Final Written Study Guide Flashcards
Explain the process of converting An analog electrical signal to a digital signal
Then sound goes to the analog to digital converter to go from an analog electric signal to a digital signal
ADC - when converting an analog signal to a digital signal the signal is sampled at discrete intervals (sampling rate)
Takes snapshots of the analog audio - must have 2 snapshots per wave cycle in order to accurately represent the frequency (one at the peak and one at the trough) - the higher the sample rate the more precise the recreation of the original audio is
The quality of the captured audio sample is determined by
sample rate - how many samples of the original signal are taken in periods of time
bit depth - determines the amount of possible amplitude values that can be recorded for each sample amplitude; The smaller the bit, the more noise you get; The greater the bit depth, the greater the detail in the audio
Quantization - each sampled value is rounded to the nearest value within a set of discrete levels and these levels are defined by the resolution of the ADC
Quantization error = noise floor
results in front-end distortion we see because for every bit is only 6 dB DR
16-bit digital word = DR of 96 dB
Digital signal now has numbers associated w/ it and moves to DSP
Explain the process of converting
A digital signal back to electrical
converts amplified electric signal back to an acoustic signal
armature (flexible strip of metal balanced between two magnets like a diving board) is magnetized as the electrical current flows through the coil and it moves up in the positive direction towards that magnetic and down to the negative magnetic mimicking the electrical sine wave which is attached to the diaphragm above it and as the armature moves so does the diaphragm and this diaphragm movement causes the push and pull of the air creating an acoustic signal
what is sampling rate
number of times per second an analog signal is sampled to create a digital signal - how we capture frequency information
Sampling rate takes regular snapshots of the continuous analog electric signal wave at evenly spaced moments in time & DSP only uses those snapshots (sampled points) and ignores everything else
In order to capture the wave correctly how many shots for each cycle is needed
2
one at the highest point and one at the lowest point
If you do not get enough snapshots, the wave’s frequency will not be able to be accurately recognized
what sampling rate do you want
high rate
more snapshots taken- the more accurate the original continuous wave is represented due to more sampling points
what is the nyquist theorem
Nyquist rate is 2x the given frequency to be measured accurately
minimum sample rate for the highest frequency wanting to be measured
what is bit depth
Measures the amplitude of the signal - horizontal measurement
Higher bit depth = more possible vertical amplitude values & more precisely the exact amplitude of a given sample can be recorded
It also means a wider dynamic range
what is Quantization (Bit Resolution)
Depending on the bit depth, the exact amplitude value is rounded up or down to the nearest value using quantization
Higher quantization = more fluid sample
Lower = more choppy sample
what is quantization error
The difference between the original acoustic signal and the transduced digital signal
Creates noise = noise floor
Every bit only has —-dB of dynamic range
6
16-bit digital word = DR of
96 dB
Noise floor in the hearing aid is noise created in the circuit due to quantization error
TRUE
waht is an algorith
Analytical calculations applied to the digital signal
They add, subtract or multiply strings of digital words
It creates a step by step set of decisions to achieve the desired result
explain how algorithm works
ha is listening to environment and to the acoustic scene and deciding when combo of characteristics happen (spectral, temporal, amplitude) will automatically change programs and improvise the digitized signal processing to match any given listening environment
Explain front end limitations associated with 16-bit processing, and how this impacts microphone sensitivities. How are these limitations resolved?
The smaller the bit, the more noise you get
The greater the bit depth, the greater the detail in the audio
16-bit processing has a dynamic range of 96 dB - it can collect sounds up to this but past this the sound becomes distorted
We get this dynamic range because each digital bit increase the front end dynamic range by 6dB
Solution: the dynamic range shifts by lifting the 96 dB higher to collect more sound sounds but it does this by sacrificing soft sounds
what is an auditory filter, its function, effect of LF masking on a damaged cochlea and frequencies impacted by noise
Cochlea is a series of overlapping band pass filters (frequencies that are grouped together because they are close together on the cochela) that allow certain regions on it to stimulate to a specific frequency region while ignoring frequencies outside of the band
The filters have a lot of overlap with each other so HF bands pick up LF signals from adjacent critical bands and as a result, noise can mask signals from adjacent critical bands
In normal hearing, sharp tuning curves allow for precise frequency discrimination & perception of sounds
w/ HL, the curve is broader and noise can easily affect perception of the desired signal
Broadening of the filters is mainly on the LF side, leading to the increasing of LF masking. So the LF noise masks the region it normally would but also spills over to the overlapping HF bands around it making it more difficult for the PT to use the cues to understand the speech over the noise
how many bands are in the as
25 bands
LF bandwidths are narrow - only 160 Hz wide
HF bandwidths are wide- up to 2500 hz
what is upward spread of masking
Intense 250 hz LF noise will mask that frequency region but the masking will also spill over to the overlapping HF critical bands
Noise energy peaks around 250 Hz but upward spread of masking impacts audibility up to about 1500 Hz
Types of Noise that Impact Intelligibility
Steady state signals
Random noise with an intensity frequency spectrum like speech (speech like sounds)
10-talker babble
4 talker babble
2 talker babble
Which is harder to hear in?
Room reverberation
what are the methods of sound cleaning technology
spatial domain
temporal domain
spectral
Differentiate modulation rate and depth for speech and noise.
Speech & noise signals have time differences
Modulation rates
Speed of the signal
Speech = slow rate
Noise = fast rate
Modulation depth
Amplitude variations bw loudest and softest portions of the signal
Intensity of the variations
Speech = highly variable
Noise = steady over time
How is poor SNR determined in a hearing aid?
it looks at the mod rate and depth
for noise, mod rate is slow and mod depth is steady over time so it takes this and eliminates it from the signal
how does digital noise reduction work
Steady state noise
Idling engine, hair dryer, vacuum etc.
Only acts on fast mod rates & low mod depths
Varying degrees can be applied to each frequency range
doesn’t improve speech intelligibility
Can improve listening comfort, reduce listening effort, reduce cognitive load
does DNR improve speech intelligibility
no
doesn’t improve speech intelligibility
Can improve listening comfort, reduce listening effort, reduce cognitive load
Methods of sound cleaning technology used in the spectral domain
HA here looks at a signal’s frequency to control the signal
Theory is that If noise is below 1.5 and then understanding speech comes from mid and high frequencies reducing the output in low frequencies will improve speech intelligibility in noise
if we attenuate and reduce lf amp and leave in hf, then we will probably improve speech intelligibility
issue: there is noise still in HF not just all in the LF
Methods of sound cleaning technology used in the spatial domain
noise here is managed by directional mic technology
by turning in the direction of the signal the ha provides spacial separation from the speech and noise
adaptive vs fixed
beamforming
superior when only a few noise sources are present
adaptive
superior in the presence of multiple noise sources
fixed
what is automatic mic switching
ha automatically listen to seen and when they think there is enough noise in the room, switch from omnidirectional to directional
Will switch to a fixed direction (one polar plot) or an adaptive directional mic (multiple polar plots)
purpose of automatic mic switching
Switching occurs automatically and continuously
Microphones revert to omni-directional mode in quiet
Feature has an override switch
Polar plots allow reverse directionality to supply a side or backwards zoom
sometimes you want to hear what is behind you and not what is in front
driving in the car
sitting with group in class or in movie theatre and friends are behind you
what is adaptive directional mic
turns directional mics on and off when needed
Directionality of a mic sensitivity is determined by external and internal time delay
Here time delay is digitally manipulated to shift the azimuth of a null based on sounds location
Automatically changes the null as locations of undesired signals moves behind the listener - null steering
Can steer across a BB frequency range or in narrower frequency ranges
what are broadband and multiband mics
Broadband and multiband mics varies null point for different frequency ranges (bunch of polar plots stacked on top of each other
what is beamforming
Has a very narrow beamwidth (only azimuths of + 25°, + 35°, + 50°
Monitors overall intensity of environment
When signal is <55 SPL widest beamwidth activates
As environmental intensity increases (>75 SPL) smallest beamwidth is activated
ONLY thing we can do in digital ha tech to improve speech intelligibility is to
enable directional mics
Explain Weiner filter function and limitations
Spectral subtraction approach, measures short term noise spectrum during gaps in speech
Improves acceptance of background noise but doesn’t improve intelligibility
Worlds good on steady noise but not fluctuating so it is not effective in real time noisy situations
Reads between the lines in desired signal and takes out background noise found in the modulations of speech
what are 3 methods to reduce external feedback
Reduce external feedback loop
Digital notch filtering
Digital feedback cancellation
explain Reduce external feedback loop
Increase snugness of mold to reduce size of slit leaks
Or decrease vent size to stop feedback path
Limitation: both increases OE
explain digital notch filtering
Removes frequencies around the noise - reduces gain around 2-4 kHz where feedback occurs
HA between this range with notch creates a notch in frequency response so we don’t amplify sound in those regions -
reduction in gain from 2-4 kHz and if you don’t turn the volume up it won’t cause feedback
Limitation 35% of intelligibility comes from this range alone so you stop feedback but stopped audibility of important speech sounds so reduced speech intelligibility
explain digital feedback cancellation
When HA detects feedback (identified due to steady state noise bw 2-4 kHz) an algorithm creates an out of phase clone of the signal (duplicate of the feedback) and this causes the clone to be subtracted from the amp path and in turn attenuates the feedback
Describe the limitations of each feedback reduction method (DFS)
Algorithms can cause brief feedback until the out of phase tone can activate
Any sustained tone can start an algorithm (whistle, violin, etc.)
Audible beep when an external signal is mistaken for feedback - entrainment
Feedback cancellation can distort or attenuate parts of speech
Faster battery drainage and life
what are the 3 types of frequency lower
linear frequency transposition
nonlinear frequency compression
spectral envelope warping
what is frequency lowering
Only works from high to low
Best for steeply sloping HF HL
tries to improve HF audibility by shifting it down to LF
do we turn FL on for adults initially
measure signal audibility, turn feature on later
use auditory closure skills so it is less imperative we turn it on
do we turn frequency lowering on for children initially
they are learning language and audibility of s and z is important for them, as soon as it is identified it is turned on
explain linear frequency transpostion
CUT AND PASTE into LF, takes the highs and shoves it into the lows
Improves HF audibility by moving HF band one octave down to LF region
explain nonlinear frequency compression
piano keys
HF range is compressed into a LF range; squishes it down into the audible region
maintain tonotopic order more, lowered all the frequencies, squished into lower frequency space, close to each frequencies close to their original spot
explain spectral envelope warping
Leave the HF where it is but also COPY and PASTE into LF
Keeping a portion in HF but also transposing portion down to LF range
*oticon colors
how does digital wind noise reduction work
Wind turbulence only affects one diaphragm so the HA talk to each ither and take the signal on the opposite side and transmit it and overlay it so it seems like the noise went away and improves snr in that environment
LF filtering can be applied - energy of wind is around 300 hz
Describe the uses and benefits associated with wireless binaural processing technology
HA’s talk to each other
Wind noise management - how does this work
Dual phone - ha streams call into opposite ear to get binaural herring
Volume control
Program
how does binaural help with ILD
How does it support improved awareness of ILD in HA’s
As it arrives to second ear it is lower and wdr adds more gain on that same side so the ILD are gone so you cannot hear well in noise so binaural wireless the one ha tells that ha to not add as much wdrc
One ear will tell the other ear to not add as much wdrc so it doesn’t lose ILD
what is digital data logging
Helps with counseling topics & ways to customize them to meet their lifestyle needs
Which is better for high noise environments? fixed or adaptive? why?
Fixed is better for more noises because it attenuates all the signals
Adaptive is better for a few noise sources because it is not smart enough to determine which ones to attenuate; it doesn’t know how to function
The more BB the noise the greater the effect on speech
works on steady state noise, improves listening effort, working memory, reduces cognitive load but doesn’t improve speech intelligibility
Can improve listening comfort, reduce listening effort, reduce cognitive load
digital noise reduction (temp domain)
Most noise is <1500 Hz, if we reduce output here it will reduce background noise but it assumes all competing signals are in the LFs and this is not true
Background noise is in HF too
spectral domain
what is the test box for
Acoustic chamber
Reduces reflections
Low ambient room noise
Calibrated sound sourccw
what are the speakers for
Emit measurement signal
what is the refrence mic
Calibrates SPL output from speakers
waht is the coupler mic
Measurement mic collecting output from the HA
Simulates size of the canal
Custom products - ITE & ITC
Uses fun tak to attach HA to it
ha 1
Traditional BTE w/ earhook
Attach directly to hook
ha 2
