JPH04295727A - Impulse-response measuring method - Google Patents

Abstract

PURPOSE:To improve S/N by measuring the impulse response for every divided band, taking out the band wherein S/N is to be improved, and generating an acoustic signal with, a speaker and the like after amplification. CONSTITUTION:A signal for measuring impulse response, e.g. a time extending pulse (TSP), is supplied to an input terminal 1. The TSP signal is sent into a speaker 11 in a chamber 10 which is an acoustic system to be measured, and a sound is generated. The response of the sound by the echo and the like is collected with a microphone 12. Only the high-region component of the response signal from the microphone 12 is taken out through a high-pass filter(HPF) 3 and outputted through an output terminal 2. The TPS signal is sent into a low-pass filter(LPF) 5 by the same way, and only the low-region component is taken out. The measured results of the impulse responses with respect to the low-region component and the high-region component are added, and the impulse responses in the entire band are obtained. The highly accurate impulse responses can be measured in a short time, and the required S/N in the entire bands can be readily obtained.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an impulse response measuring method in which an impulse signal is applied to an acoustic system to be measured such as an audio room or a music hall and the response thereof is measured.

0002

2. Description of the Related Art In order to accurately measure the impulse response of an acoustic system to be measured in a room or the like, it is important to improve the S/N ratio. However, since impulse is a signal with the smallest power compared to its peak value, in reality, S/
It is difficult to earn N. If the pulse width is widened, the energy can be increased, but the disadvantage is that the frequency characteristics are no longer flat.

[0003] Therefore, various methods have been proposed to improve the S/N ratio. For example, it is known to repeat the impulse measurement described above many times and average all the measurement results (so-called synchronous addition). This is done in consideration of the fact that by repeating additions, the noise level hardly increases, but only the level of the signal component increases. However, the repetition period must be longer than the time at which the response of the acoustic system under test can be considered to have converged to zero, for example, twice the reverberation time. If you try to find it, it will take several tens of minutes or more.

[0004] Also, in the Journal of the Acoustical Society of Japan, Vol. 45, No. 1, pp. 44-50, IEICE Techniques EA89-33, etc., there is a time stretched pulse (TSP).
) is shown to measure the impulse response. This TSP is a signal obtained by passing an impulse through an all-pass filter whose group delay is proportional to frequency. For example, the power of a 4096-point TPS signal is about 31 dB higher than that of an impulse signal, and the same S/N ratio as 1200 synchronous calculations using impulses can be obtained in one measurement. Furthermore, synchronous addition is also performed by repeating this process.

[0005]

[Problems to be Solved by the Invention] These impulse response measurement methods all involve sending a signal with a uniform spectrum over the entire frequency band to an electro-acoustic transducer such as a speaker in the acoustic system under test. The impulse response is calculated backward from the signal obtained by generating a sound and collecting the response with an acoustic-to-electrical converter such as a microphone.

On the other hand, in real environments, noise is often concentrated in some bands, especially in the low range, and when looking at the measurement results, even if the S/N ratio is sufficient in the high range, the noise is often concentrated in some bands, especially in the low range. Since the S/N ratio in the area is insufficient, it is currently necessary to further perform synchronous addition or the like to improve the overall S/N ratio.

The present invention has been made in view of the above circumstances, and it is possible to measure impulse responses with high precision in a short time, and it is possible to measure the impulse response with high precision in a short period of time. It is an object of the present invention to provide an impulse response measurement method that can easily obtain a high S/N ratio.

[0008]

[Means for Solving the Problems] According to the impulse response measurement method according to the present invention, in the impulse measurement method in which an impulse signal is applied to the acoustic system under test and the response is measured, the frequency band is divided into a plurality of parts and each This is an impulse response measurement method that measures the impulse response for each band and synthesizes each measurement result.For the band where you want to improve the S/N, after amplifying the signal through a filter, - solving the above problem by sending it to an acoustic conversion means to generate an acoustic signal;

Here, for bands other than the band in which the S/N is desired to be improved, the signal of the corresponding band is amplified via a filter and the electro-acoustic conversion means (speaker etc.)
However, the signals of all bands may be sent to the above electric wire.
The response signal may be sent to an acoustic conversion means and passed through a filter to extract a signal in the corresponding band.

0010

[Operation] By dividing the band and extracting and amplifying the band in which the S/N is desired to be improved, the signal is sent to an electro-acoustic conversion means such as a speaker to generate an acoustic signal, so that noise in this band is reduced. It is possible to achieve high-precision impulse response measurement in a short period of time.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Some preferred embodiments of the impulse response measuring method according to the present invention will be described below with reference to the drawings. First, in FIG. 1, as a first embodiment of the present invention, steps A and B show how to measure an impulse response by dividing the entire frequency band into two bands, a high band and a low band. . In this case, the order of A and B may be either first.

That is, first, in A of FIG. 1, a signal for impulse response measurement, for example, a time stretched pulse (TSP) is input to the input terminal 1.
lse). This TSP signal is sent to the speaker 11 in the room 10, which is the acoustic system to be measured, and is emitted as sound, and the response due to echoes in the room 10 is collected by the microphone 12. The response signal from the microphone 12 is passed through a high-pass filter (HPF) 3 to extract only high-frequency components, and is then taken out from the output terminal 2 via an amplifier 4. A TSP response signal is obtained from output terminal 2, but an impulse response (however, HP
Inverse filtering for restoring the high-frequency response that is the passband of F3 is performed using, for example, FFT (fast Fourier transform). The impulse response of the inverse filter is simply the time axis of the TSP reversed, so if we find the cross-correlation function between the TSP response and the TSP itself, we can calculate the impulse response (for high frequencies) of the room 10, which is the acoustic system under test. response).

Next, in FIG. 1B, a TS similar to the above
The P signal is passed through input terminal 1 to a low-pass filter (LP
F) 5 and extracts only the low frequency components. Here LPF
5 and the HPF 3 are set so that their frequency responses add up to 1. The output signal from the LPF 5 is amplified by the amplifier 6 and sent to the speaker 11 in the room 10.
and a sound is emitted. The response is collected by the microphone 12 and taken out from the output terminal 2. This output signal is also a TSP response signal, and by inverse filtering using FFT, an impulse response for the low frequency range of the acoustic system to be measured (room 10) is obtained. By adding this measurement result to the above-mentioned impulse response measurement result for the high range, the impulse response for the entire band can be obtained.

That is, the measurement of the impulse response in the embodiment of the present invention is carried out in two steps: the procedure A in FIG. 1 and the procedure B in FIG. and becomes 1, so the overall impulse response obtained by adding each output will be the same as the original system. H
The purpose of PF3 is to remove low-frequency noise, and
The power of the signal added to the low frequency range is increased by the PF5 and the amplifier 6. That is, in this embodiment, it is assumed that noise is concentrated in the low frequency range, and in order to increase the S/N on the low frequency side, a TSP signal equivalent to an impulse signal is used.
After extracting only the low-frequency component in advance with the LPF 5 and amplifying it with the amplifier 6, the speaker 1 in the room 10, which is the acoustic system to be measured, is
1. This increases the power in the low range compared to the case where the TSP signal is directly supplied to the speaker 11. If the passband of the LPF 5 is 1/N of the entire band, the peak value of its impulse response will also be 1/N. Therefore, if the peak value of the signal applied to the speaker 11 is kept constant, the power spectral density in the low range will be increased by N2 times. Note that the amplifier 4 after the HPF 3 is used to compensate for the attenuation caused by passing through the HPF 3 and to match the levels between the outputs of the two measurements, and may be omitted. Furthermore, either of the steps A and B in FIG. 1 may be performed first. Furthermore, the HPF 3 and LPF 5 may be configured by hardware or may be realized by software.

FIG. 2 shows specific examples of the frequency characteristics of the HPF 3 and LPF 5. All of these filters are linear phase FIR filters,
It is set so that each frequency response adds up to 1. In the example of Fig. 2, the frequency value of fs /32 on the frequency axis (
The passing and cutoff bands of the HPF 3 and the LPF 5 are exchanged with each other at a boundary (where fs is a sampling frequency), and this sampling frequency fs is, for example, 44.
It is 1kHz. Also, Figure 3 shows the TS of 1024 points.
An example of P is shown.

Next, actual impulse response measurement using the method of the first embodiment and its results will be explained. As room 10, a listening room with a reverberation time of about 0.3 seconds was used. Also, using TSP (time stretching pulse), 4
An impulse response of 096 points was obtained. The sampling frequency is 44.1kHz, and for LPF5,
The cutoff frequency is set to the entire band (f) as shown in Figure 2 above.
It is set to 1/16 of fs /2), that is, fs /32, and created as an FIR filter with 501 taps.
After designing F, HP is calculated by subtracting it from 1.
I created F3. The signal used for low frequency measurement is the above TS
Insert 15 0s between each sample of P, and this is the LPF
A waveform to be interpolated was created in advance by a computer, and an impulse response including the LPF was calculated from the response obtained. On the other hand, for the high frequency range, HPF characteristics were convoluted during the conventional calculation process. Synchronous addition was performed 10 times each for these low and high frequencies.

FIG. 4 shows the results of measurements conducted under the above conditions. In FIG. 4, characteristic curve a represents the spectrum of impulse response, and characteristic curve b represents the spectrum of background noise when no impulse (actually, TSP) is emitted.
are shown respectively. As a comparative example for this, Figure 5
2A and 2B respectively show a spectral curve a of the impulse response when the band division is not performed and a spectral curve b of the background noise when no TSP is emitted. In this comparative example, the conditions other than dividing the band are the same. Comparing these FIGS. 4 and 5, first of all, in FIG. 5 of the comparative example, the S/N in the high range, that is, about 1 kHz or higher, is
is about 60 dB, but it drops to 20 dB in the low range, and in order to increase this, it is necessary to perform synchronized addition many more times, which takes time. On the contrary,
The response characteristics of the embodiment of the present invention shown in FIG.
(below), the background noise level is apparently reduced by 24 dB, and the S/R is 60 dB in the band above about 100 Hz.
N is obtained. This corresponds to obtaining an S/N equivalent to 2560 synchronous additions in the conventional method in the low frequency range.

As described above, the S/N is improved by 24 dB by dividing the band and pouring 16 times the power into the band where the S/N is desired to be improved, specifically the low range which is 1/16 or less of the entire band. The number of necessary synchronous additions was reduced to 1/256. This makes it possible to achieve highly accurate impulse response measurements in a short time. Note that the signal supplied to input terminal 1 is
Although SP is preferable, it goes without saying that even if impulses are used, the number of synchronous additions can be significantly reduced compared to the case without band division. It is also possible to control the S/N for each band by setting the number of synchronous additions for each divided band.

Next, a second embodiment of the present invention will be described with reference to FIG. This FIG. 6 shows an example in which the band is divided into three, and the individual measurement results of each measurement process of A, B, and C are added to obtain one impulse response result. The order of these three measurement steps can be changed arbitrarily, for example, B, C, A or C, A.
, B, and so on. In FIG. 6, parts corresponding to those in FIG. 1 are given the same reference numerals, and explanations thereof will be omitted.

That is, in FIG. 6, the input terminal 1 is supplied with, for example, a time stretching pulse (TSP) similar to that in the first embodiment. In the measurement step A, the input TSP signal is sent directly to the speaker 11, and the response signal from the microphone 12 is transmitted at a cutoff frequency of f1.
high pass filter (HPF) 21, and this HPF
A signal from 21 is taken out from output terminal 2 via amplifier 22. In the measurement step B, the TSP signal is passed through a low pass filter (LPF) 2 with a cutoff frequency of f1.
3, amplified by amplifier 24, and sent to speaker 11. The response signal from the microphone 12 is sent to the HPF 25 whose cutoff frequency is f2 (f2 < f1), and the signal from the HPF 25 is taken out from the output terminal 2 via the amplifier 26. Therefore, in this step of measuring B, response results for the frequency band f2 to f1 are obtained. Next, in the measurement step C, the TSP signal is sent to the LPF 27 with a cutoff frequency of f2, amplified by the amplifier 28 and sent to the speaker 11, and the response signal from the microphone 12 is taken out from the output terminal 2 as it is.

In the second embodiment shown in FIG.
The band is divided into three bands: above f1, between f2 and f1, and below f2, and the band between f2 and f1 and the band below f2 are amplified before being supplied to the speaker 11, thereby adjusting the signal level relative to the noise level. is increased to improve the S/N. For example, f1 = 1kHz, f2 =
A value such as 100Hz is possible. In this second embodiment, for example, the noise is roughly in the band f2 to f1 and f
This is suitable when the data are distributed over 2 or less bands. Regarding the frequency response of each filter in the three bands,
Of course, the setting is made so that the result of adding these values is 1.

Next, FIG. 7 shows, as a third embodiment of the present invention, an example in which the band is divided into two and the signals in each band are amplified before being supplied to a speaker. That is, in the measurement step A, the TSP signal supplied to the input terminal 1 is passed through the HPF 31, amplified by the amplifier 32, and sent to the speaker 11. In addition, in the measurement step B, T
The SP signal is passed through the LPF 36 and then amplified by the amplifier 37,
It is being sent to speaker 11. Each response signal obtained from the microphone 12 in each of these steps may be taken out as it is from the output terminal 2, but in the third embodiment of FIG. Further, in B, the signals are taken out from the output terminal 2 via the LPF 38 and the amplifier 39, respectively.

In this case, the HPFs 31 and 33 are, for example, f
For example, the LPFs 36 and 38 pass signals of fs /32 or higher (fs is the sampling frequency).
It passes the following signals, and is designed so that the sum of all these frequency responses becomes 1. Further, the amplifiers 34 and 39 are for output level adjustment, and may be omitted. The functions and effects of this third embodiment are similar to those of the first embodiment shown in FIG. 1, so common parts are given the same reference numerals and explanations will be omitted.

Next, FIG. 8 shows a fourth embodiment of the present invention in which the band is divided into a large number of parts (for example, n+2 parts). In other words, f1
Above, f1 to f2, f2 to f3,..., fn
~fn+1, and less than or equal to fn+1 (f1 > f2 >
f3 >>>fn>fn+1), and in each measurement process A, B, C,..., D, and E, the impulse signal supplied to input terminal 1 (e.g. The above TSP signal) is filtered from these bands by HPF 41, band pass filter (BPF) 511,
BPF512,..., BPF51n, and LPF
Take out at 46, amplifiers 42, 521, 5 respectively
22, . In addition, the response signals obtained from the microphone 12 in each measurement process are
F43, BPF531, BPF532,..., B
Amplifiers 44, 541, 542, ..., 54n, and 4 are connected via PF53n and LPF48, respectively.
9, and are taken out from the output terminal 2, respectively.

[0025] Here, the order of each measurement process A to E is arbitrary, and since the operation is almost the same as that of the first embodiment shown in Fig. 1, common parts are given the same reference numerals. The explanation will be omitted. In this fourth embodiment, the above-mentioned first to
Not only can the same functions and effects as in the third embodiment be obtained, but also the power of each band can be increased by further subdividing the frequency band.

It should be noted that the present invention is not limited to the above-mentioned embodiment, but for example, the LPF in the configuration shown in FIG.
The frequency band may be divided into four or more by increasing the number of sets of 23 and HPF 25 to a plurality of sets. Alternatively, the number of times of synchronous addition may be set for each of the plurality of divided bands, and the S/N may be set for each of these bands.

[0027]

Effects of the Invention As is clear from the above explanation, according to the impulse response measurement method according to the present invention, the impulse response is measured for each band by dividing the band, and the S/
The band in which the N is desired to be improved is extracted using a filter, amplified, and then sent to an electro-acoustic conversion means such as a speaker to generate an acoustic signal. This makes it possible to improve the S/N in this band and to Highly accurate impulse response measurements can be achieved in a short period of time.

[Brief explanation of drawings]

FIG. 1 is a block diagram for explaining a first embodiment of an impulse response measurement method according to the present invention.

FIG. 2 is a diagram showing frequency characteristics of the HPF and LPF of the first embodiment.

FIG. 3 is a waveform diagram showing a specific example of TSP (time extension pulse).

FIG. 4 is a diagram showing frequency characteristics of impulse response measurement results of the first example.

FIG. 5 is a diagram showing frequency characteristics of conventional impulse response measurement results.

FIG. 6 is a block diagram for explaining a second embodiment of the present invention.

FIG. 7 is a block diagram for explaining a third embodiment of the present invention.

FIG. 8 is a block diagram for explaining a fourth embodiment of the present invention.

[Explanation of symbols]

1... Input terminal 2... Output terminal 3... High pass filter (HPF) 4, 6...
...Amplifier 5...Low pass filter (LPF) 10...
・Room (acoustic system to be measured) 11...Speaker 12...Microphone

Claims (1)

[Claims] Claim 1: In an impulse measurement method in which an impulse signal is applied to an acoustic system to be measured and its response is measured, a frequency band is divided into a plurality of bands, an impulse response is measured for each band, and each measurement result is synthesized. This impulse response measurement method is characterized in that for a band in which the S/N is desired to be improved, a signal passed through a filter is amplified and then sent to an electro-acoustic conversion means to generate an acoustic signal. Measuring method.