auditory object and speech recognition Flashcards
Explain the problem of scene segregation
spectral cues
(pitch and timbre)
and temporal cues
(timings)
this is what allows us to distinguish between object and background
what are the different ways in which object recognition is used daily?
recognising an object, whether its an action footstep, breaking glass, animal vocalisation or words
viewpoint invariance such that you are able to distinguish between men, women , and children
semantic aspect
What is the main difference between auditory and visual systems?
that auditory systems are more complex as there is an inherent time dimension, whereas visual ones don’t.
They both rely on waves, whether they are light or sound waves.
in visual system, light waves are so high, it happen instantly, but in auditory system, due to the inherent time dimension, they are not as fast as light waves, such that if a sound was at 1 Hz, we most likely could not hear it, as the lowest frequency we can hear is around 20 Hz with a period of 50 milliseconds.
what are cues for segmentation? (temporal/ spectral cues)
spectral cues
timbre- the pattern of the harmonics
pitch
temporal cues:
onset time
Describe acoustic cues that can be used in auditory scene segregation
timbre
pitch
timing
how many primary auditory fields were identified in humans vs macaque monkeys
2 in humans and 3 in monkeys
how are vowel sounds distinguished?
the spectral envelope as they are different.
the formants, where a certain harmonic’s power is relatively enhanced compared to others. This acoustic structure of certain vowels allows them to be distinct and distinguishable.
explain categorical perception of constant vowel syllables ‘ba’ ‘ga’ ‘da’
they will always be in one category or another, even if you change frequencies and cannot produce sounds in between, They’ll always be perceived in one category or another
Discuss the neural correlates of categorical perception
that when looking at the spatial organisation of neurons responding to ‘ba’ ‘ga’ ‘da’, they look different suggesting categorical perception (shown at a specific snapshot of time-doesn’t work well if you don’t have the right time window)