SMPTE ST 2095-1-2015 Calibration Reference Wideband Digital Pink Noise Signal (Includes Access to Additional Content).pdf

1、 Access to Content for (SMPTE ST 2095-1:2015) (Click here to view the publication) This Page is not part of the original publication: This page has been created by IHS as a convenience to the user in order to provide access to the content as authorized by the Copyright holder of this document. Click

2、 the link(s) below to access the content and use normal procedures for downloading or opening the files. SMPTE_Additional Data Information contained in the above is the property of the Copyright holder and all Notice of Disclaimer or that a certain course of action is preferred but not necessarily r

3、equired; or that (in the negative form) a certain possibility or course of action is deprecated but not prohibited. The keywords “may“ and “need not“ indicate courses of action permissible within the limits of the document. The keyword “reserved” indicates a provision that is not defined at this tim

4、e, shall not be used, and may be defined in the future. The keyword “forbidden” indicates “reserved” and in addition indicates that the provision will never be defined in the future. A conformant implementation according to this document is one that includes all mandatory provisions (“shall“) and, i

5、f implemented, all recommended provisions (“should“) as described. A conformant implementation need not implement optional provisions (“may“) and need not implement them as described. Unless otherwise specified, the order of precedence of the types of normative information in this document shall be

6、as follows: Normative prose shall be the authoritative definition; Tables shall be next; followed by formal languages; then figures; and then any other language forms. 3 Normative References The following standards contain provisions that, through reference in this text, constitute provisions of thi

7、s recommended practice. At the time of publication, the editions indicated were valid. All standards are subject to revision, and parties to agreements based on this recommended practice are encouraged to investigate the possibility of applying the most recent edition of the standards indicated belo

8、w. SMPTE RP 200:2012, Relative and Absolute Sound Pressure Levels for Motion-Picture Multichannel Sound Systems Applicable for Analog Photographic Film Audio, Digital Photographic Film Audio and D-Cinema. AES17-1998 (R2009), AES Standard Method for Digital Audio Engineering Measurement of Digital Au

9、dio Equipment. SMPTE ST 2095-1:2015 Page 5 of 17 pages 4 Terms and Acronyms 4.1 0 dB FS The RMS level of a 997-Hz sine wave whose positive peak value reaches Full Scale Digital. (Note: This may be depicted as “0 dBFS” in some literature. The meaning is the same.) 4.2 Audio sampling rate The number o

10、f audio samples per second, measured in Hz or kHz (one kHz being 1,000 Hz). For example, 48,000 samples per second can be expressed as either 48,000 Hz, or 48.00 kHz. 4.3 Average Responding Meter An average responding meter measures the mean of the absolute value of sample values (or instantaneous v

11、alues) but is calibrated to indicate the RMS value of the waveform when driven with a sine wave. Because the ratio of average to RMS varies with the type of waveform, such a meter can read the RMS value of sine waves accurately, but such a meter will read the RMS value of other waveforms incorrectly

12、. 4.4 Crest factor The ratio of peak value to RMS value of a waveform, in dB. 4.5 dB FSD RMS value relative to FSD. (Note: This may be depicted as “dBFSD” in some literature. The meaning is the same.) 4.6 FSD Full Scale Digital. The maximum numeric value capable of being stored in the bit depth of a

13、 digital signal path. Example: A full scale sine wave has positive peaks with a value of 7FFFFF (hex) and negative peaks with a value of 800000 (hex) in a 24-bit digital audio system. 4.7 Kurtosis A measure of whether the data are peaked or flat relative to a normal distribution. The Kurtosis for a

14、normal distribution is 3. The “excess kurtosis” for a normal distribution is 0. 4.8 Peak value The absolute magnitude of the highest peak amplitude excursion of a signal. 4.9 Pink noise Noise whose power spectral density (W/Hz) is inversely proportional to frequency and whose power per octave (W/oct

15、ave) is constant. 4.10 Pinking filter A filter that converts white noise to pink noise. A pinking filter has a frequency response of -3 dB/octave. 4.11 Root mean square (RMS or rms) The square root of the mean of the squares of sample amplitudes. Different waveforms with the same RMS value will deli

16、ver the same amount of power. The RMS value of a sine wave is 1/2 times its peak value.SMPTE ST 2095-1:2015 Page 6 of 17 pages 4.12 Skewness A measure of the lack of symmetry. The skewness for a normal distribution is 0. 4.13 Unique signal period Period over which no portion of a noise signal repeat

17、s itself. 4.14 White noise Noise whose power spectral density (W/Hz) is constant and whose power per octave (W/octave) is proportional to frequency. 4.15 Wideband pink noise A testing reference signal that includes all frequencies within the bandwidth of a measuring system. 5 Summary of Characterist

18、ics Table 1 summarizes key characteristics of the Calibration Reference Wideband Digital Pink Noise Signal. Table 1 Digital pink noise characteristics Parameter Value Reference RMS Level, full spectrum -18.5 dB FS Section 6.2 RMS Level, 22.4 Hz 22.4 kHz -19 dB FS Section 6.1 Target RMS Level, any 1/

19、3-octave from 20 Hz to 16 kHz -33.74 dB FS Section 6.2 Crest factor 11.5 12 dB Section 6.3 Pink Noise signal bandwidth 10 Hz 22.4 kHz Sections 7.2, 7.3 Energy uniformity 0.25 dB for any 1/3-octave band from 20 Hz 16 kHz Section 7.4 Minimum unique signal period 10 seconds Section 9.1 6 Signal Amplitu

20、de 6.1 Level Criteria (Informative) In order to ensure consistency with the calibration level of dubbing theatres and the level of soundtracks produced over the past several decades, the digital pink noise level defined in this Standard is intended to be consistent with widely accepted reference noi

21、se signals that are currently used in movie soundtrack production. It should be noted that historical discussions of pink noise level are based on SMPTE RP 2002, which defines the reference level of wideband pink noise as being -20 dB FS when measured by an average-responding voltmeter with a bandpa

22、ss filter of 22 Hz to 22 kHz. SMPTE ST 2095-1:2015 Page 7 of 17 pages Regarding meter type, SMPTE RP 200:2012 defines an average responding meter as “a meter which provides a voltage indication proportional to the average value of the rectified signal, with ballistics as described in IEC 60268-17.”

23、In contrast, this Standard defines pink noise levels in the digital domain with RMS measurements. The difference in meter type affects the numerical level readings. For example, the -20 dB FS level per SMPTE RP 200 reads -19 dB FS with an RMS meter even though the actual signal level through the pre

24、scribed bandpass filter is equivalent. Bandwidth also affects the numerical level readings. The spectrum of the pink noise defined in this Standard extends to 10 Hz to better cover the low frequency range of modern loudspeaker system s. When measured without a bandpass filter, the RMS level reads 0.

25、5 dB higher, -18.5 dB FS. 6.2 RMS Value (Normative) The RMS value of the digital pink noise signal having the bandwidth defined in Section 7.1 shall be -18.5 dB FS. The broadband level tolerances are as follows: a. The value for any interval greater than or equal to the unique signal period shall be

26、 within 0.10 dB. b. The value of any one second interval shall be within 0.75 dB. c. The value of any 125 millisecond (ms) interval shall be within 2.00 dB. When analyzed on a 1/3-octave scale, the target level for any 1/3-octave band between 20 Hz and 16 kHz (inclusive) is -33.74 dB FS. See Section

27、 6.1 for an informative comparison to current analog pink noise signals. 6.3 Crest Factor (Normative) The Crest Factor, measured over the unique signal period, shall be between 11.5 dB and 12 dB, inclusive. 6.4 Statistical Criteria (Informative) The heart of a pink noise signal is the white noise so

28、urce from which it is derived. Wideband noise must be uniformly distributed across the spectrum of interest. Another way to say this is it must be totally random. However, not all random noise is perfectly random. In a digital noise signal, true randomness implies the value of each sample within a d

29、efined range is equally probable; i.e., a uniform distribution, with each successive sample being statistically independent of the others. The predictability of computers makes it a challenge to generate perfect randomness in a compact algorithm. Section 9 describes the pseudorandom number generator

30、 used in the example pink noise algorithm. Ideal randomness in a noise signal can be shown by how well it conforms to a normal distribution, also known as a Gaussian distribution the bell-shaped curve. Kurtosis is one measure of distortion in the shape of the Gaussian curve. A high kurtosis distribu

31、tion has a sharper peak and fatter tails, while a low kurtosis distribution has a more rounded peak and thinner tails. Another form of distortion in the distribution curve is skewness, or asymmetry, wherein the tails on either side of the peak have different slopes, rather than being symmetrical mir

32、ror images. Figure 1 shows the ideal Gaussian distribution for the example noise generator with kurtosis better than 3 0.2, and skewness better than 0 0.01. The x axis of the graph shows normalized full scale amplitude (min/max -1 to +1). The y axis is frequency of occurrence, normalized to max bin

33、= 1.0. The graph shows the relative number of samples falling into 71 amplitude bins, each with a width of 0.01, between -0.35 to +0.35. The example noise generator algorithm limits the amplitude of peaks to +/- 0.335 and so the curve goes to 0 at +/- 0.35. SMPTE ST 2095-1:2015 Page 8 of 17 pages Sk

34、ewness and Kurtosis figures are calculated directly from the sample values for pink noise signals generated by the example algorithm at 48 kHz and 96 kHz sample rates, with periodicity ranging from 512 K to 4096 K samples, using the SKEW() and KURT() functions in Microsoft Excel. Figure 1 Normalized

35、 amplitude frequency distribution 7 Signal Spectral Content 7.1 Bandwidth Characteristics The bandwidth of the calibration reference wideband digital pink noise signal shall be 10 Hz to 22.4 kHz. 7.2 Low Frequency Roll-off Characteristics The low frequency roll-off characteristics shall be such that

36、 the signal is attenuated by -3 dB at 10 Hz, and decreases at a minimum rate of 21 dB per octave below 10 Hz, which is the result of applying a conventional 4th-order Butterworth highpass filter with a cutoff frequency of 10 Hz to a pink noise signal with a 3 dB per octave slope. Note that a digital

37、 filter with these characteristics may be realized from the continuous time (analog) prototype using the matched pole-zero mapping method (matched Z transform) or equivalent. 7.3 High Frequency Roll-off Characteristics The high frequency roll-off characteristics shall be such that the signal is atte

38、nuated by -3 dB at 22.4 kHz, and decreases at a minimum roll-off rate of 36 dB per octave above 22.4 kHz. It is noted that for sample rates of 48.00 kHz and 96.00 kHz, a digital filter realized from a 4th-order, continuous-time Butterworth lowpass filter prototype by means of the bilinear transform

39、with frequency pre-warping will easily satisfy this specification. 7.4 Spectral Uniformity The RMS signal level of any 1/3-octave band between 20 Hz and 16 kHz shall fall within 0.25 dB of the nominal target level when measured over a period of time equal to at least one unique signal period. 8 Digi

40、tal Signal Parameters 8.1 Sample Rate The calibration reference wideband digital pink noise signal shall support sample rates of 48.00 kHz and 96.00 kHz. SMPTE ST 2095-1:2015 Page 9 of 17 pages 8.2 Sample Word Size The calibration reference wideband digital pink noise signal shall have a have an int

41、eger word size of at least 24 bits/sample. 9 Signal Duration 9.1 Minimum Unique Signal Period (Normative) The minimum unique signal period for a calibration reference wideband pink noise signal is 10 seconds. 9.2 Signal Duration To extend the duration of the signal for purposes of acoustic measureme

42、nts, the unique signal period may be looped or repeated as required. Alternatively, the unique signal period can be extended beyond the minimum prescribed 10 seconds. 9.3 Measurement Implications of Signal Duration (Informative) In acoustic measurements, factors such as reverberation and background

43、noise often will necessitate measurement durations on the order of one minute or more for full settling and stability, especially at low frequencies. If the minimum unique signal period is repeated, it is expected to provide a stable measurement within the settling time of the room and measurement s

44、ystem. For cases where an extended signal with a longer unique signal period is used, measurement time may need to be longer for the measurement signal to converge to the specified tolerances. 10 Calibration Reference Wideband Digital Pink Noise Signal Algorithm (Informative) 10.1 Noise Generator An

45、nex A describes an available executable noise generator program, and Annex B provides a pseudocode version. The algorithm demonstrates a method of generating band-limited “pink“ noise by filtering the output of a pseudorandom number generator (PRNG). The PRNG used is a linear congruential generator

46、(LCG). The cyclicality of lower order bits in full-period LCGs with modulus 2n (often regarded as a flaw in many other applications) makes it possible to limit its periodicity while maintaining a uniform probability distribution by simply discarding some of the higher order bits. This is useful for

47、our purposes, as it makes it easy to support multiple sampling rates and obtain faster, more repeatable results in spectrum analysis and sound level measurements. 10.2 Spectral Shaping and Filtering The filter network used to shape and band-limit the output spectrum consists of a “pinking“ filter ba

48、nk and a bandpass filter. The pinking filter is a parallel network of six, first-order lowpass filters. The bandpass filter is comprised of a fourth-order highpass and lowpass filter, implemented as two pairs of second-order (biquad) filters. 10.3 Scaling the Output Level To obtain a desired output

49、level, given specific highpass and lowpass filter cutoff frequencies, a gain stage can be applied anywhere in the signal chain. Alternately, scaling gain can be built into the pinking filter (saving one mult per iteration) by multiplying the beta/b1 coefficient of each component filter by a desired gain factor. SMPTE ST 2095-1:2015 Page 10 of 17 pages Annex A Example Band-Limited Pink Noise Generator Executable Script (Informative) A.1 Python Script Included with this standard is the file ST-2095-Generator.py. This file is an executable compu