1. Central Auditory Processing
Part II: Lent Term 2014: ( 2 of 4)
Central Auditory Processing
Roy Patterson
Centre for the Neural Basis of Hearing
Department of Physiology, Development and Neuroscience
University of Cambridge   
Lecture slides on CamTools
Lecture slides, sounds and background papers on

2. The Overture
Act I: the information in communication sounds (animal calls, speech, musical notes)
Act II: the perception of communication sounds and the robustness of perception to changes in acoustic scale
Act III: the processing of communication sounds in the
auditory system (signal processing)
Act IV: the processing of communication sounds
(anatomy, physiology, brain imaging)

3. Auditory perception is robust to changes in Ss and Sf
Decreasing VTL
Increasing GPR
(1/Sf )
( Ss )
Kawahara and Irino (2004). Principles of speech manipulation system STRAIGHT. In Speech separation by humans and machines, P. Divenyi (Ed.), Kluwer Academic,

4. Rana catesbeiana
(1/Sf )
Decreasing VTL
Increasing GPR
( Ss )

5. [Patterson, Smith, van Dinther and Walters (2008)]
Low
High
Pitch
Long
Short
VTL
( Ss )
(Sf )
Time
Time

6. Spectra on a linear frequency axis
Low
High
Pitch
Long
Short
VTL
( Ss )
(Sf )

7. Recognition of Scaled Vowels
Smith, Patterson, Turner, Kawahara and Irino JASA (2005)
pdn.cam.ac.uk/groups/cnbh/teaching/lectures/SPTKIjasa05.pdf
/u/
mean

8. waveform and spectrum of a child’s /a/Sf Ss
Frequency on a logarithmic axis (octaves)

9. Psychophysical experiments confirm:
The Perception of Communication Sounds: Summary
Psychophysical experiments confirm: I:
Humans can extract the content of the communication without being confused by size differences

10. Speaker Size estimates for vowels varying in GPR and VTL
pdn.cam.ac.uk/groups/cnbh/teaching/lectures/SPjasa05.pdf
Smith and Patterson (2005) JASA
log(VTL)
log(GPR)

11. Psychophysical experiments also confirm:
The Perception of Communication Sounds: Summary
Psychophysical experiments also confirm: I:
Humans can extract the content of the communication without being confused by size differences
II:
Humans can extract the size information without being confused by differences in the content

12. Discrimination of Ss and Sf
Ss : the semitones on the keyboard differ by 5.9%
Experiments with vowels show that people can
reliably discriminate a 2% difference in Ss
Sf : the discrimination of Sf is more tricky
Present two vowels and ask:
Which vowel came from the larger speaker?

13. waveform and spectrum of a child’s /a/Sf Ss
Frequency on a logarithmic axis (octaves)

14. Ives, Smith and Patterson (2005) JASA
Syllable database
Sonorants
Stops
Fricatives
ma na la ra wa ya ba da ga pa ta ka sa fa va za xa ha
me ne le re we ye be de ge pe te ke se fe ve ze xe he
mi ni li ri wi yi bi di gi pi ti ki si fi vi zi xi hi
mo no lo ro wo yo bo do go po to ko so fo vo zo xo ho
mu nu lu ru wu yu bu du gu pu tu ku su fu vu zu xu hu
am an al ar aw ay ab ad ag ap at ak as af av az ax ah
em en el er ew ey eb ed eg ep et ek es ef ev ez ex eh
im in il ir iw iy ib id ig ip it ik is if iv iz ix ih
om on ol or ow oy ob od og op ot ok os of ov oz ox oh
um un ul ur uw uy ub ud ug up ut uk us uf uv uz ux uh
aa ee ii oo uu
CV’s
VC’s
Ives, Smith and Patterson (2005) JASA
vowels
large (voiced)
small (voiced)
mi en ka it so us
Kawahara and Irino (2004). The vocoder STRAIGHT. Kluwer Academic

15. Ives, Smith and Patterson (2005) JASA
Speaker-size discrimination with syllables
Present two intervals of syllables and ask: Which is the larger speaker?
The syllables are randomly chosen for the intervals.
The overall level is varied randomly between the intervals.
The pitch contours are different.
VTL = x
VTL = x + Δx
The only consistent cue is a difference in VTL ( Sf )
interval 1
interval 2
pitch
/wa/
/et/
/am/
/ku/
/ma/
/te/
/om/
/se/
Ives, Smith and Patterson (2005) JASA

16. Experiment:
Sf discrimination thresholds for five different people
Glottal pulse rate / Hz 80
160
320
SER
0.92
1.22
1.65
SMALL MALE
LARGE MALE
SMALL CHILD
DWARF
CASTRATO
Ives, Smith and Patterson, JASA (2005)
( Ss )
VTL/cm 14 10 19 Sf

17. Results: all subjects, all syllables
DWARF
SMALL CHILD
% reported larger
% reported larger
Trials test as smaller
Trials test as smaller
SMALL MALE
Ives, Smith and Patterson (2005) JASA
% reported larger
Trials test as smaller
LARGE MALE
CASTRATO
% reported larger
% reported larger
Trials test as smaller
Trials test as smaller

18. Results: all subjects, all syllables
DWARF
SMALL CHILD
average JND across syllable category for specific speaker type.
SMALL MALE
grand average JND for the experiment
basically 5%, independent of acoustic scale
CASTRATO
LARGE MALE

19. Psychophysical experiments confirm:
The Perception of Communication Sounds: Summary
The acoustic scale values in communication sounds tell us which individual, within a population, is speaking or which instrument, within a family, is playing
Psychophysical experiments confirm: I:
Humans can extract the content of the communication without being confused by the size information
II:
Humans can extract the size information without being confused by the content of the communication
III:
Auditory perception is amazingly robust to changes in acoustic scale (Ss and/or Sf ) in communication sounds

20. End of Act II
Thank you
Smith, D. R. R., Patterson, R. D., Turner, R., Kawahara, H., and Irino, T. (2005). "The processing and perception of size information in speech sounds," J. Acoust. Soc. Am.117, 
Smith, D. R. R. and Patterson, R. D. (2005). "The interaction of glottal-pulse rate and vocal-tract length in judgements of speaker size, sex and age," J. Acoust. Soc. Am. 118,
Ives, D. T., Smith, D. R. R. and Patterson, R. D. (2005). "Discrimination of speaker size from syllable phrases," J. Acoust. Soc. Am. 118 (6),

21. Concurrent Speech and the cocktail party
Colin Cherry (1952)

22. Syllable database
Sonorants
Stops
Fricatives
ma na la ra wa ya ba da ga pa ta ka sa fa va za xa ha
me ne le re we ye be de ge pe te ke se fe ve ze xe he
mi ni li ri wi yi bi di gi pi ti ki si fi vi zi xi hi
mo no lo ro wo yo bo do go po to ko so fo vo zo xo ho
mu nu lu ru wu yu bu du gu pu tu ku su fu vu zu xu hu
am an al ar aw ay ab ad ag ap at ak as af av az ax ah
em en el er ew ey eb ed eg ep et ek es ef ev ez ex eh
im in il ir iw iy ib id ig ip it ik is if iv iz ix ih
om on ol or ow oy ob od og op ot ok os of ov oz ox oh
um un ul ur uw uy ub ud ug up ut uk us uf uv uz ux uh
aa ee ii oo uu
CV’s
VC’s
vowels

24. concurrent-speech: experimental paradigm
Identify the syllable in the interval that stays lit wu
osh
Vestergaard et al (2009) JASA

25. Robustness of speech perception

26. 600
400
200
Target
Distracter ms

27. Concurrent-speech paradigm
Sonorants (semivowels)
Stops (plosives)
Fricatives
600
400
200
Target
Distracter ms de mi ki lu ez
osh
Target triplet: de mi osh
Masker triplet: ki lu ez
Concurrently at 0 dB SNR
Pre-cursor, 0 dB SNR
Vestergaard et al (2009) JASA

28. Vestergaard, Fyson and Patterson, JASA, 2009Ss Sf
Vestergaard, Fyson and Patterson, JASA, 2009