Digital Signal Processing Flashcards
What is digital signal processing?
It converts an analog electric signal into digital data (numbers)
It can be manipulated and isolated
How is sound converted into digital?
Analog to digital converter
What do analog digital converters require?
Sound sampling and quantization
What is a sampling rate?
Measures equally spaced moments of time along the analog electric signal
Are high sampling rates better?
Yes, they better represent the original signal
What is nyquists frequency?
It quantifies the highest frequency that can be reconstructed into a signal for a particular sampling rate
A sampling rate is 2x higher than the hearing aids bandwidth (a 20 kHz sampling rate becomes a 10 kHz bandwidth)
What is bit depth?
It is the vertical measurement representing the amplitude of the signal
What is quantization?
Combines bit depth and sampling rate
Creates a digitalized version of the signal
What is the bit value of most hearing aids today?
16 bit
What is quantization error?
The difference between the original and the digitized signal
What does quantization error create?
Noise
Soft random noise referred to as the noise floor
What is an algorithm?
The analytical calculations applied to the digital signal
These are applied to acoustic scenes
Can algorithms analyze and classify acoustic scenes to determine which algorithmic signal processing scheme to apply?
Yes
They can identify steady state, transient, or reverberant signal and apply calculations to attenuate them
Can the accuracy of algorithms affect how the user hears?
Yes
What happens to the signal after digital signal processing?
It is converted back to an analog electric signal
What are the two ways to return a digital signal to an analog electric one?
Digital signal is converted to an analog signal and then goes to the receiver
Digital signal is converted and then amplified by an output amplifier (AGC-o) before entering the receiver
What is the processing speed?
The time it takes an acoustic signal to be picked up by the mic, amplified, digitized, processed, back to electric signal, converted to acoustic signal and delivered to the ear canal
What is the average processing speed for modern hearing aids?
About 2 to 10 msec
When are slow processing speeds an issue?
When direct signals are mixed with amplified signals
Sound quality issues are greatest for open domes
Comb filtering effect occurs
What is the solution for direct signals and amplified signals mixing?
Faster processing speeds
0.5 ms preferred
Do normal hearing adults understand speech in poor SNR settings?
Yes because speech is highly redundant
Do auditory filters need to be intact to efficiently process complex signals in the cochlea?
Yes
The cochlea analyzes signals using a bank of overlapping band-pass filters
These filters allow regions of the cochlea to paw attention to a specific freq region while ignoring freq outside the band
How big are the critical bands in the low freq?
Narrow
How big are the critical bands in the high freq?
Wide
Does noise mask signals in adjacent critical bands?
Yes
Filters overlap
Does poor freq resolution alter the shape of the critical bands?
Yes
It broadens the filter mainly on the low freq side of the filter
This increases the susceptibility to low freq masking
How much can an intense 250 Hz noise mask?
Not only masks the 250 Hz region, but masking spills over to overlapping higher frequency bands
Impacts audibility up to 1.5 kHz
What are the three different types of noise that significantly impact speech intelligibility?
Steady state signals
Random noise with an intensity-frequency spectrum like speech
Room reverberation
What three domains are analyzed in hearing aids before applying an algorithm?
Spatial domain (location) - identifies a signals location to control the signal
Temporal domain (time) - identifies a signals timing to control the signal
Spectral domain (frequency) - identifies the signals frequency to control the signal
How is noise in the spatial domain managed?
With directional microphone technology
What is automatic microphone switching algorithms?
A device can automatically change from an omnidirectional to a directional mic in the presence of background noise
When a device switches algorithms based on background noise, what are the two options it can switch to?
Fixed directional (one single polar plot) or adaptive directional (multiple polar plots)