Method and apparatus for determining the energy of a signal

Abstract

The invention describes two alternative methods and corresponding devices for producing an energy signal yn whose amplitude values represent the energy of an electrical signal sn. A first method and the corresponding device calculate the energy signal yn according to the equation 1yn=tau·sn22·yn-1+(1-tau2)·yn-1

Description

BACKGROUND OF THE INVENTION

[0001] The invention relates generally to the field of signal processing, and in particular to the field of channel source audio selection circuitry for an audio signal processor.

[0002] The invention relates to methods and devices for determining the energy of a signal according to the preambles of method claims 1 and as well as device claims 4 and 9.

[0003] An ideal exponentially weighted root mean square (erms) detector for determining the energy of an analog signal s(u) is well known from the theory of electrical signal processing. The energy may be determined sequentially by executing three method steps: squaring the signal, integrating the squared signal, and extracting the root of the integrated signal. These method steps are also reflected in the following equation: 3$y\left(s\left(u\right),t\right)=\sqrt{\frac{1}{T}{\int }_{-\infty }^t{s}^2\left(u\right)·e^{-\frac{\left(t-u\right)}{T}}du}$

[0004] which describes the functional principle of the ideal analog detector. The detector determines the energy y of the signal s(u) as its RMS value, weighted exponentially with a time constant T, as a function of time t.

[0005] The conventional digital ERMS detectors used in practice receive a digital signal sn at their input in order to deliver a digital energy signal yn from the output, with the amplitude values of the energy signal representing the energy of the signal sn. They are based on method steps known from theory. To convert these method steps, as a rule as shown in FIG. 4a, they are formed as a series circuit consisting of a squaring element 1, a low-pass filter 2, and a root extractor 3.

[0006] FIG. 4 b shows one possible digital implementation for such a series circuit. Accordingly, squaring element 1 is formed from a first multiplying element 410 that multiplies the digital signal sn by itself in order to provide the squared signal s2n and its output. The squared signal is then supplied as the input signal to the digital low-pass 2 weighted with a factor tau.

[0007] Within the low-pass 2, the input signal is fed as a first summand to an adding element 420 which delivers at its output the desired energy signal but squared as y2n. As the second summand, the output signal y2n fed back through a state memory 430 weighted with the factor 1−tau is supplied to adding element 420.

[0008] Then the squared energy signal y2n is subjected by a root extractor 3 connected downstream to root extraction. The root extractor 3 comprises a second adding element 440 that receives the squared energy signal y2n and outputs the desired energy signal yn at its output. To calculate the energy signal yn, adding element 440 adds the squared energy signal to two additional signals. These are firstly the energy signal yn-1 fed back through a second state memory 450 from its own output and secondly a signal y2n-1 that is obtained by squaring and negating from the fed-back energy signal yn-1.

[0009] The conventional calculation of the energy signal yn shown here suffers from the following disadvantages:

[0010] By squaring the signal sn, its dynamic range is sharply increased. It is only possible to store the squared signal in a memory with a very large word width.

[0011] The square root routines used in conventional extraction of square roots converge slowly, often as a function of the magnitude of the amplitude of their input signal. They are therefore unsuitable for use in systems that require rapid convergence behavior of the detector, such as compander-expander systems for example.

SUMMARY OF THE INVENTION

[0012] The goal of the invention is to improve the methods and devices according to the species for determining the energy of a signal in such fashion that they exhibit faster convergence behavior and reduced computation cost, as well as less storage space.

[0013] This goal is achieved by the method steps claimed in method claims 1 and 8, as well as by the features claimed in device claims 4 and 9. Additional advantageous embodiments of the invention are the subjects of the subclaims.

[0014] According to a first embodiment of the method according to the invention, as shown in independent method claim 1, the goal is achieved especially by the fact that an energy signal yn is calculated whose amplitude values represent the energy of signal sn, according to the equation 4$\begin{array}{cc}{y}_n=\frac{\mathrm{tau}·s_n^2}{2·y_{n-1}}+\left(1-\frac{\mathrm{tau}}{2}\right)·y_{n-1}\text{with}\text{tau: a specified parameter and}\text{n}\text{: the clock pulse, whereby in Equation 1 the method steps of squaring, lowpass filtration, and root extraction are combined.}& \left(1\right)\end{array}$

[0015] The parameter tau determines the time constant for the exponential weighting. It bears the following relationship approximately with the time constant T in the analog formula (page 3): 5$\mathrm{tau}=\&LeftBracketingBar;0,5·\frac{\left(1-e\frac{i}{T·\mathrm{fs}}\right)}{\left(0,5-e\frac{i}{T·\mathrm{fs}}\right)}\&RightBracketingBar;$$\text{where:}$$I=\mathrm{sqrt}\left(-1\right)$$\mathrm{fs}=\text{sampling frequency of the digital system.}$

[0016] Example:

[0017] with fs=48 kHz and T=20 ms, tau is approximately 0.00104.

[0018] The explanations of the parameters tau and n likewise apply to all the following equations in the specification.

[0019] In the method performed according to Equation 1, the integrated method step of extracting the root exhibits a quadratic convergence behavior which has a very advantageous effect on the dynamic behavior of the entire process.

[0020] In addition, the method step of root extraction in the step of low-pass filtration is integrated, so that the calculation expense is reduced.

[0021] To work the method, it is no longer necessary to store the squared signal s2n with its large dynamic range; instead, it is sufficient to store the amplitude values found for the energy signal yn, which, because of the fact that the root has been extracted, exhibit a value range considerably reduced by comparison with the dynamic range of the squared signal s2n. For this reason, in the method according to the invention, a memory with a relatively small word width can be used.

[0022] According to one advantageous improvement on the method, the calculation of the first summand in Equation 1 includes the step “multiply signal sn by an auxiliary signal tau/2yn-1 to obtain a product signal” and “multiply the product signal by the signal sn” or alternatively the steps “square the signal sn and “multiply the squared signal s2n by the auxiliary signal tau/2yn-1”. Depending on the selected implementation of the method, one alternative or the other can contribute to reducing the computation and storage cost.

[0023] The advantages given for the first embodiment of the method according to the invention apply similarly to a corresponding device.

[0024] It is advantageous for the corresponding device according to the invention to have an inverter to receive the energy signal yn and to output an inverted signal bn=tau/2yn, designed so that it, by the equation 6$\begin{array}{cc}{b}_n=b_{n-1}·2\frac{k}{\mathrm{tau}}·\left(\frac{\left(1+k\right)·\mathrm{tau}}{2k}-\left(b_{n-1}·y_n\right)\right)& \left(2\right)\end{array}$

[0025] with

[0026] k: a constant preferably between 0.5 and 1, it provides the specified link between the inverted signal bn and the energy signal yn. The convergence behavior of the inverter can be influenced by the choice of the constant k; in this manner, the quadratic convergence behavior of the entire device can be optimized as well.

[0027] (Note: the explanation for the constant k likewise applies to all the following equations in the specification.)

[0028] According to a second embodiment of the method according to the invention, as shown in the independent method claim 8, the goal is achieved especially by virtue of the fact that the energy signal yn is calculated according to the equation 7$\begin{array}{cc}{y}_n=y_{n-1}+\frac{\mathrm{tau}}{2y_{n-1}}\left(s_n^2-y_{n-1}^2\right)& \left(3\right)\end{array}$

[0029] with the method steps of squaring, low-pass filtration, and root extraction being combined in Equation 3.

[0030] In the method performed according to Equation 3, the method step of root extraction exhibits a quadratic convergence behavior, which has a highly advantageous effect on the dynamic behavior of the entire method.

[0031] In addition, the method step of low-pass filtration is integrated in the step of root extraction, so that calculation cost is reduced.

[0032] To work the method, it is no longer necessary to store the squared signal s2n with its wide dynamic range; instead, it is sufficient to store the currently determined value for the energy signal yn, which, because of the root extraction that has been performed, has a value range that is considerably reduced by comparison with the dynamic range of the squared signal s2n. For this reason, in the method according to the invention, a memory with a much reduced word width can be used.

[0033] The listed advantages of the second embodiment of the method according to the invention also apply to a corresponding device.

[0034] For the device according to the second embodiment, it is likewise advantageous for the reasons given for it to have the inverter described above.

[0035] These and other objects, features and advantages of the present invention will become more apparent in light of the following detailed description of preferred embodiments thereof, as illustrated in the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWING

[0036] FIG. 1 a is an implementation of a detector according to the invention according to a first embodiment;

[0037] FIG. 1 b is an alternative embodiment of the detector according to FIG. 1a;

[0038] FIG. 2 is an implementation of a detector according to the invention according to a second embodiment;

[0039] FIG. 3 is an implementation of an inverter according to the invention;

[0040] FIG. 4 a is a block diagram of a conventional detector; and

[0041] FIG. 4 b is an implementation of the conventional detector from FIG. 4a.

DETAILED DESCRIPTION OF THE INVENTION

[0042] A detailed description of the two preferred embodiments of the invention follow, with reference to FIGS. 1a, 1b, 2, and 3.

[0043] The first embodiment of a detector according to the invention, shown in FIG. 1a, calculates an energy signal yn whose amplitude values represent the energy of the signal sn according to the following equation: 8$\begin{array}{cc}{y}_n=\frac{\mathrm{tau}·s_n^2}{2·y_{n-1}}+\left(1-\frac{\mathrm{tau}}{2}\right)·y_{n-1}.& \left(1\right)\end{array}$

[0044] For hardware or software implementation of Equation 1, the detector has a first multiplying element 110 and a second multiplying element 120, an adding element 130, and a state memory 140.

[0045] The input signal sn is supplied to two inputs of the first multiplying element 110 so that the squared input signal s2n is available at its output. The squared signal is then fed to the second multiplying element 120 that multiplies it by an auxiliary signal tau/2yn-1. A product signal calculated by the second multiplying element 120 is supplied to a first input of adding element 130, which provides the energy signal yn to be determined at its output. To calculate the energy signal yn, the latter is fed back through state memory 140, weighted with a factor 1−tau/2, to a second input of adding element 130.

[0046] With the first embodiment of the detector according to the invention wired in this fashion, a low-pass filter is obtained in which a root extractor is integrated.

[0047] An alternative design, shown in FIG. 1b, of the first embodiment of the detector differs from the design according to FIG. 1a only in the wiring of the two multipliers 110 and 120. According to the alternative design according to FIG. 1b, the multiplying element 110 multiplies the signal sn initially by the auxiliary signal tau/2yn-1. The product signal at the output of the first multiplying element is then supplied to multiplying element 120, which multiplies the product signal times the signal sn.

[0048] In the alternative designs of the detector according to FIGS. 1a and 1b, the product signals at the outputs of the second multiplier 120 are identical.

[0049] According to the second embodiment of the detector according to the invention shown in FIG. 2, an energy signal yn whose amplitude values represent the energy of a signal sn are calculated according to the following equation: 9$\begin{array}{cc}{y}_n=y_{n-1}+\frac{\mathrm{tau}}{2y_{n-1}}\left(s_n^2-y_{n-1}^2\right).& \left(3\right)\end{array}$

[0050] For hardware or software implementation of Equation 3, the detector according to the second embodiment has a first multiplying element 210 and a second multiplying element 230, a first adding element 220 and a second adding element 240, a state memory 260, and a squaring element 250.

[0051] The signal sn is supplied to the two inputs of the first multiplying element 210, so that the squared signal s2n is provided at the output of the first multiplying element. The squared signal is delivered to a first input of the first adding element 220 whose output signal is supplied to the second multiplying element 230. The second multiplying element 230 multiplies the output signal of the adding element 220 times the auxiliary signal tau/2yn-1 and delivers the resultant product signal to a first input of the second adding element 240, which delivers the energy signal yn to be produced at its output. The output signal of adding element 240 is fed back through the state memory 260 to a second input of the second adding element 240. In addition, the output signal yn-1 of the state memory 260 is fed back following negation through the squaring element 250 connected downstream to the second input of the first adding element 220.

[0052] With the second embodiment of the detector according to the invention wired in this fashion, a root extractor is formed into which a low-pass filter is integrated.

[0053] FIG. 3 shows the implementation of an inverter for converting the energy signal yn into a signal bn=tau/2yn. The inversion takes place according to the invention by the following equation: 10$\begin{array}{cc}{b}_n=b_{n-1}·2\frac{k}{\mathrm{tau}}·\left(\frac{\left(1+k\right)·\mathrm{tau}}{2k}-\left(b_{n-1}·y_n\right)\right).& \left(2\right)\end{array}$

[0054] For a hardware or software implementation of Equation 2, the inverter has a first multiplying element 310 and a second multiplying element 330, an adding element 320, and a state memory 340.

[0055] The first multiplying element 310 multiplies the received energy signal yn by the output signal bn from the inverter fed back via the state memory 340. The output signal of the first multiplying element 310 is negated and supplied to adding element 320, which adds it to the summand (1+k)*tau/2k. The sum signal at the output of adding element 320 is weighted with the factor 2k/tau and supplied to the second multiplying element 330, that multiplies it by the output signal bn-1 of the state memory 340. At the output of the second multiplying element 330, the output signal bn from the inverter is provided, and supplied simultaneously to the state memory 340 as an input signal.

[0056] The following is true of all versions of the erms detector:

[0057] The method supplies inaccurate results if yn suddenly rises sharply, so that: yn>c*yn-1. Here c is a constant which can be chosen depending on the desired accuracy. A good accuracy is obtained with c<1.5.

[0058] In order to improve the accuracy in this case, the following iterations can be performed as well. This is necessary only with an abrupt change in signal energy and strict requirements for accuracy. This results in the following modification: 11$y_{n,m}=\mathrm{tau}·\frac{s_n^2}{2y_{n,m-1}}+\left(I-\frac{\mathrm{tau}}{2}\right)·y_{n-1}$$\mathrm{and}$$y_{n,m}=y_{n-1}+\frac{\mathrm{tau}}{2y_{n,m-1}}\left(s_n^2-y_{n-1}^2\right)$

[0059] where:

[0060] m=additional iteration index.

[0061] The iterations are continued until 12$\frac{1}{c}<\&LeftBracketingBar;\frac{y_{n,m}}{y_{n,m-1}}\&RightBracketingBar;<c$

[0062] As a rule, even with sharp jumps in the input signals, two to three iterations will suffice.

[0063] Although the present invention has been shown and described with respect to several preferred embodiments thereof, various changes, omissions and additions to the form and detail thereof, may be made therein, without departing from the spirit and scope of the invention.

LIST OF REFERENCE NUMBERS

[0064] energy signal . . . yn

[0065] signal . . . sn

[0066] inverter signal . . . bn

[0067] squaring element . . . 1

[0068] low pass . . . 2

[0069] root extractor . . . 3

[0070] first multiplying element . . . 410

[0071] second multiplying element . . . 440

[0072] adding element . . . 420

[0073] first state memory . . . 430

[0074] second state memory . . . 450

[0075] squaring element . . . 460

[0076] First embodiment of the detector

[0077] first multiplying element . . .110

[0078] second multiplying element . . . 120

[0079] adding element . . . 130

[0080] state memory . . . 140

[0081] Second embodiment of the detector

[0082] first multiplying element . . . 210

[0083] second multiplying element . . . 230

[0084] adding element . . . 220

[0085] state memory . . . 260

[0086] squaring element . . . 250

[0087] Inverter

[0088] first multiplying element . . . 310

[0089] second multiplying element . . . 330

[0090] adding element . . . 320

[0091] state memory . . . 340

Claims

1. Method for determining the energy of a signal (sn), comprising the steps of: a) squaring the signal (sn) and providing a squared signal (s2n) indicative thereof; b) low-pass filtering of said squared signal (s2n) and providing a low-pass-filtered signal indicative thereof; and c) extracting the root of said low-pass-filtered signal to produce an energy signal (yn) whose amplitude values represent the energy of signal (sn); characterized in that the energy signal (yn) according to the equation (1) 13yn=tau·sn22·yn-1+(1-tau2)·yn-1with tau: a specified parameter and n: the clock pulse is calculated from signal (sn), with the steps a to c being combined into equation (1) and with step (c) of the root extraction being integrated into step b) of low-pass filtration.

2. Method according to claim 1, characterized in that calculation of the first summand comprises the following steps in equation (1): multiplying the signal (sn) by an auxiliary signal (tau/2yn-1) to produce a product signal; and multiplying the product signal by the signal (sn).

3. Method according to claim 1, characterized in that the calculation of the first summand in equation (1) includes the following steps: squaring the signal (sn); and multiplying the squared signal (s2n) by the auxiliary signal (tau/2yn-1).

4. A device for determining the energy of a signal (sn), comprising: a squaring element to receive the signal (sn) and to output the squared signal (s2n); a low-pass filter for low-pass filtration of the squared signal (s2n); and a root extractor for extracting the root of the low-pass-filtered squared signal and producing an energy signal (yn) whose amplitude values represent the energy of signal (sn); characterized in that the device is designed so that it is created by the equation 14yn=tau·sn22·yn-1+(1-tau2)·yn-1with tau: a specific parameter and n: the clock pulse, if creates the specified link between the signal (sn) received by it and the energy signal (yn) output by it, with the link simulating the squaring element, low-pass filter, and root extractor in such fashion that the root extractor is integrated into the low-pass filter.

5. The device according to claim 4, characterized by a first multiplying element (110) which multiplies the received signal (sn) with an auxiliary signal (tau/2yn-1) and delivers a product signal at its output; and a second multiplying element (120) which multiplies the product signal times the signal (sn).

6. The device according to claim 4, characterized by a first multiplying element (110) which squares the received signal (sn) and a second multiplying element (120) that multiplies the squared signal by an auxiliary signal (tau/2yn-1).

7. The device according to claim 4, characterized in that it comprises an inverter for receiving the energy signal (yn) and for outputting an inverted signal bn=(tau/2yn), designed so that it creates, by the equation (2)

8. The device according to claim 5, characterized in that it comprises an inverter for receiving the energy signal (yn) and for outputting an inverted signal bn=(tau/2yn), designed so that it creates, by the equation (2)

9. The device according to claim 6, characterized in that it comprises an inverter for receiving the energy signal (yn) and for outputting an inverted signal bn=(tau/2yn), designed so that it creates, by the equation (2)

10. Method for determining the energy of a signal (sn), comprising the steps of: a) squaring the signal (sn); b) low-pass filtration of the squared signal (s2n); and c) extracting the root of the low-pass-filtered signal from step b) to produce an energy signal (yn) whose amplitude values represent the energy of signal (sn); characterized in that the energy signal (yn) according to the equation (2) 18yn=yn-1+tau2⁢yn-1⁢(sn2-yn-12)with tau: a specified parameter and n: the clock pulse, is calculated from the signal (sn), with steps a) to c) being combined in equation (2) and the step b) of low-pass filtration being integrated into the step c) of root extraction.

11. A device for determining the energy of a signal (sn), comprising: a squaring element to receive signal (sn) and to output the squared signal (s2n); a low-pass filter for low-pass filtration of the squared signal (s2n); and a root extractor for extracting the root of the low-pass-filtered squared signal and producing an energy signal (yn), whose amplitude values represent the energy of signal (sn); characterized in that the device is designed so that, by the equation (3) 19yn=yn-1+tau2⁢yn-1⁢(sn2-yn-12)with tau: a specified parameter and n: the clock pulse, creates the specified link between the signal (sn) received by it and the energy signal (yn) output by it, with the link simulating the squaring element, low-pass filter, and root extractor in such fashion that the low-pass filter is integrated in the root extractor.

12. The device according to claim 11, characterized in that it has an inverter to receive energy signal (yn) and to output the signal bn=(tau/2yn) designed so that the equation (2)