FIG. 1 is an example of a result of time-frequency analysis of an impulse response of an actual sound field.
In FIG. 1, the solid line is an equal level curve connecting equal portions of the signal level (dB value), and the relationship between the signal levels L1, L2, L3, and L4 is as follows.
L1>L2>L3> L4
It has become.
As shown in FIG. 1, in general, the reverberation time tends to be longer in lower frequency components, and the reverberation time tends to be shorter in Takagi. For this reason, processing for a long integration time is necessary to perform low-frequency correction, but processing for a shorter integration time may be used for a higher frequency.

That is, simply comparing, when the same bandwidth is set, the processing load increases as the frequency range decreases, and the processing load decreases as the frequency range increases.
Therefore, when performing the filtering process by dividing each band, efficient processing can be performed by performing the band division and the filtering process in the band as follows.

(1) When dividing the frequency band, the lower the band, the narrower the band, and the higher the band, the wider the band.
(2) Re-sampling (re-sampling) each signal after dividing it for each frequency band to lower the sampling frequency.
(3) A sound field correction process is performed on the resampled signal for each castle.
(4) Up-sample the corrected signal and re-synthesize it.

  As a result, the lower the bandwidth, the narrower the bandwidth, the lower the sampling frequency at the time of resampling, the smaller the number of filter stages required in the sound field correction processing, and the lower the processing speed required for each filter. The effect that it is possible can be obtained.

[2] Specific Embodiment Next, a specific embodiment will be described.
FIG. 2 is a block diagram of the sound reproduction system of the embodiment.
The sound reproduction system 10 performs a sound field correction on the input R channel input signal RIN, outputs an R channel correction signal RC, and amplifies the R channel correction signal RC to a R channel. The R channel amplifier 12R that outputs the amplified signal RA, the R channel speaker 13R that outputs the R channel acoustic signal SR corresponding to the R channel amplified signal RA, and the input L channel input signal LIN are subjected to sound field correction, L channel sound field correction device 11L that outputs L channel correction signal LC, L channel amplifier 12L that amplifies L channel correction signal LC and outputs it as L channel amplification signal LA, and L channel corresponding to L channel amplification signal LA And an L-channel speaker 13L that outputs an acoustic signal SL.

  In this case, since the R channel sound field correction device 11R and the L channel sound field correction device 11L have the same configuration, the R channel sound field correction device 11R will be described as an example.

  FIG. 3 is a schematic configuration block diagram of the R channel sound field correction device 11R. In FIG. 3, a case where the frequency band is divided into n bands (n is an integer of 3 or more) will be described.

  The R channel sound field correction device 11R is broadly divided into bands of the input R channel input signal RIN, and resampling (downsampling) at a sampling frequency lower than the sampling frequency of the original R channel input signal RIN for a predetermined band. A frequency division / downsampling unit 21 that performs sampling), a sound field correction unit 22 that performs sound field correction on a signal of each frequency band that has been divided and banded by the band division / downsampling unit 21, and a sound field correction The signal subjected to the sound field correction in the unit 22 is resampled (upsampling) so as to have the same sampling frequency as that of the original R channel input signal RIN for a predetermined band, and the signal of each band after the sound field correction is performed. Upsampler that generates R channel correction signal RC - a synthesis section 23, and a.

  The band division / downsampling unit 21 includes a filter F11 (= high-pass filter) that passes the component of the highest frequency band (= first band) of the R channel input signal RIN, and the lowest frequency band of the R channel input signal RIN (= When the band other than the highest frequency band and the lowest frequency band of the R channel input signal RIN is divided into (n−2) bands, (N-2) filters F1m (= bandpass filters) that pass components in the mth band (1 <m <n; m is a natural number) and (n-2) filters corresponding to each of the filters F1m A downsampling unit Dm and a downsampling unit Dn corresponding to the filter F1n are provided. In this case, the pass frequency band of each of the filters F11 to F1n is set stepwise so as to become narrower toward the lower frequency side. Thereby, since the pass frequency band is set narrower toward the lower frequency side, the sampling frequency of each of the down-sampling units Dm and Dn can be set lower toward the lower frequency side.

  The filters F1m and F1n also function as a decimation filter for removing components above the Nyquist frequency of the data after downsampling and avoiding interference due to aliasing.

The sound field correction unit 22 includes sound field correction circuits C1 to Cn that perform sound field correction processing using an FIR filter (not shown) or the like for each of the divided n bands.
The upsampling / synthesizing unit 23 includes a filter F21 (= high-pass filter) that passes only a predetermined frequency band component of the signal after the sound field correction processing corresponding to the sound field correction circuit C1, and (n−2) sound fields. (N-2) upsampling units Um corresponding to the correction circuits Cm, upsampling units Un corresponding to the sound field correction circuits Cn, and (n-2) upsampling units Um, respectively. (N−2) filters F2m (= bandpass filters) that pass only the band components corresponding to the sound field correction circuit Cm, and filters F2n that pass only the band components corresponding to the sound field correction circuit Cn; And a combining unit 24 that combines the output signals of the filters F21 to F2n and outputs the combined signal as an R channel correction signal RC.

  In this case, the filters F2m and F2n also function as interpolation filters for avoiding interference due to data aliasing after upsampling.

Next, the operation will be described. In the following description, the operation of the R channel sound field correction device 11R will be mainly described.
First, an R channel input signal RIN is input to the R channel sound field correction device 11R of the sound reproduction system 10, and an L channel input signal LIN is input to the L channel sound field correction device 11L.

Here, the operation of the R channel sound field correction device 11R will be described in detail. The R channel input signal RIN is input to the band division / downsampling unit 21.
The R channel input signal RIN input to the band division / downsampling unit 21 is input to each of the filters F11 to F1n.
As a result, the filter F11 passes the component of the highest frequency band (= first band) in the R channel input signal RIN and outputs it to the sound field correction circuit C1 of the sound field correction unit 22. The filter F1n passes the component of the lowest frequency band (= nth band) out of the R channel input signal RIN and outputs it to the downsampling unit Dn.

Similarly, the (n−2) filters F1m pass the corresponding frequency band of the R channel input signal RIN, that is, the component in the mth band (1 <m <n; m is a natural number). Output to the downsampling unit Dm.
The downsampling unit Dn performs resampling (downsampling) at a sampling frequency lower than the sampling frequency of the original R channel input signal RIN for the lowest frequency band (= nth band), and the sound field of the sound field correction unit 22 Output to the correction circuit Cn.
Each down-sampling unit Dm performs resampling (down-sampling) on the frequency band corresponding to the corresponding filter 1m at a sampling frequency lower than the sampling frequency of the original R channel input signal RIN. Output to the sound field correction circuit Cm.

The signal input from the band division / downsampling unit 21 to the sound field correction unit 22 is processed based on a known sound field correction process by a corresponding sound field correction circuit.
The sound field correction circuit C1 outputs the signal after the sound field correction process to the filter F21 (= high-pass filter), and the (n-2) sound field correction circuits Cm output the signal after the sound field correction process. The sound field correction circuit Cn outputs the signal after the sound field correction processing to the upsampling unit Un.
As a result, the filter F21 passes only the high-frequency component of the signal after the sound field correction processing by the sound field correction circuit C1 and outputs it to the synthesis unit 24.

The upsampling units Um and Un perform upsampling of the signal after the sound field correction processing by the corresponding sound field correction circuits Cm and Cn, and corresponding filters F2m and F2n as the same sampling frequency as the original R channel input signal RIN. Output to.
As a result, the filter F2m passes the component of the corresponding frequency band in the output signal of the upsampling unit Um and outputs it to the synthesis unit 24.

The filter F2n passes the corresponding frequency band component of the output signal of the upsampling unit Un and outputs it to the synthesis unit 24.
At this time, the filters F2m and 2n also function as an interpolation filter for avoiding interference due to data aliasing after upsampling, and remove excess frequency components.

As a result, the synthesizer 24 synthesizes the output signals of the filters F21 to F2n and outputs them to the R channel amplifier 12R as the R channel correction signal RC.
The R channel amplifier 12R amplifies the R channel correction signal RC and outputs it to the R channel speaker 13R as the R channel amplified signal RA. The R channel speaker 13R outputs an R channel acoustic signal SR corresponding to the R channel amplified signal RA.

  Similarly, the L channel sound field correction device 11L performs sound field correction on the input L channel input signal LIN, and outputs an L channel correction signal LC to the L channel amplifier 12L. The L channel amplifier 12L Channel correction signal LC is amplified and output to L channel speaker 13L as L channel amplified signal LA. The L channel speaker 13L outputs an L channel acoustic signal SL corresponding to the L channel amplified signal LA.

As a result, the listener can enjoy the sound with the optimum sound field correction based on the R channel sound signal SR and the L channel sound signal SL that have been corrected.
Note that overhead is generated in the band splitting and recombining, but the total processing load can be optimized by optimizing the number of band divisions and the bandwidth.

  As described above, according to the present embodiment, it is possible to reduce the signal processing load and impair the acoustic performance as compared with the case of performing the filtering process with a large number of stages at a high sampling rate without dividing the frequency band at all. It is possible to configure a sound field correction apparatus that does not occur.

It is an example of the result of having carried out the time frequency analysis of the impulse response of an actual sound field. It is a block diagram of the sound reproduction system of an embodiment. It is a general | schematic block diagram of an R channel sound field correction apparatus. It is a conceptual diagram of the conventional audio reproduction system.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Sound reproduction system 11R R channel sound field correction apparatus 11L L channel sound field correction apparatus 12R R channel amplifier 12L L channel amplifier 13R R channel speaker 13L L channel speaker 21 Band division and downsampling part 22 Sound field correction part 23 Upsampling Synthesizer 24 Synthesizer C1-Cn Sound field corrector F11 Filter (HPF)
F1m filter (BPF, decimation filter)
F1n filter (LPF, decimation filter)
F21 filter (HPF)
F2m filter (BPF, interpolation filter)
F2n filter (LPF, interpolation filter)
LA L channel amplified signal LC L channel corrected signal LIN L channel input signal RA R channel amplified signal RC R channel corrected signal RIN R channel input signal SL L channel acoustic signal SR R channel acoustic signal

Claims (8)

In the sound field correction method for dividing the input acoustic signal into a plurality of frequency bands and performing sound field correction by performing a filtering process for each frequency band,
A band dividing process of dividing the frequency band of the low frequency side narrower than the frequency band of the high frequency side, and dividing the frequency band into the frequency band;
A down-sampling process for down-sampling a signal in each frequency band after the division at a sampling frequency corresponding to the bandwidth;
A sound field correction process for performing a sound field correction process on each signal of each frequency band of the downsampling;
A sound field correction method comprising: The sound field correction method according to claim 1,
Up-sampling process of up-sampling the signal of each frequency band after the sound field correction processing at a sampling frequency corresponding to the sampling frequency of the down-sampling;
A synthesis process of synthesizing the upsampled signal and outputting it as an output acoustic signal;
A sound field correction method comprising: In the sound field correction method according to claim 1 or 2,
The sound field correction method according to claim 1, wherein the bandwidth is set based on a reverberation time of the input acoustic signal in each frequency band in the band dividing process. In the sound field correction apparatus that divides the input acoustic signal into a plurality of frequency bands and performs sound field correction by performing a filtering process for each frequency band,
A bandwidth dividing unit that divides the frequency band on the low frequency side into a frequency band narrower than the frequency band on the high frequency side,
A downsampling unit that performs downsampling of the signal in each frequency band after the division at a sampling frequency associated with the bandwidth;
A sound field correction unit that performs sound field correction processing on the signals of each frequency band of the downsampling,
A sound field correction apparatus comprising: The sound field correction apparatus according to claim 4,
A plurality of filters that pass a predetermined frequency band of the input acoustic signal,
At least a part of the plurality of filters functions as a decimation filter for avoiding interference due to aliasing at the time of downsampling. In the sound field correction device according to claim 4 or 5,
An upsampling unit for upsampling the signal of each frequency band after the sound field correction processing at a sampling frequency corresponding to the sampling frequency of the downsampling;
A synthesizing unit that synthesizes the signal after the upsampling and outputs it as an output acoustic signal;
A sound field correction apparatus comprising: The sound field correction apparatus according to claim 6,
The upsampling unit includes a plurality of filters that pass a predetermined frequency band among signals in each frequency band after the sound field correction processing,
At least a part of the plurality of filters functions as an interpolation filter for avoiding interference due to aliasing during the upsampling. The sound field correction device according to claim 5 or 7,
The sound field correction apparatus according to claim 1, wherein a bandwidth of the filter is set based on a reverberation time of the input acoustic signal in each frequency band.

Images