5 - Digital Technology Flashcards
What are three types of HAs?
1) analog
2) programmable (digitally controlled analog)
3) digital
What are 3 components of analog HAs?
1) The electrical voltage is analogous to the acoustic signal
2) Analog HA are now obsolete
3) Processing was 1:1 up to the output limit of the hearing aid
What are 2 components of programmable HAs?
1) Programming is performed through digital technology but signal processing remains analog
2) Also obsolete
What are 2 components of digital HAs?
1) Digital processing of signals
2) Classified as entry, intermediate, and advanced levels depending on features available.
How does processing happen in analog HAs?
Mic—amplifier—receiver
How does processing happen in digital HAs?
Mic—preamp—LP filter—digital conversion—digital signal processing—analog conversion—LP filter—receiver
How does a HA process sound?
- Sound: condensations and rarefactions of air (particles moving back and forth over time)
- HAs break up this continuous information, break it down, store it, and put it back together
- HAs are recording this, and reproduce after the signal has been manipulated
What are the 4 steps of how digital HAs process sound?
1) Mic picks up signal and converts it into electrical energy
2) Analog to Digital Conversion (ADC or A/D)
3) Digitized numerical codes are assigned binary digits (bits) or “0” and “1”.
- Signal processing occurs through manipulation of these digits
4) Digital to Analog Conversion (DAC or D/A)
How can a HA represent sound well?
- There has to be enough samples to accurately represent the information
- Specifically, we must consider the NYQUIST theorem.
- We need at least double the sampling points as the highest frequency we want to represent.
- The Nyquist FREQUENCY is half the sampling rate.
What is the aliasing error?
When the input frequency is greater than the Nyquist frequency, a “new” sine wave, not present in the original signal, is created
- aliasing signal
- distortion
What is a visual example of the aliasing error?
Visual example: car wheels that appear to turn backwards in movies
- Camera samples the movement at specific times
- The wheel turns faster than the sampling frequency
How are HAs understanding sound?
HAs understand sounds as discrete points of information
What if there aren’t enough sound points to represent the sine wave?
- If not enough points to represent the sine wave, it will fill in the blanks
- If we don’t have a good signal understanding on the input, the output cannot be beneficial and accurate
Analog to digital conversion - how do you avoid aliasing in HAs?
- Problem of aliasing if input signal contains frequencies above the Nyquist frequency.
- Low-pass filter (anti-aliasing filter) rejects the frequencies higher than the Nyquist prior to A/D conversion.
Where does the anti-aliasing filter happen in digital HAs?
Anti-aliasing happens before the digital conversion of the signal
What are BITS?
- Not using a “standard” 10 base number system
- HAs are using Binary digits (BITS) 1 or 0
- It’s as easy (or as complex) as 0 1, 1 0, 1 1
- Basically, we can represent EXPONENTIAL AMOUNTS OF INFO!
Why do HAs want more BITS?
More BITS, smaller error, more dynamic range!
How many BITS does a HA generally use?
- HAs generally use 12 to 24 bit processing—-16 bits usually… or higher
- 16 bit: can represent a wide range of frequencies (speech frequencies and a good representation of DR)
Each BIT has a ____dB dynamic range
6 dB
16-bit A/D converter has a dynamic range of ____dB.
96 dB
What do HAs need a lot of processing for?
Need a lot of processing to represent music well in a HA
What happens to very loud inputs in a HA?
- Inputs that exceed the upper limit of the A/D converter’s dynamic range will be peak clipped, or HA uses compression to reduce input level before processing.
- A distorted signal at the front-end cannot be improved at a later stage.
What happens when speech is in very loud noisy backgrounds?
- Both speech and noise are distorted!
- We need the processing on the front end so the HA can manage delivery and noise
What are the 3 steps of digital to analog conversion?
1) Numerical format is converted back into an analog signal.
2) Through this process, some unwanted noise (images) is created.
3) An anti-imaging filter (low pass filter) rejects the noise.
Where does the anti-imaging filter happen in digital HAs?
An anti-imaging filter is on the back end (the anti-aliasing is on the front end
What is the processing delay?
Digital processing takes time: analog sound is sampled, converted to digital info which is then manipulated, and then re-converted into analog signal.
How long does the processing delay take?
1) Complex processing takes longer.
2) Delay is about 5-6 ms in many HA.
3) Ideally 1 or 2 ms (the shorter the better).
4) Problematic if 10-15 ms or longer (this is when patients perceive sound quality issues).
Processing delay (echo effect) may be perceived by people with what 3 things?
1) Normal hearing in one ear and a unilateral fit on the poor ear.
2) Open fittings: sound also enters the ear naturally.
3) Bilateral fittings: processing delay may cause the lip movements to be out of sync with the signal heard.
When trying to match targets, what 2 problems can occur if your test input is not speech or speech-like?
1) Digital noise reduction system can interpret the test signal as noise and reduce the gain.
2) Feedback cancellation feature of HA may think that test signal is feedback.
What 3 ways do we mitigate the problems that can occur if your test input isn’t speech or speech-like?
- Some manufacturers have a “verification setting” in the software that turns off noise reduction features for REMs
- We use speech input (e.g., “carrot passage”, or the ISTS signal)
- If you only have access to a sweep or composite noise test input, deactivate the automatic features (e.g., noise reduction) before real- ear verification of targets.
How do entry level HAs differ from high end HAs?
- Entry and high-end HAs have the same processor
- Compared to entry level HAs, high-end HAs generally have:
- More activated features
- In entry level HAs, several of the features are disabled within the firmware (not visible to the audiologist)
- More bands/channels for frequency shaping and for signal processing (e.g., noise reduction algorithm)
- More automatic programs
What are the 3 values of having more bands or channels?
1) Greater fitting precision for frequency shaping (gain and MPO)
2) Better control of compression parameters
3) Better signal processing performance (e.g., noise reduction, feedback management)
Is more channels better, e.g., is a 20-channel HA better than an 8-channel HA?
- Manufacturers, and many audiologists, seem to believe that more is better
- However, with at least 6 channels of processing, your patient should do well with the HA
Who does the number of channels effect?
- Even if the patients select the most premium devices, if we don’t verify targets and meet their audibility, we aren’t providing them with effects of the premium HAs
- The number of channels make a difference for us as audiologists, not the patients
What 3 things does having more channels mean (cons)?
1) Increase in processing load and group delay
2) Increased battery consumption
3) Higher HA cost
How many channels does an entry level and premium level HA have?
- Entry level = 6 channels
- Premium level = 20-25 channels
Is advanced level better than entry level?
- Results showed no difference between technology levels for improvements in speech perception and quality of life.
- Participants did not report better outcomes with advanced technology HAs for any of the measures
What did the study find that advanced HAs did better than entry level?
Entry and advanced HA had similar performance except that advanced level HA performed better for localization of HF stimuli in the quiet condition.
Is newly launched technology really better?
- Usually lots of hype when a new product or feature is introduced.
- New products are typically launched by the manufacturer at a major conference (e.g., CAA or AAA conference).
- Manufacturer may have “evidence” to support the new feature, e.g., “in clinical trials, X number of users preferred this new technology”.
- How strong is this evidence? Is it real evidence or is it the placebo effect? Yes (only 2% preferred)
