Abstract
A method, apparatus, and system for measuring and analyzing the effects of dynamics modifying processors on a signal. This new approach utilizes statistical analysis techniques to provide a direct comparison and evaluation between the processed signal and the unprocessed signal's dynamic characteristics. The method identifies and quantifies Effective Dynamic Range, Clip Tolerance, Lower Limit Tolerance, Crest Factor, and Diminuendo Factor, using either peak or r.m.s values. In an alternative embodiment, the invention allows for user adjustment and control of the relative relationship of Crest Factor and Diminuendo Factor, which the user may perceive as loudness.
Claims
1. A quantitative evaluation method for use on a signal dynamics modifier device, comprising using a signal dynamics modifier device which modifies at least one input signal into at least one output signal; measuring the at least one input signal and at least one output signal; displaying the at least one input signal and at least one output signal; using a software application, embodied on a non-transitory computer readable medium, for characterizing the effects of the signal dynamics modifier device on signal dynamic properties by using statistical analysis of digitized signal samples as time variant amplitude value data; calculating, over selected time period, for both the at least one input signal and the at least one output signal, the probability density functions for both signal peaks and signal short term r.m.s. values; and comparing the characteristics of these two sets of probability density functions by calculating median r.m.s. value; maximum peak value considered statistically significant by a user; minimum r.m.s. value considered statistically significant by the user; ratios of the aforementioned median, maximum and minimum values; other statistical characteristics including standard deviation, skew and kurtosis; and composites of the above characteristics; and adjusting the signal dynamics modifier device, based on one or more of the calculations.
2. The quantitative evaluation method for used on a signal dynamics modifier device according to claim 1, wherein the the step of measuring the at least one input signal and at least one output signal is performed by a computer sound card which provides digitized signal in the form of contiguous time variant sample data for both the at least one input signal and at least one output signal to the software application.
3. The quantitative evaluation method for used on a signal dynamics modifier device according to claim 1, in which the software application is resident upon, and the statistical analysis is performed by a circuitry of a dedicated device capable of simultaneous evaluation of at least one input signal to the signal dynamics modifier device and at least one output signal from the signal dynamics modifier device.
4. The quantitative evaluation method for used on a signal dynamics modifier device according to claim 1, in which the statistical analysis and the software application are executed by a digital signal processing algorithm on a digital signal processing device, wherein the algorithm is embodied on a non-transitory computer readable medium, running on the digital signal processing device, and its associated circuitry.
5. The quantitative evaluation method for used on a signal dynamics modifier device described in claim 4, wherein the digital signal processing device is a processor of the signal dynamics modifier device.
6. The quantitative evaluation method for used on a signal dynamics modifier device according to claim 1, wherein the step of adjusting the signal dynamics modifier device further comprises allowing user selection and/or setting of clip tolerance, lower limit tolerance, crest factor, and diminuendo factor.
7. The quantitative evaluation method for used on a signal dynamics modifier device described in claim 1, wherein the step of adjusting the signal dynamics modifier device further comprises allowing user selection and/or setting of short term time window (Tr.m.s.), and long term time window (LTr.m.s).
8. The quantitative evaluation method for used on a signal dynamics modifier device according to claim 1, wherein the step of adjusting the signal dynamics modifier device further comprises dividing either or both of the input and output signals into multiple frequency bands, allowing for the simultaneous analyses and comparison.
9. The quantitative evaluation method for used on a signal dynamics modifier device according to claim 1, wherein the step of adjusting the signal dynamics modifier device further comprises supplying a negative feedback between the output of the signal dynamics modifier device and the input, to more closely meet the dynamics modification effect behavior desired by an end-user.
10. The quantitative evaluation method for used on a signal dynamics modifier device according to claim 1, in which the output signal is an acoustic signal from a loudspeaker system received at a measurement microphone which is passed from a measurement preamplifier to a post processed analysis allowing evaluation of the effects of power compression on signal dynamics in the acoustic domain.
11. The quantitative evaluation method for used on a signal dynamics modifier device according to claim 1, wherein the step of measuring the at least one input signal and at least one output signal is performed by an external signal interface which provides digitized signal in the form of contiguous time variant sample data for both the at least one input signal and at least one output signal to the software application.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
(1) FIG. 1 shows a system architecture diagram of the present invention, using an embodiment that allows user measurement and analysis of a dynamics-modifying device, such as a processor, and the interaction to various pre- and post-processed parameter ratios. FIG. 1B shows a system architecture diagram of the present invention, using an embodiment that allows the user the ability to measure, analyze, and adjust a dynamics-modifying device, such as a processor, and the parameters and interactions that can be examined and adjusted.
(2) FIG. 2 shows a representative chart or graph of a Probability Density Function of signal peaks with upper bound curtailment.
(3) FIG. 3 shows a representative chart or graph of the Probability Density Function of pre- and post-processed median r.m.s. values, taken over LT.sub.r.m.s. (or T.sub.pop, the upper limit of LT.sub.r.m.s.), and their relationship to pre-processed diminuendo and crest factors.
(4) FIG. 4 shows a representative chart or graph of the Probability Density Functions for the original r.m.s. signal voltage and compressed voltage output values, taken over LT.sub.r.m.s, in which lower bound curtailment is present.
(5) FIG. 5 shows a chart of Probability Density Function for r.m.s. voltage output values, taken over LT.sub.r.m.s, and its relationship to LLT, Crest Factor, and Clip Tolerance Level.
(6) FIG. 6 shows a chart of Probability Density Function for r.m.s. voltage output values, taken over LT.sub.r.m.s., and its relationship to Diminuendo Factor and Crest Factor.
(7) FIG. 7 shows a chart of Probability Density Function for r.m.s. voltage output values, taken over LT.sub.r.m.s., and its relationship to Effective Dynamic Range.
(8) FIG. 8 shows a chart of Probability Density Function for the original r.m.s. and compressed voltage output values, taken over LT.sub.r.m.s., with the resultant curves normalized to the mean level to expose the changed nature of the distribution curve tails.
(9) FIG. 9 shows an example of a Probability Density Function chart for peak voltages, highlighting the upper bound of the minimum significant probability and its relationship to Clip Tolerance.
(10) FIG. 10 shows an example of a Probability Density Function, highlighting the relationship between the lower level signal tolerance (LLT), and the lower bound for the minimum significant probability density.
(11) FIG. 11 shows a Probability Density Function chart for r.m.s. value or voltage, measured over LT.sub.r.m.s., showing lower bound curtailment caused by the noise floor.
(12) FIG. 12 shows a chart relating Probability Density Function of pre- and post-processed median r.m.s. values and their relationship to post-processed Diminuendo and Crest factors.
(13) Note that for FIGS. 2 through 12 the horizontal axis is signal level in dB and the vertical axis is unscaled probability.
DETAILED DESCRIPTION AND SUMMARY OF THE INVENTION
(14) This description does not limit the invention, but rather illustrates its general principles of operation. Examples are illustrated with the accompanying drawings. A variety of drawings are offered, showing the present invention and the probability density functions that are used in its algorithm.
(15) A Probability Density Function for audio input or output voltage, whether peak or r.m.s., shows a sequence of data values that, ideally, monotonically increase, until a maximum, and then monotonically decrease, until a minimum.
(16) When the sequence of signal absolute value data values in the data stream reverses direction, it indicates an inflection point. When the inflection point in the absolute value changes from an increasing sequence to a decreasing sequence this indicates a signal peak event. Most often, such an inflection point will be a single point of data indicating a momentary peak in the modulus of signal absolute value, but it can also be a sequence of values, when the signal is clipped.
(17) FIG. 1 shows the present invention, along with a probability density analysis chart. The invention relies on an input signal 1. The signal is passed through a dynamics processor 2, which is the device under test (DUT) 2. The DUT 2 is any device that has a processor that can affect the signal dynamics: compression, expansion, dynamic range, or clipping of a signal. The DUT 2 can include, but is not limited to, a DSP, a signal processor, an integrated amplifier with a DSP that has a compressor, a computer with sound card and software, or an audio system. The DUT 2 outputs an output signal 3. The invention has a processor that can compare the pre-processed probability density 4 of the signal and the post-processed probability density 5 of the signal. The processor can received value settings for root mean square time constant, T.sub.rms, 6; long-term root mean square time window, LT.sub.rms, 7; total performance time, T.sub.pop, 8; Clip Tolerance, CT, 9; and Low Level Tolerance, LLT, 10.
(18) The Probability Density Function pair 11 shows the relationship between a typical pre-processed 13 and post-processed 14 Probability Density Functions. The processor can easily compare the pre-processed 4 and post-processed 5 Probability Density Functions. The processor can assess a number of parameters and ratios 16, between pre- and post-processed values, including, but not limited to, Dynamic Range, Crest Factor, Diminuendo Factor, Relative Loudness, Median r.m.s., Kurtosis, and Skew.
(19) The median value to the noise floor is the Diminuendo Factor 15. In this drawing, the Diminuendo Factor 15 is for the post-processed median r.m.s. value. The post-processed Crest Factor 12 is the CT Level to the median r.m.s. value (corresponding to the peak of the probability density distribution) of the Probability Density Function.
(20) FIG. 1B shows an alternative embodiment for this analysis apparatus used to evaluate the effect of power compression in a loudspeaker system. In this case the output signal 103 is derived from an acoustic signal 106, which is the output of the DUT 109. In this embodiment, the DUT 109 is an electro-acoustic system 109, composed of, at a minimum, an amplifier 108 and a loudspeaker 107. The DUT receives an input 1. A measurement microphone 105 is connected to a measurement preamplifier 104 to provide a signal to the post compression probability density analysis computation 5. The measurement output signal 103 shows the result of such system. Several measurements are consistent between FIG. 1 and FIG. 1B: the pre-processed Probability Density Function 4, T.sub.rms 6, LTrms 7, Tpop 8, Clip Tolerance 9 or CT 9, and Low Level Tolerance 10 or LLT 10.
(21) FIG. 2 shows an example of a Probability Density Function 112 chart for signal peak levels 111 in which the upper level curtailment typical of overload or clipping 110 is evident in a sharp discontinuity in the upper tail of the curve 110. The curve shows a sharp upward discontinuity 110 caused by repeated level events at the system maximum which, absent this level constraint, would occur at higher levels than the level at 110.
(22) FIG. 3 shows a graph 11 between pre-13 and post-processed 14 Probability Density Functions 13, 14. The relationship between the pre-processed Diminuendo Factor 18 and pre-processed Crest Factor 19 is illustrated.
(23) FIG. 4 shows a Probability Density Function graph 11, comparing the original signal r.m.s. 118 and compressed 17 Probability Density Function. The tail 119 of the original r.m.s. and the compressed 17 curves is shown for an example of lower bound curtailment due system constraints caused by the system noise floor.
(24) FIG. 5 shows an r.m.s. Probability Density Function graph 118, specifically calling out its median level as the peak of the curve. The minimum significant Probability Density Function (upper) 21 defined as Clip Tolerance probability value 23 and minimum significant Probability Density Function (lower) 20 defined as Lower Limit Tolerance probability value 219 are shown. The level at Clip Tolerance 23 is obtained from the intersection of peak level Probability Density Function curve 51 of FIG. 9 and the minimum significant probability density (upper) 21. The Crest Factor 22 is the difference between the Clip Tolerance Level (CT) 23 and the median r.m.s. level (peak of 118). The Lower Level Signal Tolerance (LLT) 219 is the intersection of the r.m.s. Probability Density Function curve 118 and the minimum significant probability density (lower) 20, which is usually considered the noise floor.
(25) FIG. 6 shows an r.m.s. Probability Density Function 118, specifically calling out its median level as the peak of the curve. The minimum significant Probability Density Function (upper) 21 defined as Clip Tolerance probability value 23 and minimum significant probability density (lower) 20 defined as Lower Limit Tolerance (LLT) value 219 are shown. The level at Clip Tolerance 23 obtained from the peak level Probability Density Function 51 of FIG. 9 is shown, and, as the peak level corresponding to the minimum significant probability density (upper) 21. The Crest Factor 22 is the difference between the Clip Tolerance Level (CT) 23 and the median r.m.s. level (peak of 118). The Lower Level Signal Tolerance (LLT) 219 is the intersection of the r.m.s. probability density 118 and the minimum significant probability density (lower) 20. The Diminuendo Factor 24 is the difference between the Median r.m.s. level peak (callout of 118) and the Lower Level Signal Tolerance (LLT) 219.
(26) Comparing FIG. 6 to FIG. 7 shows that the Effective Dynamic Range 25 (FIG. 7) is equal to the Diminuendo Factor 24 plus Crest Factor 22 (FIG. 6).
(27) FIG. 8 shows the same curves as FIG. 12, being original signal r.m.s. Probability Density Function 26 compared to a compressed probability density function 27. In this chart, the curves have been normalized to the median level to visually highlight the characteristic of the tails 28, 29 of each curve in which the compressed curve tails show more kurtosis, i.e. steeper tails, than the original signal curve.
(28) FIG. 9 shows an example Probability Density Function chart for signal peak levels in which the minimum significant probability density (upper) for highest level signal peaks considered statistically significant has been set 21. This Clip Tolerance 23 probability corresponds to the intersection of the peaks curve 51, and the minimum significant probability density 21.
(29) FIG. 10 shows an example of a Probability Density Function for r.m.s. voltage value 118, using a T.sub.rms time window, highlighting the LLT 219 and the minimum significant probability density (lower) 20. The LLT 219 is the intersection of the lower portion fo the r.m.s. Probability Density Function 118 and the minimum significant probability density (lower) 20. The minimum significant probability density (upper) 21 is also shown for reference.
(30) FIG. 11 shows a curve 63 of a Probability Density Function 62 for the signal short term r.m.s, taken over time value LT.sub.rms. The lower portion of the curve shows the curtailment of the lower bound at the noise floor 61.
(31) FIG. 12 shows a comparison between pre- 13 and post-processed 14 Probability Density Functions 11. The relationship between the Post-processed Diminuendo Factor 15 and Post-processed Crest Factor 12 is apparent. Similarly the same quantities could be calculated for the Pre-processed values, as shown in FIG. 3.
