JP6663297B2 - Sampling frequency shift amount estimating apparatus, shift amount estimating method, and shift amount estimating program - Google Patents

Description

  The present invention relates to a sampling frequency shift amount estimating apparatus, a shift amount estimating method, and a shift amount estimating program.

  As a method for measuring the propagation characteristics of an acoustic space, an impulse response characteristic measurement is known. In an impulse response characteristic measuring device for measuring an impulse response characteristic, generally, a reproduction system for reproducing a measurement signal and a recording system for recording a measurement signal use signal processing at the same sampling frequency generated from the same oscillator. In a loop configuration.

  On the other hand, the playback system and the recording system may not have a loop configuration. For example, a case where an in-vehicle device (reproducing device) having no external input is used as a reproducing system, and a recording device other than a reproducing device such as a smartphone is used as a recording system. In this case, since the sampling frequency of the recording device is asynchronous with that of the reproduction device, the impulse response characteristics cannot be accurately obtained.

  For example, Patent Literature 1 and Patent Literature 2 disclose a method of improving the measurement accuracy of impulse response characteristics when the sampling frequency is asynchronous by performing a Fourier transform on the measurement signal recorded by the recording system and correcting the phase. Has been described. However, in this method, when there is a large difference between the sampling frequencies of the two devices, the amount of phase rotation becomes too large, and the exact amount of phase correction becomes unknown. In addition, since a phase variation occurs due to a noise component included in the measurement signal, the phase is not accurately corrected, and the impulse response characteristics cannot be accurately determined.

JP-A-2002-365320 JP 2011-22055 A

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a sampling frequency suitable for accurately measuring impulse response characteristics even when the sampling frequency is asynchronous between a reproduction system and a recording system. It is an object of the present invention to provide a deviation amount estimating apparatus, a deviation amount estimating method and a deviation amount estimating program.

  A sampling frequency shift amount estimating apparatus according to an embodiment of the present invention includes a recording unit that records a measurement signal, and a recording signal obtained by recording the measurement signal by the recording unit. Estimating means for estimating a deviation amount of the sampling frequency from the transmission source. The estimating means resamples the recorded signal recorded by the recording means while shifting the sampling frequency within the first frequency range at a first interval, and performs an S / N ratio of the plurality of first recorded signals after resampling. A second frequency range that is narrower than the first frequency range, based on the recording signal, while shifting the sampling frequency within the second frequency range at a second interval that is smaller than the first interval. Resampling, setting a third frequency range narrower than the second frequency range based on the SN ratios of the plurality of second recorded signals after resampling, in a third frequency range and a second interval At a narrower third interval, the sampling frequency of the second recorded signal having the highest SN ratio is converted by linear interpolation, and based on the SN ratios of the plurality of third recorded signals after linear interpolation, the measurement signal Estimating a shift amount of the sampling frequency of the source.

  Further, in one embodiment of the present invention, the estimating means obtains a cross-correlation function between each of the recorded signals resampled within the first frequency range and the first reference signal, and calculates each of the first recorded signals. Is calculated, the S / N ratio of each of the first recorded signals is calculated based on the calculated impulse response characteristics of each of the first recorded signals, and each of the resampled signals within the second frequency range is calculated. A cross-correlation function between the recorded signal and the second reference signal is obtained by calculation, an impulse response characteristic of each second recorded signal is calculated, and the respective impulse response characteristics of each second recorded signal are calculated based on the calculated impulse response characteristics. Calculate the SN ratio of the second recorded signal, calculate the cross-correlation function between each recorded signal whose sampling frequency has been converted within the third frequency range and the third reference signal, and calculate each third recorded signal. Signal, the SN ratio of each third recorded signal is calculated based on the calculated impulse response characteristics of each third recorded signal, and the SN ratio of the third recorded signal having the highest SN ratio is calculated. A configuration may be adopted in which the amount of deviation of the sampling frequency from the source of the measurement signal is estimated based on the sampling frequency.

  In one embodiment of the present invention, the first reference signal is, for example, a signal obtained by filtering a measurement signal at a cutoff frequency corresponding to a first interval. The second reference signal is, for example, a signal obtained by filtering the measurement signal at a cutoff frequency corresponding to the second interval.

  In one embodiment of the present invention, the SN ratio is, for example, a ratio of a peak power, which represents the maximum amplitude of the impulse response characteristic, to an average value of the effective power of at least a part of the impulse response characteristic. It is.

  Further, the sampling frequency deviation amount estimating apparatus according to one embodiment of the present invention may be configured to include an output unit that selects and outputs, as a measurement result, an impulse response characteristic of the third recorded signal having the highest SN ratio.

  A method for estimating a deviation amount of a sampling frequency according to an embodiment of the present invention includes a recording step of recording a measurement signal, and a measurement signal based on a recording signal obtained by recording the measurement signal in the recording step. Estimating the amount of deviation of the sampling frequency from the transmission source of the In the estimation step, the recording signal recorded in the recording step is resampled while shifting the sampling frequency within the first frequency range at a first interval, and the plurality of first recording signals after resampling are resampled. Based on the S / N ratio, a second frequency range smaller than the first frequency range is set, and the recording signal is shifted within the second frequency range by a sampling frequency at a second interval smaller than the first interval. While resampling, based on the SN ratio of the plurality of second recorded signals after resampling, set a third frequency range narrower than the second frequency range, and in the third frequency range and the second At a third interval smaller than the interval, the sampling frequency of the second recorded signal having the highest SN ratio is converted by linear interpolation, and measured based on the SN ratios of the plurality of third recorded signals after linear interpolation. Estimating a shift amount of the sampling frequency of the signal originates.

  A sampling frequency shift amount estimation program according to an embodiment of the present invention is a program for causing a computer to execute the above-described sampling frequency shift amount estimation method.

  According to an embodiment of the present invention, a sampling frequency deviation amount estimating apparatus, a deviation amount estimating method, and a deviation amount suitable for accurately measuring impulse response characteristics even when the sampling frequency is asynchronous between a reproduction system and a recording system. A quantity estimation program is provided.

It is a block diagram showing composition of an acoustic system concerning one embodiment of the present invention. FIG. 7 is a diagram illustrating a flowchart of sampling frequency shift amount estimation processing executed in the recording device included in the audio system according to the embodiment of the present invention. FIG. 7 is a diagram illustrating frequency characteristics when a reference signal (the same M-sequence signal as a measurement signal) is not filtered. FIG. 6 is a diagram illustrating frequency characteristics when a reference signal is filtered in one embodiment of the present invention. FIG. 3 is a diagram illustrating SNRs of 13 impulse response characteristics obtained as a result calculated in processing step S15 of FIG. 2. FIG. 3 is a diagram illustrating an SNR of an impulse response characteristic calculated in processing step S15 of FIG. 2 when a reference signal is not filtered at a cutoff frequency Cf. It is a figure which shows the impulse response characteristic (difference DF is -150Hz) when filtering a reference signal in one Embodiment of this invention. It is a figure which shows the impulse response characteristic (difference DF is -150Hz) when a reference signal is not filtered. FIG. 3 is a diagram showing SNRs of eleven impulse response characteristics obtained as a result calculated in processing step S19 of FIG. 2. FIG. 3 is a diagram showing SNRs of 26 impulse response characteristics obtained as a result calculated in processing step S23 of FIG. 2. FIG. 7 is a diagram illustrating an impulse response characteristic output as a measurement result in one embodiment of the present invention. FIG. 4 is a diagram illustrating a frequency characteristic of an impulse response output as a measurement result in one embodiment of the present invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following, an acoustic system will be described as an embodiment of the present invention.

[Configuration of acoustic system 1]
FIG. 1 is a block diagram illustrating a configuration of an acoustic system 1 according to an embodiment of the present invention. The acoustic system 1 according to the present embodiment includes a reproduction device 10 and a recording device 20.

  The playback device 10 is, for example, a vehicle-mounted audio device mounted on a vehicle. The playback device 10 decodes a measurement signal input from a sound source such as a CD (Compact Disc) or a DVD (Digital Versatile Disc), and outputs the decoded signal from a speaker in the vehicle compartment. Output. An M-sequence signal (Maximum Length Sequence), which is pseudorandom noise, is used as the measurement signal.

  The recording device 20 is a portable terminal that can be brought into the vehicle interior, such as a smartphone, a feature phone, a PHS (Personal Handy phone System), a tablet terminal, a notebook PC, a PDA (Personal Digital Assistant), a PND (Portable Navigation Device), and a portable game machine. As shown in FIG. 1, it has a recording unit 21, a sampling frequency conversion unit 22, an impulse response calculation unit 23, and a sampling frequency estimation unit 24.

[Shift amount estimation processing]
FIG. 2 is a diagram illustrating a flowchart of the shift amount estimation processing executed in the recording device 20 according to the embodiment of the present invention. This flowchart is started, for example, when a predetermined recording standby operation is performed by the user. The recording of the measurement signal by the recording device 20 is performed at an arbitrary position in the vehicle compartment such as a driver's seat or a passenger seat.

[S11 in FIG. 2 (Recording of measurement signal)]
In this processing step S11, during recording standby, a sound signal (here, a measurement signal reproduced by the reproduction device 10) that satisfies predetermined conditions (level, length, etc.) is attached to the microphone attached to the recording device 20 or a built-in microphone. Is input to the microphone, the input measurement signal is recorded in the recording unit 21. The length of the measurement signal to be recorded (hereinafter referred to as “recording signal”) is set so that there is no excess or deficiency even if the sampling point is expanded due to conversion of the sampling frequency (resampling described later). It is about twice the length of the M-sequence signal.

[S12 in FIG. 2 (resampling of recorded signal)]
In this processing step S12, the recording signal recorded in the processing step S11 (recording of the signal for measurement) is resampled by the sampling frequency conversion unit 22. Here, resampling is to convert a signal sampled at a certain sample point sequence into a signal sampled at another sample point sequence, and to perform interpolation processing to increase the sampling frequency to an integral multiple and to reduce the sampling frequency to a fraction of an integer. This is realized by performing a combination of the thinning processes.

[S13 in FIG. 2 (calculation of impulse response characteristics)]
In this processing step S13, the cross-correlation function between the recorded signal and the reference signal resampled in the processing step S12 (resampling of the recorded signal) is calculated by the impulse response calculation unit 23, and the impulse response characteristic is calculated. Is calculated. The reference signal is the same M-sequence signal as the measurement signal reproduced by the reproduction device 10 and is stored in the recording device 20 in advance. In executing the processing step S13, the reference signal is filtered at a cutoff frequency Cf corresponding to an interval Dt between the sampling frequencies after resampling.

  FIGS. 3 and 4 show the frequency characteristics of the M-sequence signal whose sampling frequency Sf after resampling is 44.1 kHz and whose signal length is 32,767 samples. FIG. 3 shows frequency characteristics when the same M-sequence signal as the measurement signal is not filtered, and FIG. 4 shows frequency characteristics when the M-sequence signal is filtered. In each of FIGS. 3 and 4, the vertical axis indicates the level (unit: dB), and the horizontal axis indicates the frequency (unit: Hz). The interval Dt is 50 Hz, and a secondary Butterworth low-pass filter is used for filtering. In this case, the cutoff frequency Cf is set to the following value.

Cf = Sf / Dt = 44.1 kHz / 50 Hz = 882 Hz

  In the present embodiment, the sampling frequency sf1 of the playback device 10 is 44.1 kHz, and the sampling frequency sf2 of the recording device 20 is 48 kHz. The reference of the sampling frequency Sf after resampling is 44.1 kHz, and the range of resampling is Sf ± 300 Hz. Therefore, in the present embodiment, in processing step S12 (resampling of recorded signals), the sampling frequency is shifted at intervals of 50 Hz within the range from 43.8 kHz to 44.4 kHz, and the sampling frequency after each resampling is processed. Step S13 (calculation of impulse response characteristics) is executed. As a result, a total of 13 impulse response characteristics are obtained.

  Note that the above “± 300 Hz” is a range determined in consideration of the maximum deviation from the sampling frequency of the playback device 10 when the recording device 20 processes the recording signal at the asynchronous sampling frequency. I have. Hereinafter, for the sake of convenience of description, the difference between the sampling frequencies of both devices is referred to as “sampling frequency difference”.

[S14 in FIG. 2 (completion determination)]
In this processing step S14, by executing the processing steps S12 (resampling of recorded signals) and S13 (calculation of impulse response characteristics) a plurality of times, the impulse response characteristics of all (a total of 13) recorded signals to be processed are changed. It is determined whether the request has been made.

[S15 in FIG. 2 (Narrowing down of processing target range)]
This processing step S15 is executed when it is determined in step S14 (completion determination) that the impulse response characteristics of all target recorded signals have been obtained (S14: YES). In this processing step S15, in order to estimate the sampling frequency deviation amount, the sampling frequency estimating unit 24 calculates the SNR (Signal) based on the peak power S and the effective power N of the noise for each of the impulse response characteristics of the 13 recorded signals. Noise Ratio) is calculated. The peak power S is obtained by converting the impulse response characteristic into an absolute value and expressing the maximum amplitude in power. The effective power N is an average value of the effective power in a predetermined section (for example, a section from 1 sample to 2048 samples).

  In the present embodiment, when estimating the sampling frequency shift amount, an impulse response characteristic in which noise is suppressed by a pulse compression effect is used. Therefore, the estimation accuracy of the sampling frequency is improved as compared with a configuration in which phase characteristics are detected and estimation and correction are performed (Patent Documents 1 and 2).

  FIG. 5 shows the SNRs of the 13 impulse response characteristics obtained as a result of the calculation in step S15. In FIG. 5, the vertical axis shows the SNR (unit: dB) of the impulse response characteristics, and the horizontal axis shows the difference DF (unit: Hz) with respect to the reference sampling frequency Sf. In the present embodiment, the difference DF at which the SNR peaks within a range of ± 300 Hz (from 43.8 kHz to 44.4 kHz) is estimated to be the sampling frequency shift amount.

  In this processing step S15, in order to accurately estimate the sampling frequency shift amount, the sampling frequency estimating unit 24 narrows the processing target range to a narrower range including the difference DF having the peak SNR. Specifically, the range is narrowed from the difference DF having the largest SNR to the difference DF having the larger SNR among the difference DFs adjacent to the difference DF. In the example of FIG. 5, the range is narrowed down to a range from −150 Hz (difference DF having the largest SNR) to −100 Hz (difference DF having a larger SNR among adjacent difference DFs).

  FIG. 6 exemplifies the SNR of the impulse response characteristic calculated in step S15 when the reference signal is not filtered at the cutoff frequency Cf (see FIG. 3). 6, the vertical axis indicates the SNR (unit: dB) of the impulse response characteristic, and the horizontal axis indicates the difference DF (unit: Hz) with respect to the reference sampling frequency Sf.

  Also in the example of FIG. 6, when the difference DF is -150 Hz, the SNR of the impulse response characteristics becomes maximum. However, the maximum SNR is 18 dB, which is 15 dB lower than the example of FIG.

  7 and 8 show the impulse response characteristics when the difference DF is -150 Hz. FIG. 7 shows an impulse response characteristic when the reference signal is filtered at the cutoff frequency Cf (see FIG. 4), and FIG. 8 shows an impulse response when the reference signal is not filtered at the cutoff frequency Cf (see FIG. 3). The response characteristics are shown. 7 and 8, the vertical axis indicates amplitude, and the horizontal axis indicates samples (time).

  In the example of FIG. 8 (and FIG. 6), since the reference signal is not filtered at the cutoff frequency Cf, the noise component is large (the SNR is small). Therefore, when the level of the measurement signal reproduced by the reproduction device 10 is low or the noise in the measurement environment is large, it becomes difficult to accurately estimate the sampling frequency shift amount. In other words, in the example of FIG. 7 (and FIG. 5), the noise component is reduced (the SNR is increased) by filtering the reference signal at the cutoff frequency Cf. Even if the signal level is low or the noise in the measurement environment is high, the sampling frequency shift amount can be accurately estimated.

[S16 in FIG. 2 (resampling of recorded signal)]
In the present processing step S16, the recording signal recorded in the processing step S11 (recording of the signal for measurement) by the sampling frequency converter 22 in the range narrowed down in the processing step S15 (refining the range of the processing target). Resampled. Specifically, the recorded signal is resampled at intervals of 5 Hz within a range from 43.95 kHz to 44 kHz (difference DF is from -150 Hz to -100 Hz).

[S17 in FIG. 2 (calculation of impulse response characteristics)]
In this processing step S17, the impulse response calculation unit 23 calculates the cross-correlation function between the recorded signal re-sampled in the processing step S16 (recorded signal resampling) and the reference signal to calculate the impulse response characteristics. Is calculated. Note that the reference signal is subjected to a cutoff frequency Cf (Sf / Dt = 44.1 kHz / 5 Hz = 8.82 kHz) corresponding to the sampling frequency interval Dt (5 Hz) after resampling when executing the processing step S17. Be filtered.

  In processing step S16 (resampling of the recorded signal), the sampling frequency is shifted at intervals of 5 Hz within the range from 43.95 kHz to 44 kHz, and processing step S17 (calculation of impulse response characteristics) is performed for each sampling frequency after resampling. Is executed. As a result, a total of 11 impulse response characteristics are obtained.

[S18 in FIG. 2 (completion determination)]
In this processing step S18, by executing the processing steps S16 (resampling of recorded signals) and S17 (calculation of impulse response characteristics) a plurality of times, the impulse response characteristics of all the target (total 11) recorded signals are improved. It is determined whether the request has been made.

[S19 in FIG. 2 (Narrowing down of processing target range)]
This processing step S19 is executed when it is determined in the processing step S18 (completion determination) that the impulse response characteristics of all (a total of 11) target recording signals have been obtained (S18: YES). In this processing step S19, the sampling frequency estimating unit 24 calculates the SNR for each of the impulse response characteristics of the eleven recorded signals based on the peak power S and the effective power N of the noise in order to estimate the sampling frequency deviation amount. Is done.

  FIG. 9 shows the SNRs of the eleven impulse response characteristics obtained as a result of the calculation in step S19. In FIG. 9, the vertical axis shows the SNR (unit: dB) of the impulse response characteristics, and the horizontal axis shows the difference DF (unit: Hz) with respect to the reference sampling frequency Sf.

  In the present processing step S19, in order to accurately estimate the sampling frequency shift amount, the sampling frequency estimating unit 24 narrows the processing target range to a narrower range including the difference DF having the peak SNR. In the example of FIG. 9, the range is narrowed down to a range from −135 Hz (difference DF having the largest SNR among the difference DFs adjacent to the difference DF having the largest SNR) to −130 Hz (difference DF having the largest SNR).

[S20 in FIG. 2 (linear interpolation of recorded signal)]
In resampling, when the interval Dt is narrow, interpolation and decimation for low frequency components are performed, so that a FIR (Finite Impulse Response) with a large number of taps is required. Therefore, in the processing step S20, the sampling frequency conversion unit 22 performs conversion of the sampling frequency by linear interpolation. Here, the linear interpolation is a simple method of interpolating points by connecting them with a straight line, and can be applied when the sampling frequency shift amount is small.

  Specifically, in the processing step S20, the sampling frequency of the recorded signal resampled to 43.97 kHz (frequency corresponding to the difference DF (-130 Hz) at which the SNR becomes the maximum) is determined in the processing step S19 (processing target S19). The range narrowed down by the (range narrowing) is converted by linear interpolation. Conversion of the sampling frequency by linear interpolation is performed at intervals of 0.2 Hz within the range.

[S21 in FIG. 2 (calculation of impulse response characteristics)]
In this processing step S21, a cross-correlation function between the recorded signal whose sampling frequency has been converted in processing step S20 (linear interpolation of the recorded signal) and the reference signal is calculated by the impulse response calculation unit 23, and the impulse response is calculated. Response characteristics are calculated. Since the interval Dt is 1 Hz or less (specifically, 0.2 Hz), the filtering of the reference signal is not performed.

  In processing step S20 (linear interpolation of the recorded signal), the sampling frequency is shifted at intervals of 0.2 Hz within a range from 43.965 kHz to 43.97 kHz, and processing step S21 (impulse response characteristics) is performed for each converted sampling frequency. Is calculated. Thus, a total of 26 impulse response characteristics are obtained.

[S22 in FIG. 2 (completion determination)]
In this processing step S22, by executing the processing steps S20 (linear interpolation of recorded signals) and S21 (calculation of impulse response characteristics) a plurality of times, the impulse response characteristics of all the target (26 total) recorded signals are obtained. It is determined whether the request has been made.

[S23 in FIG. 2 (estimation of sampling frequency deviation amount)]
This processing step S23 is executed when it is determined in the processing step S22 (completion determination) that the impulse response characteristics of all the target (total 26) recorded signals have been obtained (S22: YES). In the present processing step S23, the SNR is calculated by the sampling frequency estimating unit 24 based on the peak power S and the effective power N of the noise for each of the impulse response characteristics of the 26 recorded signals in order to estimate the sampling frequency deviation amount. Is done.

  FIG. 10 shows the SNRs of the 26 impulse response characteristics obtained as a result of the calculation in the processing step S23. In FIG. 10, the vertical axis indicates the SNR (unit: dB) of the impulse response characteristic, and the horizontal axis indicates the difference DF (unit: Hz) with respect to a reference (43.97 kHz which is a reference at the time of linear interpolation). Is shown.

  According to the example of FIG. 10, the SNR of the impulse response characteristic becomes maximum when the difference DF is -1.6 Hz. Taking into account the reference of 43.97 kHz at the time of linear interpolation, “−131.6 Hz” is obtained as the difference DF from the reference sampling frequency Sf (44.1 kHz). Since “−131.6 Hz” is the difference DF at which the SNR of the impulse response characteristic reaches a peak, it is estimated to be the sampling frequency shift amount (substantial difference between the oscillators of both devices). Since the estimated sampling frequency shift amount is about 0.3% (131.6 Hz / 44.1 kHz), it greatly affects the measurement accuracy of the impulse response characteristics.

  In the present embodiment, the impulse response calculation unit 23 selects the impulse response characteristics of the sampling frequency deviation amount (DF = -1.6 Hz in FIG. 10) estimated by the deviation amount estimation processing of FIG. 2 as the measurement result. Then, it is output and stored in a subsequent circuit (for example, a memory or the like in the recording device 20). That is, according to the present embodiment, the measurement result of the impulse response characteristics in which the sampling frequency deviation amount (−131.6 Hz) from the reproduction device 10 is corrected can be obtained.

  11 and 12 show an impulse response characteristic output from the impulse response calculation unit 23 and a frequency characteristic of the impulse response, respectively. In FIG. 11, the vertical axis indicates amplitude, and the horizontal axis indicates samples (time). In FIG. 12, the vertical axis indicates the level (unit: dB), and the horizontal axis indicates the frequency (unit: Hz).

  In the present embodiment, by estimating the sampling frequency shift amount after narrowing down the range of the processing target a plurality of times, noise is reduced as compared with the impulse response characteristics when the conversion of the sampling frequency is insufficient (FIG. 8). It can be seen that the impulse response characteristics are significantly reduced and have a sufficient SNR (see FIG. 11). Also, no noise was observed in the frequency characteristics, indicating that a sufficient dynamic range was secured (see FIG. 12).

  As described above, according to the present embodiment, the sampling frequency shift amount is roughly resampled and estimated at coarse frequency intervals, and then resampled and estimated at fine frequency intervals, and the sampling frequency shift amount is further reduced. By performing the conversion of the sampling frequency by linear interpolation at the stage when the sampling frequency can be obtained, it is possible to improve the accuracy of estimating the sampling frequency shift amount while suppressing the processing amount.

  The above is the description of the exemplary embodiment of the present invention. The embodiments of the present invention are not limited to those described above, and various modifications are possible within the scope of the technical idea of the present invention. For example, the embodiments of the present application also include examples and the like exemplarily shown in the specification or contents obtained by appropriately combining obvious examples and the like.

Reference Signs List 1 audio system 10 reproduction device 20 recording device 21 recording unit 22 sampling frequency conversion unit 23 impulse response calculation unit 24 sampling frequency estimation unit

Claims (11)

Recording means for recording the signal for measurement,
Estimating means for estimating the amount of deviation of the sampling frequency from the source of the measurement signal, based on the recorded signal obtained by recording the measurement signal by the recording means,
With
The estimating means includes:
The recording signal recorded by the recording means, resampling while shifting the sampling frequency within a first frequency range at a first interval, based on the SN ratio of the plurality of first recording signals after resampling, Set a second frequency range narrower than the first frequency range,
The recorded signal is resampled while shifting the sampling frequency within the second frequency range at a second interval narrower than the first interval, and the S / N ratios of the plurality of second recorded signals after resampling. Based on, set a third frequency range narrower than the second frequency range,
In the third frequency range and at a third interval narrower than the second interval, the sampling frequency of the second recording signal having the highest SN ratio is converted by linear interpolation,
Based on the SN ratio of the plurality of third recorded signals after linear interpolation, estimate the amount of deviation of the sampling frequency from the source of the measurement signal,
Sampling frequency shift amount estimation device. The estimating means includes:
The cross-correlation function between each recorded signal resampled within the first frequency range and the first reference signal is obtained by calculation, and the impulse response characteristic of each of the first recorded signals is calculated,
Calculating the SN ratio of each first recorded signal based on the calculated impulse response characteristics of each first recorded signal,
The cross-correlation function between each recorded signal resampled in the second frequency range and the second reference signal is obtained by calculation, and the impulse response characteristic of each of the second recorded signals is calculated.
Calculating the S / N ratio of each of the second recorded signals based on the calculated impulse response characteristics of each of the second recorded signals,
The cross-correlation function between each recording signal and the third reference signal whose sampling frequency has been converted within the third frequency range is obtained by calculation, and the impulse response characteristics of each of the third recording signals are calculated.
Calculating the SN ratio of each third recorded signal based on the calculated impulse response characteristics of each third recorded signal,
Based on the sampling frequency of the third recording signal having the highest SN ratio, estimating the amount of deviation of the sampling frequency from the source of the measurement signal,
The sampling frequency deviation amount estimating apparatus according to claim 1. The first reference signal is
The measurement signal is filtered at a cutoff frequency according to the first interval,
The second reference signal is
The measurement signal is filtered at a cutoff frequency according to the second interval,
The sampling frequency deviation amount estimating apparatus according to claim 2. The SN ratio is
It is a ratio of the peak power, which represents the maximum amplitude of the impulse response characteristics, to the average value of the effective power of at least a section of the impulse response characteristics.
The sampling frequency deviation amount estimating device according to claim 2 or 3. An output unit that selects and outputs an impulse response characteristic of the third recording signal having the highest SN ratio as a measurement result,
The sampling frequency deviation amount estimating apparatus according to any one of claims 2 to 4. A recording step for recording the signal for measurement,
An estimation step of estimating a deviation amount of a sampling frequency from a source of the measurement signal based on a recording signal obtained by recording the measurement signal in the recording step,
including,
In the estimating step,
The recording signal recorded in the recording step, re-sampled while shifting the sampling frequency at a first interval within a first frequency range,
Based on the SN ratio of the plurality of first recorded signals after resampling, set a second frequency range narrower than the first frequency range,
The recorded signal is resampled while shifting the sampling frequency within the second frequency range at a second interval narrower than the first interval,
Based on the SN ratio of the plurality of second recorded signals after resampling, set a third frequency range narrower than the second frequency range,
In the third frequency range and at a third interval narrower than the second interval, the sampling frequency of the second recording signal having the highest SN ratio is converted by linear interpolation,
Based on the SN ratio of the plurality of third recorded signals after linear interpolation, estimate the amount of deviation of the sampling frequency from the source of the measurement signal,
Estimation method of sampling frequency shift amount. In the estimating step,
The cross-correlation function between each recorded signal resampled within the first frequency range and the first reference signal is obtained by calculation, and the impulse response characteristic of each of the first recorded signals is calculated,
Calculating the SN ratio of each first recorded signal based on the calculated impulse response characteristics of each first recorded signal,
The cross-correlation function between each recorded signal resampled in the second frequency range and the second reference signal is obtained by calculation, and the impulse response characteristic of each of the second recorded signals is calculated.
Calculating the S / N ratio of each of the second recorded signals based on the calculated impulse response characteristics of each of the second recorded signals,
The cross-correlation function between each recording signal and the third reference signal whose sampling frequency has been converted within the third frequency range is obtained by calculation, and the impulse response characteristics of each of the third recording signals are calculated.
Calculating the SN ratio of each third recorded signal based on the calculated impulse response characteristics of each third recorded signal,
Based on the sampling frequency of the third recording signal having the highest SN ratio, estimating the amount of deviation of the sampling frequency from the source of the measurement signal,
The method for estimating a shift amount of a sampling frequency according to claim 6. The first reference signal is
The measurement signal is filtered at a cutoff frequency according to the first interval,
The second reference signal is
The measurement signal is filtered at a cutoff frequency according to the second interval,
A method for estimating a shift amount of a sampling frequency according to claim 7. The SN ratio is
It is a ratio of the peak power, which represents the maximum amplitude of the impulse response characteristics, to the average value of the effective power of at least a section of the impulse response characteristics.
The method for estimating a shift amount of a sampling frequency according to claim 7. An output step of selecting and outputting an impulse response characteristic of the third recorded signal having the highest SN ratio as a measurement result,
The method for estimating a deviation of a sampling frequency according to any one of claims 6 to 9.   A sampling frequency shift amount estimation program for causing a computer to execute the sampling frequency shift amount estimation method according to any one of claims 6 to 10.

Images