Method and system for performing transfer function measurements on a device

Abstract

A method for performing transfer function measurements on a DUT using a dual-channel transfer function analyzer without a real-time copy of the test signal from a device such as a loudspeaker. A virtual reference test signal, serving as a copy of the DUT test signal, is played back in a channel of the transfer function analyzer at the estimated start time of the DUT test signal and at an estimated playback speed. The known playback speed of the virtual reference test signal is compared with the estimated playback speed of the DUT test signal, and any detected differences between the two signals are compensated by adjusting the playback speed of the virtual reference test signal or the playback speed of the DUT output, or both. A system for performing this method includes a virtual reference module having a detection section and an adjustment section.

Description

Background Technology

[0001] This invention relates to systems and methods for measuring the transfer function of a device, and more particularly, to systems and methods for measuring the transfer function of a device without accessible input. The invention is particularly applicable to measuring the transfer function of a loudspeaker or other device having one or more loudspeakers as device components.

[0002] Loudspeakers and devices with one or more loudspeakers, such as mobile phones and wireless headsets, accept a time-varying signal as input and produce a detectable sound output represented by the time-varying signal. The relative change made by a loudspeaker or other device (hereinafter collectively referred to as a device) between the input and output signals is called the transfer function.

[0003] Systems for measuring the transfer function of a device are known. Such measurement systems include a two-channel analyzer that receives copies of both the input and output from the device under test (sometimes referred to herein as the DUT). When measuring the DUT, a third device, called a playback device, may be needed to provide the input signal. (Sometimes the two-channel analyzer can be used as a playback device. Sometimes this might be a connection analyzer.)

[0004] Typically, a dual-channel analyzer receives copies of both the input and output signals of the DUT simultaneously; that is, they receive “real-time” copies of both signals. Alternatively, the real-time input and output signals can be recorded simultaneously and then played back to the dual-channel analyzer later (“offline”).

[0005] However, sending a real-time copy of the input signal to the device is often inconvenient or impossible: the input connector may be proprietary, the input signal may be encrypted, or the input signal may be transmitted using a wireless point-to-point protocol. Furthermore, dual-channel analyzers using a real-time copy of the input signal cannot continuously measure the playback speed of a playback device, nor can they correct for its side effects. Generally, in the context of "smart speakers," open-loop measurement is desirable.

[0006] This invention facilitates the measurement of transfer functions when a real-time copy of the DUT output is available but a real-time copy of the input is not. Summary of the Invention

[0007] The present invention relates to a method and system for measuring the transfer function of a device capable of receiving input and generating output in response to the input, wherein the measurement is performed by a transfer function analyzer having a first channel for receiving a test signal input and a second channel for receiving an output from a device under test (DUT), the output being generated when the DUT receives a test signal from a playback device.

[0008] The method includes first providing a test signal with desired characteristics in one or more frequency bands. This test signal will reside in a DUT playback device as a DUT test signal. The method also includes storing a copy of the DUT test signal separately from the DUT playback device as a virtual reference test signal (sometimes referred to herein as a "virtual reference signal" or "reference signal"). This virtual reference signal is played separately from the DUT test signal residing in the playback device to a first channel of a function analyzer.

[0009] The method also includes having a DUT playback device play the DUT test signal into the DUT at playback speed to generate a DUT output played into a second channel of the transfer function analyzer. The DUT output is monitored to determine if the DUT test signal is being played into the DUT. When it is determined that the DUT test signal is being played into the DUT, the start time of the DUT test signal playback and the playback speed of the DUT test signal are estimated. A virtual reference test signal begins playback into a first channel of the transfer function analyzer at the estimated start time of the DUT test signal and at the estimated playback speed of the DUT test signal. This occurs when the monitored DUT output is played into the second channel of the transfer function analyzer.

[0010] The method also provides a comparison of the known playback speed of the virtual reference test signal with the estimated playback speed of the DUT test signal. If any difference is detected between the two signals, the playback speed of the virtual reference test signal or the playback speed of the DUT output, or both, is adjusted to compensate for any detected difference between the two signals.

[0011] It should be understood that the "estimated start time" of the DUT test signal can be the time after the DUT test signal actually begins playback. The start time of the virtual reference test signal can be estimated to coincide with the DUT output segment generated after the DUT test signal begins playback. In another aspect of the invention, the virtual reference test signal can begin playback at an offset corresponding to the difference between the estimated start time of the DUT and the amount of time taken to detect the estimated start time of the DUT output.

[0012] It should also be understood that the steps described above—comparing the playback speed of the virtual reference test signal to the estimated playback speed of the DUT test signal and making the necessary adjustments to one or both of these signals—can be performed before the virtual reference test signal begins playback. The virtual reference test signal will begin playback at the estimated start time after the signal playback speed adjustment.

[0013] The virtual reference test system according to the present invention includes a virtual reference module configured to play the DUT output to a second channel of a transfer function analyzer. A copy of the test signal stored in the DUT playback device is stored in the virtual reference module as a virtual reference test signal. The virtual reference module consists of a detection section, an adjustment section, and a virtual reference player. The detection section is configured to: The presence of an output from the DUT is detected; this output is generated in response to a test signal being played into the DUT. Estimate the start time of the DUT test signal playback from the DUT output, and Estimate the speed at which the DUT test signal is played into the DUT from the DUT output.

[0014] The adjustment section of the virtual reference module is configured as follows: The playback speed of the virtual reference test signal is compared with the estimated playback speed of the DUT test signal, and Adjust the playback speed of the virtual reference test signal or the playback speed of the DUT output played to the second channel of the transfer function analyzer, or both, as needed, to compensate for any difference between the estimated playback speed of the DUT test signal and the playback speed of the stored virtual reference test signal.

[0015] The virtual reference player is configured to play the stored virtual reference test signal to the first channel of the transfer function analyzer at a speed adjusted by the adjustment section of the virtual reference module when the presence of the DUT output is detected. Therefore, the speed of the virtual reference test signal is aligned with the speed of the DUT test signal.

[0016] Other aspects of the invention will become apparent from the following description and claims. Attached Figure Description

[0017] Figure 1 This is a diagram of a conventional system that uses a dual-channel transfer function analyzer to measure the transfer function of a loudspeaker or a device under test (DUT) with a loudspeaker assembly.

[0018] Figure 2 This is a diagram of a system that uses a separately stored copy of the device test signal as the input signal to a dual-channel analyzer to measure the transfer function of the DUT.

[0019] Figure 3 This is a diagram of a system used to measure the transfer function of a DUT (Device Under Test). The system uses a separately stored copy of the device test signal as the input signal to a dual-channel analyzer. Figure 2 As shown, a detection section has been added to detect whether there is a DUT output generated in response to a test signal played into the DUT.

[0020] Figure 4 This is a diagram of a system used to measure the transfer function of a DUT. The system uses a separately stored copy of the device test signal as the input signal to a dual-channel analyzer, which has features such as... Figure 3 The detection section shown is supplemented with an adjustment section that compares the playback speed of the virtual reference test signal with the estimated playback speed of the DUT test signal, and adjusts either or both of the playback speed of the virtual reference test signal or the playback speed of the DUT output as needed to compensate for any difference between the estimated playback speed of the DUT test signal and the playback speed of the stored virtual reference test signal.

[0021] Figure 5 yes Figure 3 and Figure 4 The illustration shows the playback speed compensation performed by the adjustment section of the virtual reference module.

[0022] Figure 6 This is a diagram illustrating an example of how playback speed or DUT test signal drifts over time and thus needs to be constantly adjusted over time. Detailed Implementation

[0023] Now refer to the attached diagram, Figure 1 This illustrates a common method for measuring the transfer function of a loudspeaker under test (DUT) using a two-channel transfer function analyzer. For example... Figure 1 As shown, the playback device 11, which stores the test signal x[n], plays the stored test signal, which varies with time denoted as X(t), to the DUT 13, and simultaneously plays a real-time copy of the test signal to the first channel (denoted as C1) of the transfer function analyzer 15. It should be understood that although the playback device 11 is shown as separate from the DUT, the device can be built into the DUT.

[0024] In response to the input test signal X(t) played into the DUT, the DUT produces an acoustic output denoted as Y(t), which is picked up by microphone 17. Microphone 17 converts the acoustic output into an electrical signal Y(t)', which can then be compared with X(t). The signal Y(t) is played into the second channel (denoted as C2) of the transfer function analyzer so that the analyzer can use it to determine the transfer function of the DUT. (It is worth noting that the transfer function is usually implemented in the frequency domain because it is difficult to compute in the time domain. When there is no noise, the transfer function is the ratio of the Fourier transform of the signal: Y(f) / X(f)).

[0025] Figure 2This paper illustrates a method for measuring the transfer function of a loudspeaker using a dual-channel transfer function analyzer when a real-time copy of the DUT test signal is not readily accessible. Here, a copy of the DUT test signal x[n] stored in the DUT playback device 13 is stored separately from the playback device in which it is accessible. This copy is referred to as the "virtual reference test signal" because it is the same signal as the DUT test signal x[n] stored in the playback device 11 used to play the test signal to the DUT. The location where the virtual reference test signal is stored is called the "virtual reference module." Figure 2 It is represented by the number 19. This module can be integrated into a transfer function analyzer or it can be a standalone device.

[0026] As can be seen, the playback device 11 and the virtual reference module 19 are driven by different clocks, designated as clock A in the case of the playback device, which plays the test signal to the DUT, and designated as clock B, which plays the virtual reference test signal to the C1 channel of the transfer function analyzer. As will be described later, the ability to adjust the virtual reference clock is a possible embodiment of the system and method of the present invention.

[0027] By providing an accessible, separately stored virtual reference test signal, a virtual test signal can be played into the C1 channel of the transfer function analyzer instead of the real-time copy of the DUT test signal when a real-time copy of the DUT test signal is inaccessible. Previously, attempting to do this involved detecting the consistency between the virtual reference test signal played into analyzer channel C1 and the DUT output played into analyzer channel C2.

[0028] Figure 3 and Figure 4 An improvement to the basic method of using a separately stored virtual reference test signal as the C1 channel input of the transfer function analyzer 15 is shown. Figure 3 An additional detection section 21 is shown, which is configured to detect the presence of a DUT output generated in response to a test signal being played into the DUT. Once the presence of the DUT output is detected, the detection section estimates the time at which the DUT test signal begins to play, as well as the speed at which the DUT test signal is played into the DUT. Figure 4A further adjustment section 23 is shown, configured to compare the playback speed of the virtual reference test signal with the estimated playback speed of the DUT test signal, and adjust either the playback speed of the virtual reference test signal or the playback speed of the DUT output played to channel C2 of the transfer function analyzer 15, or both, if necessary. In the virtual reference module 19, it can be seen that the start time estimated by the detection section and the playback speed adjustment made to the virtual reference test signal are implemented by the virtual reference player 25, which receives control data from the detection section 21 and the adjustment section 23, as shown in control data paths 27 and 29.

[0029] The following describes further features of the detection and adjustment section of the test signal and virtual reference module 19.

[0030] Test signal

[0031] During the time period required for the test signal to be locked in by the detection section 21, the test signal needs to have energy (e.g., a specific implementation of pink noise) across virtually all times and frequencies. In order to... Figure 3 and Figure 4 For the system to function properly, the signal loaded into playback device 11 must be the same as the signal loaded into virtual reference module 19. Therefore, it is beneficial for the signal to be a uniquely identified time series, such as a signal identified by international standards like AES-75.

[0032] Signal start time detection

[0033] The detection section 21 of the virtual reference module uses an algorithm that is a combination of the following strategies. (These strategies can be implemented in the detection section 21 and the adjustment section 23).

[0034] If the statistics of the DUT output meet the criteria (e.g., the exponential time-weighted RMS level exceeds a threshold), the start of a signal is detected. When the amplitude statistics of the DUT output meet a specific criterion, playback of the virtual reference test signal (sometimes referred to herein as the “virtual reference” or “reference signal”) may begin at the beginning of the reference file x[n]. Virtual reference playback preferably begins playback of the reference file at a predetermined offset; however, using an offset is not a requirement of this invention.

[0035] For a selected segment of the reference signal, the cross-correlation between the reference signal segment and each consecutive segment of the DUT output is calculated. Signal initiation is detected if the cross-correlation amplitude statistics meet a specific criterion, for example, if the cross-correlation peak exceeds the cross-correlation RMS by a selected amount. A sharpening filter can be applied to both the dummy reference and the DUT output signal to improve the detectability of the cross-correlation peak when the reference and measurement signals are aligned. This filter flattens the power spectrum of the reference signal and can be made adaptive, appropriately flattening the spectrum based on the selected reference signal's spectrum.

[0036] If a continuous segment of the reference signal is cross-correlated with the DUT output, statistics on the peak time of the cross-correlation can be collected. If the statistics on the peak time of the cross-correlation meet selected criteria, such as if the peak time variance is below a threshold, the start of the signal is detected.

[0037] When the cross-correlation peak time statistics meet the criteria, virtual reference playback can begin at the beginning of the reference file. Virtual reference playback can start playing the reference file from the time point of the cross-correlation calculation (such as the average peak time).

[0038] Ideally, the detection of the start of the DUT output should be continuously confirmed by calculating the cross-correlation between the DUT output and the played virtual reference. If the statistics of this cross-correlation do not meet the standard, the virtual playback of the reference signal can be stopped.

[0039] Interpolation can be used to refine peak time estimates. This interpolation can be implemented in the time domain, such as quadratic interpolation, or in the frequency domain, such as by estimating the phase slope. An improvement in frequency domain interpolation is incorporating coherence, for example, by assigning higher weights to frequencies with high coherence.

[0040] One possible additional feature is to select several segments from a virtual reference file for concurrent comparison via cross-correlation with the DUT output. If the amplitude of any of these cross-correlations meets a criterion, the start of a signal is detected.

[0041] Playback speed test

[0042] Playback speed can be estimated by comparing the DUT output with the output of the virtual reference player 25. Playback speed can be estimated by measuring the slope of the cross-correlation peak positions of two consecutive segments of the reference signal and the DUT output. The slope of the line fitted to the peak positions over multiple consecutive segments can be used to estimate the playback speed.

[0043] Alternatively, playback speed can be estimated in the frequency domain by observing the rate of phase change of the transfer function. Coherence can be incorporated into playback speed estimation, for example, by assigning higher weights to frequencies with high coherence.

[0044] Furthermore, statistical data collected from the detector can be used to better estimate the fluctuating playback rate. For example, by modeling the playback rate as a slowly changing function, the correlated mean and standard deviation of these estimates allow for the creation of a dynamic model of the DUT's playback rate. This allows for the prediction of the probability distribution of the DUT's future effective playback rate. This information can be combined with the current estimation iterations to more accurately estimate the hidden DUT's true playback rate.

[0045] Playback speed will be estimated in the detection or adjustment section, such as Figure 4 As shown.

[0046] Detector log statistics

[0047] For each consecutive segment of the reference signal and the DUT output, the following statistics can be captured: The amplitude statistics of the cross-correlation between the continuous segments of the reference signal and the DUT output, such as peak value and RMS.

[0048] Time statistics of the cross-correlation between the continuous segments of the reference signal and the DUT output, such as the average value and deviation of the peak time.

[0049] The slope and higher-order derivatives of the peak positions of the cross-correlation between the reference signal and the DUT output of a continuous segment.

[0050] The coefficients of the regression model at the peak position.

[0051] The mean moment and higher-order moments of the squared error term in the regression model at the peak position.

[0052] The quantities derived from the above include "jitter" and "drift".

[0053] Playback speed compensation

[0054] Because the playback of the virtual reference test signal and the DUT test signal is driven by different clocks, there is no guarantee that they will play back at the same speed. This can make it difficult to create an estimate of the transfer function, as the two signals will drift when they align with each other. When the DUT clock and the virtual reference module clock are constant but not synchronized, the peak time in the cross-correlation will vary at a fixed rate, such as... Figure 5 As shown.

[0055] An estimate of the effective playback rate of the DUT can be created by the slope of the recorded position of the cross-correlation peaks of the reference signal and the DUT output across consecutive segments. Alternatively, it could be the slope of a regression model fitted to the same recorded data.

[0056] Compensation for DUT playback speed differences can be achieved by first pre-delaying one of the signals (virtual reference test signal playback or DUT output) in a buffer, and then using interpolation to implement a fractional delay. The magnitude of this delay will vary proportionally to the previously determined estimated playback speed. This imposes some memory requirements on the analyzer and limits the duration of the analysis, as the length of the pre-delay buffer determines the maximum variation that can be compensated for when the two signals are aligned. This compensation will occur during... Figure 4 The adjustment section shown in box 23.

[0057] Alternatively, any discrepancies in playback rate can be addressed by adjusting the playback rate of the virtual reference player 25 in proportion to the estimated playback rate of the reference file on the DUT. This will... Figure 4 This occurs in the virtual reference player 25 shown. This method eliminates the need for pre-allocated buffers, does not limit the duration of the analysis, and does not require compensation for maximum changes in the alignment of the two signals during signal drift. Alternatively, when loading the reference file into the virtual player, it can be interpolated to a higher sampling rate to reduce the computation required during playback.

[0058] While various embodiments of the invention have been described in considerable detail in the foregoing description, this is not to imply that the invention is limited to the described embodiments or their specific details. It should be understood that variations of the described embodiments are possible without departing from the scope and spirit of the invention. For example, although the invention is described as being used in conjunction with a “dual-channel analyzer,” it is understood that the invention can be used with transfer function analyzers having more channels, provided that there are input channels for the virtual reference test signal and input channels for the DUT output.

Claims

1. A method for measuring the transfer function of a device capable of receiving an input and generating an output in response to the input, wherein the measurement is performed by a transfer function analyzer having a first channel for receiving a test signal input and a second channel for receiving an output from a device under test (DUT), the output being generated when the DUT receives the test signal from a playback device, the method comprising the steps of: A. Providing a test signal with desired characteristics on one or more frequency bands, wherein the test signal resides in the DUT playback device as a DUT test signal. B. Store a copy of the DUT test signal separately from the DUT playback device as a virtual reference test signal. C. The DUT playback device plays the DUT test signal to the DUT at playback speed to generate the DUT output played to the second channel of the function analyzer. D. Monitor the DUT output to determine whether the DUT test signal is being played into the DUT. E. When it is determined, based on the monitored DUT output, that the DUT test signal is being played back to the DUT by the DUT playback device, perform the following operations: i) Estimate the start time of the DUT test signal playback, and ii) Estimate the playback speed of the DUT test signal. F. At the estimated start time of the DUT test signal and at the estimated playback speed of the DUT test signal, while playing the DUT output monitored from the DUT to the second channel of the transfer function analyzer, begin playing the virtual reference test signal to the first channel of the transfer function analyzer, and G. Compare the known playback speed of the virtual reference test signal with the estimated playback speed of the DUT test signal, and, if necessary and as required, adjust the playback speed of the virtual reference test signal or the playback speed of the DUT output played to the second channel of the function analyzer, or both, to compensate for any difference between the estimated playback speed of the DUT detection signal and the playback speed of the stored virtual reference test signal. Steps F and G can be performed in any order.

2. The method as described in claim 1, characterized in that, The known playback speed of the virtual reference test signal is continuously compared with the estimated playback speed of the DUT test signal, and the known playback speed of the virtual reference test signal or the playback speed of the DUT output, or both, is continuously adjusted to compensate for any changes in the estimated playback speed of the DUT test signal or the playback speed of the stored virtual reference test signal, or both, over time.

3. The method as described in claim 1, characterized in that, Estimating the start time of the DUT test signal playback includes collecting and recording statistics of consecutive segments of the virtual reference test signal and the DUT output, and using the recorded statistics to estimate the start time.

4. The method as described in claim 3, characterized in that, The collected statistics used to estimate the start time include at least one of the following: the average, root mean square, or standard deviation of the difference between the DUT output and the stored virtual reference test signal.

5. The method as described in claim 1, characterized in that, Estimating the playback speed of the DUT test signal playback includes collecting and recording statistics of consecutive segments of the virtual reference test signal and the DUT output, and using the recorded statistics to determine the playback speed of the DUT output.

6. The method as described in claim 5, characterized in that, The collected statistics include at least one of the following: i) The slope of the peak position of the cross-correlation between the reference signal and the DUT output of two consecutive segments. ii) The slope of the peak-fitted line across multiple consecutive segments, iii) The rate of phase change of the transfer function in the frequency domain.

7. The method as described in claim 5, characterized in that, The known playback speed of the virtual reference test signal is continuously compared with the estimated playback speed of the DUT test signal, and if the cross-correlation statistics do not meet the standard, the virtual playback of the reference signal is stopped.

8. The method as described in claim 1, characterized in that, A copy of the stored virtual reference test signal is played into the first channel of the transfer function analyzer at a predetermined offset relative to the estimated start time of the DUT output.

9. The method as described in claim 1, characterized in that, To estimate the start time of the test signal playback, a fixed segment of the stored virtual reference test signal is compared with each segment of the DUT output to determine whether the expected correlation determined by the correlation function has been achieved.

10. The method as described in claim 9, characterized in that, The continuous segments of the virtual reference test signal are compared with each continuous segment of the DUT output until the desired correlation is achieved.

11. The method as described in claim 9, characterized in that, The method includes monitoring the peak value and RMS value of the correlation function between the stored virtual reference test signal and the DUT output to determine whether the peak value of the correlation function exceeds the RMS value by a selected amount, wherein reaching the selected amount indicates that the test signal has started playing.

12. The method as described in claim 9, characterized in that, A sharpening filter is applied to the virtual reference test signal segment and the DUT output segment to improve the detectability of peaks in the correlation function between the virtual reference test signal segment and the continuous segments of the monitored DUT output.

13. The method as described in claim 12, characterized in that, The sharpening filter is selected based on the characteristics of the test signal.

14. The method as described in claim 9, characterized in that, The timing of the peak value of the correlation function between the virtual reference test signal and the DUT output segment is estimated by interpolation.

15. The method as described in claim 14, characterized in that, Interpolation is performed in the time domain.

16. The method as described in claim 15, characterized in that, Interpolation is performed in the frequency domain.

17. The method as described in claim 1, characterized in that, Playback speed is estimated by comparing the DUT output with a stored virtual reference test signal that is played into the first channel of the transfer function analyzer.

18. The method as described in claim 17, characterized in that, The playback speed is estimated by measuring the slope of the peak position of the correlation function between the reference signal and the continuous segments of the DUT output.

19. The method as described in claim 18, characterized in that, The playback speed is estimated by fitting a line of the position of the peak in the correlation function over multiple consecutive segments and using its slope to estimate the playback speed.

20. A virtual reference test system for measuring the transfer function of a device capable of receiving an input and generating an output in response to the input, wherein the measurement is performed by a transfer function analyzer having a first channel for receiving a test signal input and a second channel for receiving an output from a device under test (DUT), the output being generated when a test signal stored in a DUT playback device is played back to the DUT, the virtual reference test system comprising: A virtual reference module is configured to play the DUT output to the second channel of the function analyzer, and a copy of the test signal stored in the DUT playback device is stored as a virtual reference test signal in the virtual reference module. The virtual reference module includes a detection section, an adjustment section, and a virtual reference player. The detection portion of the virtual reference module is configured to: i) Detect the presence of an output from the DUT, the output being generated in response to a test signal being played into the DUT. ii) Estimate the start time of the DUT test signal playback from the DUT output, and iii) Estimate the speed at which the DUT test signal is played into the DUT from the DUT output, and The adjustment portion of the virtual reference module is configured to: i) Compare the playback speed of the virtual reference test signal with the estimated playback speed of the DUT test signal, and ii) Adjust the playback speed of the virtual reference test signal or the playback speed of the DUT output played to the second channel of the transfer function analyzer, or both, as needed, to compensate for any difference between the estimated playback speed of the DUT test signal and the playback speed of the stored virtual reference test signal, and The virtual reference player is configured to play the stored virtual reference test signal into the first channel of the transfer function analyzer at a speed adjusted by the adjustment portion of the virtual reference module when the presence of the DUT output is detected.

21. The virtual reference test system as described in claim 20, characterized in that, The adjustment portion of the virtual reference module continuously compares the playback speed of the virtual reference test signal with the estimated playback speed of the DUT test signal, and continuously adjusts the playback speed of the virtual reference test signal or the playback speed of the DUT output, or both, as needed, to compensate for any changes in the playback speed of the DUT test signal or the playback speed of the stored virtual reference test signal.

22. The virtual reference test system as described in claim 20, characterized in that, To estimate the start time of the test signal playback, the detection section compares one or more fixed segments of the stored virtual reference test signal with segments of the DUT output to determine whether the desired correlation has been achieved.

23. The virtual reference test system as described in claim 20, characterized in that, In order to estimate the time when the test signal starts playing, the detection section collects and records statistical data of the virtual reference test signal and the continuous segments of the DUT output, and performs cross-correlation on the recorded statistical data.

Images