Welcome to Hearing Aid Compression, Digital Microphones & Noise Reduction Presenter: Ted Venema Audiologist, Speaker, Author IHS Organizers: Ted Annis Carrie Pedersen Senior Marketing Specialist Member Services Supervisor
Housekeeping • This presentation is being recorded • CE credit is available! Visit ihsinfo.org for details • Note taking handouts are available at ihsinfo.org on the webinar page. Feel free to download now!
Agenda • Linear Gain • Compression in Analog Hearing Aids • Compression for Severe-Profound HL • Compression for Mild-Moderate SNHL • Compression in Today’s Digital Hearing Aids • Q&A (enter your questions in the Question Box any time)
HEARING AID COMPRESSION: IT’S A GAIN ISSUE NOT A SIGNAL-TO-NOISE RATIO THING
Hearing Aids Must Do Two Things: Provide gain for the HL • Increase signal-to-noise ratio (SNR) • Compression is a gain issue: SNR is a Dmic & DNR issue • To be covered in another webinar •
Fitting the Eye Versus Fitting the Ear Intact Retina Damaged Hair Cells = Fig 3-1, Venema, T. Compression for Clinicians 2 nd edition, Cengage 2006
Normal Inner and Outer Hair Cells Fig 1-7 Venema, T. Compression for Clinicians 2 nd edition, Cengage 2006
Damaged Hair Cells (mostly outer) Fig 1-8 Venema, T. Compression for Clinicians 2 nd edition, Cengage 2006
Here’s a “Passive” Traveling Wave A wave without outer hair cells Base Apex Basilar Membrane Fig 1-4, Venema, T. Compression for Clinicians 2 nd edition, Cengage 2006
Outer Hair Cells Amplify and Sharpen the Peak! They are the “muscles” of the cochlea They usually get damaged first Apex Base Lows Highs Basilar Membrane
The “sharpening” Done by OHCs Cannot be Restored This is why hearing aids for ears aren’t like glasses for eyes Natural shape of fluid wave: 2 peaks from 2 tones close in Hz Hair cell damage results in: smaller rounded peaks Hearing aids make wave bigger: but cannot sharpen it Fig 1-6 , Venema, T. Compression for Clinicians 2 nd edition Cengage 2006
The Sharpened Traveling Waves… Increases Hz resolution… • Output Limiting WDRC • Our ability to distinguish between Hz’s close together Our ability to separate speech from background noise • • To compensate, we can increase signal-to-noise ratio (SNR) • By means of D mics, etc. A topic for another webinar •
We Can However, Amplify Traveling Waves… In the same manner as is done by the OHCs • Output Limiting WDRC Amplify soft sounds especially below 50 dB HL • • Focus amplification on soft inputs • Gradually decrease gain as inputs increase • The focus of Wide Dynamic Range Compression (WDRC)
Reduced Dynamic Range Focus of All Fitting Methods and Reason for Compression in 1 st Place O O O O O O O Dynamic X X X X X X X MCL Range MCL Dynamic Range UCL UCL
Reduced Dynamic Range Means We: Cannot “Mirror” Audiogram with Average Speech Inputs We can amplify: { • Soft inputs by full degree of HL X X X X X X X • Average inputs by { ½ degree of HL MCL { • Loud inputs by UCL little or nothing at all
Idea Behind ½ Gain Rule: Place Aided Speech into Dynamic Range 250 500 1000 2000 4000 8000 Hz 0 10 A A A 20 10 15 A 30 25 A A A A 25 40 30 30 30 30 50 60 70 80 90 100 LDL for Speech 110 120
Until Late 1980s: Almost All Hearing Aids Gave “Linear” Amplification Output = Input + Gain Peak Clipping Linear means same Gain for { 120 all inputs along 45 0 In the case shown here, for: Output 100 • Whisper 80 • Talking 60 0 20 40 60 80 100 Input Venema, T. Compression for Clinicians 2 nd edition, Thomson Delmar Learning 2006
Output Limiting Compression Came Next Similar to linear gain, but limit MPO with compression I/O Function Output = input + gain Output for 2000 Hz 120 High knee point: 60 dB SPL input 100 Linear gain: 80 Left of knee-point 60 Output increase = input increase 0 20 40 60 80 100 Input High compression ratio: For inputs > 60 dB SPL Right of knee-point compression dramatically limits MPO
Loudness Growth Became Focus in 1990s Normal Hearing vs Mild-Moderate SNHL Output Limiting WDRC Too Loud Loud Comfortable Soft Very Soft 0 10 20 30 40 50 60 70 80 90 100 dB HL
Along with Loudness Growth Came WDRC Focus on imitating OHC amplification of soft inputs I/O Function for 2000 Hz Cochlea is WDRC amplifier! Output 120 Knee point is lower: 40 dB SPL input 100 As with OLC, linear gain is: 80 Left of knee-point 60 Compression ratio also lower: 0 20 40 60 80 100 Input As inputs increase, For inputs > 40 dB SPL Gain is slowly decreased compression slowly limits MPO
OLC vs WDRC WDRC: OLC: Low Knee-point High Knee-point Low Compression ratio High Compression ratio Limiting WDRC Output Output 10:1 140 120 120 100 2:1 100 80 80 60 0 20 40 60 80 100 0 20 40 60 80 100 Input Input Fig 5-7, Venema, T. Compression for Clinicians 2 nd edition, Cengage 2006
OLC vs WDRC: Displayed as Frequency Responses Gain Gain 40 dB 40 dB 60 dB 60 dB 80 dB 80 dB Frequency Frequency
Different Ways of Adjusting Compression OLC WDRC Output Output max min max min Input Input OLC adjustment: Affects Knee-point and Output WDRC adjustment: Affects Knee-point and Gain
Summary A Clinical “Spectrum” of Compression Linear Limiting WDRC Associated with: High Compression ratio for: Low Compression ratio for: •Peak clipping •Loud inputs only •Soft & medium inputs Adjusted by: Adjusted by: Adjusted by: •MPO changes •MPO changes •TK changes
Summary: Clinical Application of Compression 250 500 1000 2000 4000 8000 Hz 0 WDRC 10 20 Mild-moderate SNHL 30 40 50 60 70 OLC 80 90 Severe-profound SNHL 100 110
Two (or more) Knee-points Linear, WDRC, Output Limiting TK 2 TK 1 Output Expansion Input
Today’s Digital Hearing Aids I/O Functions with Multiple Knee-Points Output limiting compression More compression Output ratio Linear gain WDRC Expansion 25 45 65 80 Input Note use of Linear gain for mid to slightly higher inputs
Is Linear Gain All That Bad? • As long as it does not distort, it can sound “clean” • That’s why HA manufacturers use it for some inputs • Eg. average to slightly louder speech inputs For the complaint: • “I can hear others at tables further away better than the person across the table!” • ADRO (adaptive dynamic range optimization) • Is an interesting departure
WDRC Shown On An Input/Output Function Note linear gain only for very soft inputs up to 40 dB SPL; WDRC occurs for average and louder inputs. As inputs increase from 40 to 100 dB SPL, the outputs increase by only half as much. This is a compression ratio of 2:1. This is how WDRC amplifies soft sounds by a lot, and louder sounds by progressively less and less 40 dB SPL input + 50 dB gain 130 Knee-point 90 dB SPL output 110 Output 60 dB SPL input + 40 dB gain 90 100 dB SPL output 70 80 dB SPL input + 30 dB gain 110 dB SPL output 50 0 20 40 60 80 100 Etc. Input
ADRO Shown On An Input/Output Function Greater and Lesser amounts of linear gain are provided, depending on the listener’s sound environment; in loud environments, less linear gain, in soft environments, more linear gain. The focus: listener’s comfort Maximum Output { 130 110 Output Note that a shift to the right actually shows a decrease 90 in linear gain! Eg. a 40 dB SPL input results 70 in only a 70 dB SPL output, so the gain now is only 30 dB 50 0 20 40 60 80 100 Input
WDRC: What it Does to Speech Unaided Speech The top sound wave represents an example of a sentence spoken at an average conversational loudness level. The peaks are the louder parts of speech; namely the vowels. The valleys are the softer parts; namely, the unvoiced consonants, like s, f, t, ch, sh, etc. Aided Speech The bottom sound wave represents the same sentence amplified with WDRC. Note how the overall sound wave is amplified, but the peak-to-valley contrast is decreased. WDRC amplifies soft sounds by a lot, and louder sounds by less
ADRO: What it Does to Speech Unaided Speech The top sound wave represents the same sentence spoken at an average conversational loudness level. Remember, the peaks (vowels) are the louder parts of speech, while the valleys (sounds like s, f, t,ch, sh, etc., are the softer parts. Aided Speech with ADRO The bottom sound wave represents the same sentence amplified with ADRO. Note how the overall sound wave is amplified, and the natural peak-to-valley contrast is preserved.
Recommend
More recommend
