10/21/2013 Evidence base for HI Industry research background – Starkey hearing aid features: Hearing Technologies, 2010 ‐ 2012 Realities of hearing aid features don’t always fit with marketing spin, conventional wisdom or the ʹ what, how and why ʹ of { published research technology selection, fitting and assessment. Drew Dundas, PhD Disclosure Director of Audiology, Clinical Assistant Professor of Otolaryngology UCSF Medical Center Directional What the technology is Assessing benefit: microphones intended to accomplish Objective D ‐ Mics Digital Noise How it actually works Subjective Reduction Why you might want to The take home message Frequency Lowering recommend it DNR Frequency Lowering What, How, Why… Today’s Topics 1
10/21/2013 “an abnormally rapid growth of loudness for Abnormally rapid? Critical terms and concepts sounds presented at Growth of loudness? levels greater than the threshold of detection” Foundations… Recruitment Perceptual correlate of intensity Sound must be Change in loudness is affected by: audible to have loudness Magnitude of intensity change Duration of intensity change Loudness Loudness 2
10/21/2013 The compressor of a hearing aid can be thought Objective Subjective of as a loudness control Why do we care? system. Real Ear Insertion Gain Real Ear Aided Response REAR > Threshold = Audibility REIG ≠ Audibility Gain = (Output – Input) Device gain ≠ change in audibility Response = Intensity Gain vs. Response REIG vs. REAR 3
10/21/2013 We need both gain and Channel = A subset of the audibility to produce bandwidth for signal analysis and benefit that is: processing Objective Band = A subset of the bandwidth and where you can control gain Subjective Why do we care? Channels vs. Bands The compressor of a hearing aid can be thought of as a loudness control system. Signal processing features are gain control systems. Loudness and Gain The main course 4
10/21/2013 Directional What? microphones D ‐ Mics Digital Noise Why? Reduction Frequency Lowering DNR How? Frequency Lowering Directional Microphone Today’s Topics Technology Noise is an unwanted competitor. It can also drive the compressor level estimate. This can result in decreased signal audibility, as well as poor SNR. The theory 5
10/21/2013 Displays relative sensitivity of the mic at different angles. Convention: Up is the ‘look’ direction. Convention: Where the line gets close to the center, the mic is less sensitive. Polar Response Pattern How do they work? Fixed directional Automatic directional Adaptive directional Automatic adaptive directional A little like this… The implementations 6
10/21/2013 Switches between omnidirectional and fixed directional When to switch is governed by sound environment analysis Automatic directional Dual Omni ‐ directional Vary the time delay, vary the polar response pattern Adjust response pattern to maximize overall SNR Change time delay, Adaptive Directional change response 7
10/21/2013 Adds the low noise When conditions are benefit of an appropriate – e.g., high omnidirectional SNR, low level response pattern listening. Automatic Adaptive Directional Adaptive Null ‐ Steering Typically 20% ‐ 35% when: The sound source of interest is in front and nearby Competing noise is mainly behind or surrounds the listener Reverberation is moderate or less The instrument has a high average directivity index (DI) (3.5 – 5.5 dB) Directional Benefit DI on the head (Ricketts, 2008) 8
10/21/2013 Caveat: Microphone Drift Directional Mics are good for almost Normal Hypercardioid Pattern Degraded Pattern due to everyone, but… 0.6 ‐ dB Sensitivity Mismatch They are not magic If you don’t have audibility, DI = 6.0 dB DI = 4.0 dB they can’t help. if there is a vent, they cannot provide benefit if you are not at least 0dB insertion gain The take home message Nulls are lost. DI drops by 2 dB. Identify which parts of What? sound are speech, and Simple, right? which parts are noise. …um, no. Why? Don’t amplify the noise. How? The theory Digital Noise Reduction 9
10/21/2013 10 10 8 8 6 6 4 4 SIGNAL VALUE SIGNAL VALUE Fast vs. Slow acting 2 2 Noise reduction vs. Speech preservation 0 0 -2 -2 -4 -4 -6 -6 -8 -8 -10 -10 The implementation 0 0 0 .5 0 .5 1 1 1 .5 1 .5 2 2 2 .5 2 .5 T IM E , s T IM E , s What it does 30 20 10 SNR, dB 0 Identify Noise -10 Calculate Noise spectrum For a given Time & Frequency: -20 Turn gain up when Speech -30 Turn gain down when Noise -40 0 0.5 1 1.5 2 2.5 Identify Noise TIME, s 10
10/21/2013 TC = 0.02; Slope = 0.45; Offset = 10 30 30 20 20 10 10 SNR, dB 0 SNR, dB 0 -10 -10 -20 -20 -30 -30 -40 -40 0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5 Identify Speech Apply Gain Rules TIME, s TIME, s “Strict” Detection Speech Noise Preservation Reduction “Lenient” Detection In Running Speech A Balancing Act 11
10/21/2013 * * Acceptance of Comfort and Annoyance Background Noise (Palmer, Bentler, Mueller, 2006) (Mueller, Weber, Hornsby, 2006) Cognitive Benefits DNR makes noise more acceptable (Sarampalis et al., 2009) 12
10/21/2013 Effects of DNR: Enhanced comfort What? Some listeners may May free up cognitive experience enhanced resources for other speech understanding in Why? tasks noise May make HAs more How? acceptable The take home message Frequency Lowering Some listeners may not benefit from HF audibility The theory 13
10/21/2013 Off ‐ frequency listening /S/ ‘Sensory overload’ Distortion Adverse effects on speech understanding Theory Frequency Lowering Frequency Compression Three Current Transposition Non ‐ Linear Frequency Compression Implementations Feature Synthesis The implementations 1 2 3 4 4 5 6 5 6 14
10/21/2013 Increased audibility Decreased bandwidth at all times Reduced sound quality Frequency Compression Frequency (Hz) Frequency Transposition Technique Frequency Transposition 1 2 3 4 5 6 15
10/21/2013 Preserve bandwidth Identify HF consonant sounds Generate a spectral analogue at a lower • Increased audibility frequency • Speech cue confusion Provide appropriate audibility of the analogue • Reduced sound quality re: concurrent speech sounds Feature Synthesis Frequency Transposition • Consonant sounds replicated in real time • Bandwidth preserved • Quality usually preserved The evidence Feature Synthesis 16
10/21/2013 Frequency Lowering: Adults * Frequency Lowering: Adults (Frequency Compression) (Glista et al., 2009) (Glista, Scollie, Bagatto, Seewald, Johson, 2009) Consonant Recognition Frequency Lowering: Adults (Frequency Transposition) (Galster, Valentine, Dundas & Fitz, 2011) (Kuk, Peeters, Keenan, & Lau, 2007) 17
10/21/2013 Recent research Provides measurable Probe mic measures Problem: how much suggests that can be useful with real ‐ world benefit audibility is preserving bandwidth bandwidth limited Tuning for listener appropriate? is preferred even in techniques preference is necessary patients with suspected What targets do you Demonstrates change Like fitting targets, one dead regions aim for? in audibility size does not fit all Sound quality matters Assessing benefit The take home message Directional All can provide microphones Directionality measurable benefit for D ‐ Mics Digital Noise DNR appropriately selected Reduction Frequency lowering and fit patients Frequency Lowering DNR Frequency Lowering Positive effects Summary 18
10/21/2013 1. Have to be applying gain with D ‐ Mics and DNR to change Choose settings the output signal carefully, using The 3 concepts are linked Can occur 2. Must be audible to be perceptible verification and patient 3. Magnitude of perception is dependent on the loudness perceptions as guides growth curve Gain, Audibility & Negative Effects? Magnitude of Perceptual effects Questions? Direct sound arriving through the vent may drew.dundas@ucsfmedctr.org Comparing sealed reduce signal processing coupler measurements effects to real ‐ world is not In challenging cases, and always realistic more severe hearing losses, consider less open fittings to maximize effect Direct vs. Amplified Path 19
Recommend
More recommend
