Alaphangmagassag (virtual pitch) Terhardt (1972-74): megkulombozendo “virtual pitch” es

1. Alaphangmagassag (virtual pitch) Terhardt (1972-74): megkulombozendo “virtual pitch” es “spectral pitch” dimenziok Virtual pitch: valoszinuleg (=biztos) idoelemzesbol adodik Spectral pitch: hangkepelemzes Ket eszleles: hangmagassag es hangszin

6. Alaphangmagassag erzekelese:idobe telik? (peldak) “Virtual pitch” hatara kb 1 kHz Dominans frekvenciaterulet (<1500 Hz) es felhangrendszamok (2-6) (Ritsma, 1962) Autokorrelacio reszlegesen megmagyarazza (Matlab peldak)

7. E Λ

8. Szurt zajnak autokorrelacioja alacsony, feher zajnak nulla

9. Feloldott es feloldatlan komponensek: Harmonikus felhangok linearisan (Hz) kovetik egymast, Mig a ful tonotopiai rendszere Greenwood egyenletet koveti

10. Basilar membrane (=alaphàrtya?) BM_mm = (16,7) log10 ((0,006046) freq + 1) (Greenwood egyenlőség)

11. Theorem proof obvious but no room to give it here...

12. Time-domain processing • output of cochlea temporally structured • neural circuitry specialized for time

13. Licklider

14. Licklider from cochlea

15. Licklider

16. Auditory Tuning Curve

17. period --> pitch

18. Jeffress similar model, based on binaural interaction Licklider, Jeffress: excitatory interaction

21. de Cheveigné Harmonic cancellation: inhibitory interaction

22. period --> pitch

23. medial superior olive (MSO)

24. Lateral superior olive (LSO)

25. Principle of MSO "coincidence counter" neuron Activated if impulses are simultaneous at input

26. Right ear Left ear Delay line Model by Jeffress (1948)

27. Principle of LSO "anti-coincidence counter" neuron activated except if impulses are simultaneous at input

28. Durlach model similar to Jeffress’s, based on binaural interaction (Equalization-Cancellation)

29. 2 types of model • Correlation (auto- & cross-) • (excitatory interaction) • Cancellation • (inhibitory interaction)

30. 2 types of model • Correlation (auto- & cross-) • (excitatory interaction) • Cancellation • (inhibitory interaction)

31. Basic ingredients running autocorrelation

32. Basic ingredients running autocorrelation

33. Basic ingredients running autocorrelation

34. Basic ingredients running autocorrelation

35. Basic ingredients running autocorrelation

36. Basic ingredients running autocorrelation

37. Basic ingredients running autocorrelation left

38. Basic ingredients running autocorrelation left right

39. Basic ingredients running autocorrelation left right running crosscorrelation

40. Structure 1 2 3 fast signal processing

41. 1. Licklider model of pitch E E

42. 1. Licklider model of pitch • Module 1: calculate autocorrelation & • crosscorrelation for all t, t, q • Module 2: select autocorrelation with • parameter t • Module 3: vary t while monitoring output • of 2 for a maximum

43. 2. Jeffress model of localization • Module 2: select crosscorrelation with • parameter q • Module 3: vary q while monitoring output • of 2 for a maximum

44. 4. Cancellation model of pitch

45. 4. Cancellation model of pitch

46. 4. Cancellation model of pitch autocorrelation terms

47. 4. Cancellation model of pitch • Module 2: linear combination of • autocorrelation terms (parameter t) • Module 3: vary t while monitoring output • of 2 for a minimum

48. 5. Equalization-Cancellation (Durlach)

49. 5. Equalization-Cancellation (Durlach)

50. 5. Equalization-Cancellation (Durlach) autocorrelation & crosscorrelation terms