why do We need to locate the sources of sound?
- location may be important info in itself
- may orient visual attention
- Heads need to be approx 30 cm apport to distinguish them
-spatial resolution is remarkable
coordinates of spatial hearing
1) Horizontal plane - left, right, front + back
- changing azimuth
2) Median plane - above, below
-changing elevation
Minimum audible angle (MAA)
- soundproof lab
= smallest distance between 2 objects we can distinguish
-Play a reference sound + follow it with another sound slightly to left or right
-reference sound = first sound - for low frequency sounds we are really good at distinguishing location (smaller MAA)
- when reference tone is not straight ahead, ability to distinguish is worse
How do we know where the sound is coming from
1) Interaural timing difference (ITD)
- when source of sound is on left, sound reaches left ear before right
2) Generates an Interaural Level Difference (ILD)
- Head shadow
- when source of sound is on left, head shields right ear from sound
- use them as cues to locations
ITD varying with azimuth
- How long is it going to take for the sound to reach your left ear after your right
- Average delay is 0.65 of a millisecond
- ITD decreases as you get closer to Azimuth of 0 or 180 front or back)
- Infants ITD is smaller because distance between ears is smaller
smallest ITD enabling listener to distinguish location
- play a sound at 0 ITD followed by a second tone that varies in ITD
- Adjust second tone so you can just detect whether it is left or right ear
- ITD to distinguish = 10 microsec = 1 degree from straight head
- Auditory system has exquisite sensitivity to interaural timing differences
tow does ILD vary with Azimuth
- ILD is substantial when the sound is high frequency
- when the frequency is high the level of difference is massive (good thing)
- At lower frequencies there is no intensity differences between 2 ears
-low frequency sound doesn’t have head shadow because low frequency waves wrap around the obstacle + aren’t disrupted
What is the smallest ILD distinguishable from OdB
- start with tone from straight ahead (0)
-second tone has small ILD which is adjusted until recognise the second ton e has moved - smallest ILD=1dB
-sensitivity is good across a vide range of frequencies - ILDs are only usable at high frequencies
lTDs are unambiguous at low frequencies
- Interrupted by frequency
- only work at low frequencies
- Duration between 2 peaks
- pairing of similar co-occuring patterns (peaks) principle
- Auditory system detects the blue sound comes first then red = determine location
lTDs are ambiguous at 770 Hz
- auditory system tries to pair similar Co-occuring patterns
-peaks are too close together 50 can’t determine which pattern co-occurs with which - Therefore unable to say location of sand
ITD’s are misleading above 770Hz
- pairing is the complete opposite
- The delay in relation to frequency of sound + pairing of peaks
- Difference between peaks makes it overlap appear the other way round
-tricking the brain into thinking the sound came from the opposite direction - This is based on pure tones
[TD’s are less ambiguous for modulated tones
-when its a complex tone
- follow pairing of similar co-occuring patterns
- similarity in shape of peaks is rescued and auditory system can determine peaks in right order
- peaks that co-occur (temporal proximity) are very dissimilar in shape whereas the 2 true co-occurring leans are similar in amplitude moduIation
- misleading co-occurance vanishes for complex tones
summary of localising pUre tones in azimuth
- ITDs at low Freq ( above 750Hz)
- ILDs At high frequencies (below 1,500Hz)
- gap between 750-1500 work a bit but not fully
- use ILDs for hearing impairments
calculation of ITDS in the brain
-superolive analyses location of sources of sound
-Happens early in ascending auditory system because it relies on very precise timing
measuring ITDs- delay lines + coincidence detectors - intro
- 2 types of neurons: DL +CD
- For a specific frequency have 2 DL one from left, one from right
- in between them have CD that only fire it they are triggered by 2 inputs simultaneously
- how localisation is implemented in barn owls
measuring ITDs- delay lines + coincidence detectors - front sounds
- Triggers same CD
- Input reaches ears at same time = AP travels at same time
- goes through CDs but only activates one in middle - where inputs meet
measuring ITDs- delay lines + coincidence detectors - sound away from front
- input reaches 1 ear before other
- Impulse travels down DL from 1 ear then the other
- sound travels at same speed resulting in triggering of a CD towards the side of the location of the sand
-each CD is specialised for a certain angle - Brain knows which angle the sound came from
opponent -process analysis - measuring ITDs in humans
- Has 2 populations of neurons detecting left or right field
- There are 2 channels, one for the left + one for the right
- system calculates difference between response of the 2 channels to work out where the sound is -establishes I TD
Front-back ambiguity
- sounds coming directly from front or behind generate an ITD of 0
-1) Pinna (if familiar sound) - outer flesh ear
-MAkes its own shadow of different nature from sound coming from front + back - High frequencies from back reflect, low wrap around
-Back sounds more muffled
2) rotating head - favours an ear than the sound coming from front
-Behavioural
Elevation
- Median plane = straight ahead to back
- moving objects up or down doesn’t affect ITD or ILD
-reflection of sound within pinna boost energy or reduce energy depending on frequency + depending on elevation - shape of pinna produces unique resonance patterns
- inserting mould into Pinna disrupts their judgement of elevation
judgement of elevation for modified pinnae experiment
- 16 speakers behind curtain organised in a grid
- Tones played randomly at different locations
- ppts point using lazer beam to location
-pre-test (before ModUIation) = good at it - judgement severely disrupted by modifying pinnae
- Accuracy improves with experience
- No after-effect: accuracy returns to normal once mould is removed
VR + Virtual Acoustic space (VAS)
-How do we recreate a sense of hearing using technology
- ILDS + ITDs provide compelling auditory space but this is limited to horizontal plane
- Head-related transfer functions (HRTF) simulate elevation are reproduction of binaural signal accounting for head-related resonance patterns
-play back what a normal head would hear
- use manikin
-limitation = generic pinnae used
Ben-Hur et al- is using a basic pinna for virtual reality a limitation
- extracted sound from manikin (generic HRTF) VS own (indiv HRTF) + control (ground truth- loud speakers)
- decent performance with HRTFS for elevation + front-back
- indiv HRTF sig better than generic especially f or elevation
-indiv HRTF still not as good as grund truth (room reverberation,)
Localisation when there’s reverberant sound
- sound creates resonance (Bounces of trees, walls)
- get direct + reverberant sound
- Echo suppression- precedence effect
-> direct sound hits ear first
-> echo gives impression it hits another ear
-> creates conflict - our brain cancels echo
- only works within very short window (30m) because its perceived as an echo
