Online-Seminar Psychoakustik 2 Transiente Vorgänge, tonale Komponenten und Modulation. Andreas Langmann. Siemens AG 2018

Transcription

1 Online-Seminar Psychoakustik 2 Transiente Vorgänge, tonale Komponenten und Modulation Andreas Langmann Siemens AG 2018

2 Transiente Metriken Time varying Loudness N10 Kurtosis Wavelets Tonale Metriken Tonality Tone to Noise Prominence Ratio Modulations Metriken Hilbert Envelope & Modulation Theory Fluctuation Strength and Roughness

3 Transiente Metriken Time varying Loudness N10 Kurtosis Wavelets

4 Clicks, Clunks and Pings! What is a Transient? Event less than 1 second in duration, usually in milliseconds Impulsive, changing amplitude rapidly Traditional FFT techniques are not always effective in analyzing Types of signals: keyboard clicks, injector ticks, piston slap, door slam, other human actuated sounds Page 5

5 Loudness N10

6 N10 Loudness Page 7

7 N10 Loudness Page 8

8 N10 Loudness Page 9

9 N10 Loudness Page 10

10 N10 Loudness Page 11

11 Kurtosis

12 Kurtosis - Histogram amplitude time # samples C L A S S E S C L A S S E S Page 13

13 Histogram: Square Wave amplitude time # samples C L A S S E S C L A S S E S Page 14

14 Histogram: Impact amplitude time # samples C L A S S E S C L A S S E S Page 15

15 Histogram: Gaussian Random amplitude time # samples C L A S S E S C L A S S E S Page 16

16 Distribution Square Wave Gaussian Random Impact # samples Kurtosis (k) is a unitless parameter that measures the relative sharpness or flatness of a distribution for a signal relative to a normal or Gaussian one. Page 17

17 Kurtosis Square Wave Gaussian Random Impact # samples <0 =0 >0 Page 18

18 Kurtosis Equal 0 -> Normal Distribution, "mesokurtic. ; example signal: Guassian Random Negative ( <0 ) Wide Distribution, "platykurtic ; example signal: Square/Sine Wave Positive ( >0 ) Narrow Distribution, "leptokurtic ; example signal: Impact Page 19

19 Time-Frequency Analysis: Wavelets

20 Time-Frequency Analysis: Wavelets 13.5 Hz, 5 Volts 5 seconds Page 21

21 Time-Frequency Analysis: Wavelets Freq Hz 10 In a perfect world, this is our frequency spectrum vs. time time Page 22

22 Time-Frequency Analysis: Wavelets Perform FFT T=0.5 sec T=1/(Df) Df = 2 Hz Freq 20 Leakage in frequency domain due to Df, amplitude reduced time Page 23

23 Time-Frequency Analysis: Wavelets Freq 20 Leakage in time domain due to T, incorrect signal determination Perform FFT Df = 0.5 Hz 10 T=1/(Df) T= 2 sec time Page 24

24 Time-Frequency Analysis: Wavelets Traditional FFT methods do not work well on transient events: Good Time Resolution Bad Frequency Good Frequency Resolution Bad Time Solution: Wavelets Alternative Time-Frequency Methods Not FFT based (per se) Generate large amount of information over small time duration (I.e., analyze only milliseconds worth of data) Page 25

25 Time-Frequency Analysis: Wavelets Freq. Freq. 8 Hz 8 Hz 2 Hz 2 Hz Traditional FFT Time Wavelet Time Page 26

26 FFT Wavelet Page 27

27 Tonale Metriken Tonality Tone to Noise Prominence Ratio

28 Tonal Examples Tonal noise issues characterized by: Distinct audible peaks at discrete frequencies Opposite of broadband db Hz Vuvuzela Turbocharger Mosquito Gear Whine Page 29

29 TONAL METRICS Tonality Tone-to-Noise Prominence Ratio DIN provides an iterative method to detect tones by comparing the levels of each spectral line 1 Tonality Unit (t. u.) has the tonality of a 1 khz sine 60dB ECMA-74 and ISO 7779 describe the calculation Levels of the prominent discrete tones are compared to the noise level in the same critical band ECMA-74 and ISO 7779 describe the calculation Average SPL of the critical band centered around the tone is higher than surrounding critical bands Page 30

30 TONALITY Tonality DIN provides an iterative method to detect tones by comparing the levels of each spectral line 1 Tonality Unit (t. u.) has the tonality of a 1 khz sine 60dB Page 31

31 Tonality Calculation Pure tones produce a tonality value of 1.0 Pure random noise produces a tonality value of 0.0 Page 32

32 TONE-TO-NOISE Tone-to-Noise ECMA-74 and ISO 7779 describe the calculation Levels of the prominent discrete tones are compared to the noise level in the same critical band Page 33

33 Critical Bands Page 34

34 PROMINENCE RATIO Prominence Ratio ECMA-74 and ISO 7779 describe the calculation Average SPL of the critical band centered around the tone is higher than surrounding critical bands Page 35

35 Tonal Metrics Curve PR Prominent Prominence Ratio RMS Yes Hz db(a) Pa db(a) Spectrum brand_a (A) Hz Curve PR Prominent Prominence Ratio No Hz Pa db(a) Spectrum brand_a (A) Hz Page

36 Modulations Metriken Hilbert Envelope & Modulation Theory Fluctuation Strength and Roughness

37 MODULATION METRICS Phase shift between signals causes modulation in amplitude these can be often perceived as annoying Sounds which vary in amplitude slowly over time Electric Motor warble Exhaust/Intake Growl Aircraft Turbo Props Cooling fan and engine running at same speed Page 38

38 Modulation Theory 400 Hz 405 Hz What does the sum of a 400 Hz sine wave and 405 Hz sine wave look like? What do you hear? Page 39

39 Modulation Theory 400 Hz 405 Hz Hz Page 40

40 Modulation Theory 400 Hz 405 Hz Hz 5 Modulations per Second Page 41

41 Modulation Theory No 5 Hz in FFT! Only 400 and 405 Hz. Page 42

42 Modulation Theory 0.12 Envelope done by Hilbert Transform Hilbert Transform separates slowly varying envelope from rapidly varying signal V Real :HighPass500:None 5:Envelope_of_HighPass:None 0.61 s 0.65 Page 43

43 Fluctuation Strength and Roughness

44 Roughness and Fluctuation Strength Let s take two sweeping sine tones over 10 secs: 10 Hz to 100 Hz 11 Hz to 110 Hz What is initial modulation frequency? Hz! Hz 1 Hz! What is the end modulation frequency at 10s? 0 0 seconds 10 Page 45

45 Engine Harmonics 600 Hz 6th order Hz 5th order 400 Hz 4th order 300 Hz 3th order! Hz 60 Hz 50 Hz 40 Hz 30 Hz 20 Hz 15 Hz 10 Hz Hz 150 Hz 100 Hz 600 RPM nd order 1st order Page 46

46 ROUGHNESS and FLUCTUATION STRENGTH Fluctuation Strength focuses on slower modulations, between 0 and 20 Hz, max at 4Hz Roughness focuses on faster modulations, between 20 and 300 Hz, max at 70 Hz 1 vacil is fluctuation strength produced by a 1000 Hz tone of 60 db which is 100% amplitude modulated at 4Hz 1 asper is roughness produced by a 1000 Hz tone of 60 db which is 100% amplitude modulated at 70 Hz Page 47

47 Andreas Langmann PreSales Solution Consultant Siemens PLM Simulation & Testing Solutions

48 Thank you