Non-uniform Compression Ratio for Dynamic Range Compression
- 1. LUDWIG KOLLENZ TWO PI GMBH, VIENNA Non-Uniform Compression Ratio as a tool for flexible transformation of dynamic range PAGE 1 FROM 18 Erlangen 2018-02-23
- 2. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Dynamic Range Mapping PAGE 2 FROM 18
- 3. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Effects of Time Constants 0.19 0.195 0.2 0.205 0.21 0.215 0.22 0.225 0.23 -1 -0.5 0 0.5 1 t / s Amplitude 0.19 0.195 0.2 0.205 0.21 0.215 0.22 0.225 0.23 -1 -0.5 0 0.5 1 t / s Amplitude 0.19 0.195 0.2 0.205 0.21 0.215 0.22 0.225 0.23 -1 -0.5 0 0.5 1 t / s Amplitude PAGE 3 FROM 18
- 4. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Uniform Distribution of Compression PAGE 4 FROM 18
- 5. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Uniform Mapping Output and Gain PAGE 5 FROM 18 O(I) = Dynamic Range of Output G(I) = Dynamic Range of Gain
- 6. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Non-Uniform Mapping of Output PAGE 6 FROM 18
- 7. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Non-Uniform Mapping of Gain PAGE 7 FROM 18
- 8. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Signal-Dependent Weighting Function PAGE 8 FROM 18 calculate –>
- 9. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Speech-Level Dependance PAGE 9 FROM 18 calculate –> calculate –> calculate –>
- 10. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Weighting Function Example PAGE 10 FROM 18
- 11. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria BANDS 1 2 3 4 5 6 7 8 9 10 11 12 CR 1,0 1,7 1,8 2,2 2,3 2,3 2,2 1,6 1,4 1,2 1,1 1,1 G50 1 13 21 26 27 28 29 29 24 22 21 20 fc (kHz) 0,25 0,5 0,75 1 1,25 1,65 2,1 2,7 3,5 4,5 5,7 7,1 Attack time: 10ms Release time: 150ms Performance Examples Moderate hearing loss PAGE 11 FROM 18
- 12. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Soft Speech in Quiet Sentence:»Five minutes peace.« 0 0.5 1 1.5 2 2.5 3 x 10 4 −0.2 −0.1 0 0.1 0.2 time / samples / 16kHz 0 0.5 1 1.5 2 2.5 3 x 10 4 −0.2 −0.1 0 0.1 0.2 time / samples / 16 kHz PAGE 12 FROM 18 top = uniform CR bottom = non-uniform CR
- 13. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Soft Speech in Quiet 1000 2000 3000 4000 5000 6000 7000 8000 −80 −75 −70 −65 −60 −55 −50 −45 −40 −35 −30 Freq. / Hz Spectrummagnitude/dB PAGE 13 FROM 18 red = uniform CR blue = non-uniform CR
- 14. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria PAGE 14 FROM 18 Medium Speech in Noise Sentence:»the children were having breakfast.« 0 0.5 1 1.5 2 2.5 3 3.5 4 x 10 4 −0.6 0 0.6 time / samples / 16kHz 0 0.5 1 1.5 2 2.5 3 3.5 4 x 10 4 −0.6 0 0.6 time / samples / 16 kHz top = uniform CR bottom = non-uniform CR
- 15. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria PAGE 15 FROM 18 Medium Speech in Noise 1000 2000 3000 4000 5000 6000 7000 8000 −60 −55 −50 −45 −40 −35 −30 Frequency / Hz Spectrummagnitude/dB red = uniform CR blue = non-uniform CR
- 16. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Conclusion PAGE 16 FROM 18 input speech Level/dB Level/dB output speech frequencyof occurrence frequencyof occurrence red = uniform CR blue = non-uniform CR
- 17. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Thank you! www.two-pi.com PAGE 17 FROM 18
- 18. Erlangen 2018-02-23 Ludwig Kollenz, Vienna/Austria Literature Bisgaard, n.;Vlaming, M.; Dahlquist, M.: »Standard Audiograms for the Iec 60118-15 Measurement Procedure«, in:trends in Amplification 14 2010, p. 113–120. Keidser, G.; Dillon, H.; Flax, M.; ching,t.; Brewer, S.: »the nAL-nL2 Prescription Procedure«, in:Audiological Research 1 2011, p. 88 – 90. Moore, B.c.;Sęk, A.:»Preferred compression Speed for Speech and Music and Its Relationship to Sensitivity to temporal Fine Structure«, in:trends in Hearing 2016 Sep 7. Moore, B.c.: »cochlear hearing loss«, Second edition, John Wiley & Sons 2008. Plomp, R: »the negative effect of amplitude compression in multichannel hearing aids in the light of the modulation transfer function«, in:J Acoust Soc Am. 83 1988, p. 2322-2327. PAGE 18 FROM 18