Lecture 6: Signal Processing & Compression

Question 1

What is the difference between analog and digital hearing aids?

Answer 1

Digital Processor

• In digital HAs, this processor converts acoustic signal to 0s and 1s
• Numbers are converted back to sound with an analog to digital converter
• This processor did not exist in analog hearing aids

Question 2

Define hearing aid output.

Answer 2

An overall expression of the overall sound power

• Expressed in dB SPL
• Often referred to as maximum power output (MPO)

Question 3

What is gain?

Answer 3

The hearing aid output - input

Question 4

What is harmonic distortion?

Answer 4

• The amplitude of the distortion produced by the hearing aid
• The source of the distortion is the amplifier

Question 5

What are the digital specifications in hearing aids?

Answer 5

1) Instructions per second (MIPS)
- The # of instructions that a digital processor can complete in a sentence
- The higher the instruction rate is, the higher the current consumption
- Higher instruction rates make it possible for the HA to perform complex operations (includes multiband processing, automatic feedback as supression)

2) Sampling Rate
- How many times per second the HA samples the input signal
- Must be greater than twice the highest frequency component that’s present in a complex signal
- The bandwidth of the signal processing is the limiting factor
- Power requirements can also limit the sampling rate

3) # of bits
- Derived from binary bits
- Greater than the # of bits, the greater precision of representing the signal
- If you have too few bits, the digital approximation is too coarse
- Too sloppy of a signal there is extra noise
- Use coding scheme that makes fewer bits produce as good of sound quality as more bits

4) Current consumption
- Consider battery life and size
- Dependent on the MIPS, the voltage requirements & the complexity of the integrated circuit

5) Processing Delay
- The amount of time that the input signal passes through the HA into the ear canal
- Too much time between input and output processing
- Longer delays makes for more sophisticated processing; makes the listener upset with their own voice quality
- Delay of less than or equal to 5 Ms is preferred

6) Physical size
- Physical size of the integrated circuit is still the limiting factor

7) Open Platform
- A closed platform means that you can’t fundamentally change how the compression works
- Open platform technology allows you to add updates to your hearing aids

Question 6

What does the acronym ACT stand for?

Answer 6

Audibility
Comfort
Tolerance

Question 7

What are the categories of signal processing?

Answer 7

1) Linear
- Referred to as compression
- One-to-one relationship between signal input and signal output
Output = input + gain
- Gain is constand for input up to a certain point

2) Nonlinear
- Both occur in analog and digital hearing aids

Question 8

What are the limitations of linear amplification?

Answer 8

• Harmonic distortion (from peak clipping)
• Frequency volume adjustment or insufficient volume
• The normal cochlea is a nonlinear system

Question 9

What are the consequences of OHC damage?

Answer 9

1) Hearing loss
2) Linear basilar membrane response
3) Loss of gain for low-level inputs
4) Reduced dynamic range
5) Broad auditory filters

Question 10

What are the purposes of compression?

Answer 10

• Limit maximum output

- Amplify sounds to fit within the hearing aid user’s auditory dynamic range

Question 11

Define compression.

Answer 11

The change in input level needed to produce a 1 dB change in output level

Question 12

Compute the compression ratio & draw it on a graph.

Answer 12

• See handout from class*

Question 13

What are the characteristics of compression?

Answer 13

1) Static: how much compression is taking place

Compression threshold
- equals the input level in dB SPL above which gain decreases by 2 dB from linear gain (threshold knee point)

Compression ratio
- indirectly describes how much the gain decreases and is defined as the change in input level needed to produce a 1 dB change in output level

2) Dynamic characteristics: how quickly compression is occurring

• Attack time: the time it takes for the compressor to react to an input signal that exceeds the knee point (less than 5 msec)
• Release time: the time it takes for teh gain to return to the precompressed level (less than 20 msec)
• Attack/release times are considered time constants
• If the time constant is too fast, the listener will hear a pumping sound
• If the time constants are too slow the listener may hear dead spots while the HA is recovering from compression

Question 14

What is the difference between input and output compression?

Answer 14

Input compression
- the level where the compression begins (after the microphone but before the amplifier)

Output compression
- not until the signal leaves the amplifier is the signal compressed

Question 15

How can you calculate compression ratio?

Answer 15

CR = change in input/ change in output

• creates the slope on the compression graph

Question 16

What is output limiting vs. WDRC?

Answer 16

OL does its work above the knee point

• High compression knee point
• High compression ratio

WDRC does it’s work below the knee point

• Low compression knee point
• Low compression ratio
• Keep soft wounds audible
• Keep moderate sounds comfortable
• Keep intense sounds loud but not loud enough to cause discomfort

Question 17

What is BILL?

Answer 17

Bass increase at low levels

- As input increases, the gain at the bass will decrease

Question 18

What is TILL?

Answer 18

Treble increases at low levels

- As input increases, the gain at higher frequencies will decrease

Question 19

What is PILL?

Answer 19

Programmable increases at low levels

Question 20

What are the different types of compression?

Answer 20

1) Multichannel compression
- Variable across frequency ( to accommodate hearing loss) and variable dependent upon signal level

2) Syllabic compression
- Uses a fast-acting compressor to increase gain for weak syllables or phonemes, and to decrease gain for loud syllables or phonemes

Question 21

What is loudness normalization?

Answer 21

• The goal is to make loudness seem the same for a person with hearing loss as it would be for a normal hearing person
• Compression ratio decreases as input level increases
• Different across frequencies
• Low compression threshold

Question 22

What is loudness equalization?

Answer 22

• To maximize speech intelligibility

- The goal is to make all frequency bands contribute equally to the volume of speech

Question 23

Define attack time.

Answer 23

How quickly the compression circuit responds to changes in input

Question 24

Define release time.

Answer 24

How quickly the compression circuit returns to “normal”

Question 25

What are the compression strategies?

Answer 25

1) Syllabic 2) Adaptive Dynamic Range Optimization (ADRO) - applies multiple time constants and rules in sequence - Gain is reduced to avoid exceeding loudness discomfort levels - Gain is reduced to avoid upward levels of speech exceeding comfortable range - Gain is increased to avoid softer levels of speech being inaudible - Gain is never allowed to exceed levels that would cause feedback or excessive amplification of background noise 3) WARP Processing - Takes into account the cochlea's non-linear processing - Uses fast syllabic attack and release times

Question 26

How can we optimize dynamic range and restore normal loudness perception?

Answer 26

- Use AGI-1 - Keep TK as low as possible to make weak sounds audible - Keep CRs low - AT/RT may be fast or slow, dependent upton patient - Multichannel compression better for sloping audiometric configuration

Question 27

What is expansion?

Answer 27

- A noise reduction strategy for quiet environments - Minimizes soft environmental noises ( i.e., the sound of the refrigerator, the furnace, etc.) - Minimizes the sound from the circuit noise - Applied to inputs softer than the knee point (TK)

Question 28

What is a single channel?

Answer 28

- Compression characteristics that are applied to the entire frequency response - Fine for relatively flat hearing loss

Question 29

What is a multichannel?

Answer 29

- The difference between bands and channels

Question 30

What's the difference between bands and channels?

Answer 30

Band: Section where gain can be controlled (frequency) Channel: Section where compression can be controlled (processing) - There can never be more channels than bands

Question 31

What is AGCo vs. AGCi?

Answer 31

AGCo - Most important for managing maximum output - As you increase the volume control settings, the gain increases, the output increases, but the MPO remains the same AGCi - Important for speech perception - As you increase the volume control settings, the gain increases, the output increases, and the MPO increases