Skip to main content
. 2004 Summer;8(3):83–124. doi: 10.1177/108471380400800302

Table 2.

The Characteristics of Noise Reduction Algorithms Implemented in Selected Commercially Available Hearing Aids*

Oticon-Syncro ReSound-Canta Sonic In-Natura Siemens-Triano Widex Diva
No. of Channels 8 14 9 16 15
Type of Noise Reduction Synchrony detection + Multichannel adaptive Multichannel adaptive Multichannel adaptive Multichannel adaptive Multichannel adaptive
Signal Detection and Analysis

  1. Synchrony detector to detect synchronous energy across the upper four high-frequency channels

  2. Modulation detector to detect modulations between 2–20 Hz in the envelope of the incoming signal in each channel

  3. Noise level detector to estimate the noise level in each channel

 1. Modulation detector to detect the modulation in the envelope of the incoming signal in each frequency channel
2. Maxima modulation detector to follow the maxima in the input signal. It attempts to reduce noise in running speech without reducing audibility
3. Minima modulation detector to follow the minima in the input signal. It provides the baseline for determining the modulation, and estimates the level of noise		 2. Noise detector to estimate the steady state noise based upon modulation rate. The target modulation rate changes depending on the frequency channel
3. SNR calculation based on the noise estimate vs the amplitude of the entire signal		 2. Modulation detection block to determine the modulation rate 2. Signal detector to detect the intensity pattern of the incoming in a 30–60-sec window within a frequency channel
3. Signal detector to monitor the spectral-intensity-temporal patterns of incoming signal across frequency channels
4. Level detector to estimate the sound pressure level in each channel
Decision Rules

  1. The result of the synchrony detector is used to infer the presence/absence of speech in the incoming signal (i.e., a Yes/No analysis)

  2. The results from the modulation detector and the level detector are used to estimate the SNR within each channel

  3. The results of 1 and 2 are integrated to determine the characteristics of the incoming signal:

    Speech Only: No noise reduction

    Speech-in-Noise: Restricted gain shaped by the articulation index to ensure speech intelligibility is maintained. In general, the amount of gain reduction in a frequency channel increases with an increase in noise level, a decrease in modulation depth, or a decrease in articulation index weighting of the frequency channe

    Noise Only: Maximum gain reduction in each frequency channel

  1. Determine if the signal is speech or nonspeech based on the modulation rates detected

  2. The threshold for gain reduction depends on the noise reduction setting. For the mild and moderate settings, the threshold is 15 dB modulation. For the strong setting, the threshold is 20 dB. Above this modulation depth, no gain reduction is applied

  3. The amount of gain reduction decreases linearly as the modulation depth decreases

  4. Maximum gain reduction occurs only if the modulation depth was 0 dB, and depends on the noise reduction setting. The maximum gain reduction for mild noise reduction setting is 12 dB, for moderate is 18 dB, and for strong is 24 dB

  1. Determine if signal is speech or nonspeech, based on the modulation rates of the signal detected

  2. Gain reduction at a frequency channel depends on the estimated SNR detected. If the SNR is less than 12 dB SNR, the gain of the frequency channel is reduced. Maximum gain reduction is observed at SNR < 0 dB

  3. Maximum gain reduction (6, 12, and 18 dB) depends on the noise reduction setting. The amount of gain reduction is frequency independent

  1. Determine if the signal is speech or nonspeech based on the modulation rates of the signal detected in a 12- s window. The modulation rate of speech is assumed to be between 4–6 Hz

  2. If the detected modulation rate is outside the speech range, gain is reduced at the frequency channel

  3. The amount of gain reduction depends on the modulation depth/SNR. The exact amount is described by the Wiener Filter.

  4. Maximum gain reduction of 12, 18, 24 dB depends on the noise reduction setting and provided that there is sufficient gain in the hearing aid to allow the maximum amount of gain reduction

  1. The intensity level within a frequency channel and the spectral-intensity-temporal patterns of incoming signal give a gradual reduction of the channel gain up to 14 dB

  2. The amount of gain reduction in a frequency channel increases with an increase in input level, decrease in modulation depth, and a decrease in Articulation Index weighting of the frequency channel

  3. Adaptive changes in filter characteristics (e.g., compression ratio, amount of attenuation): fast change if modulation is detected and slow change if no modulation is detected in the frequency channel
Adaptation Speed/Speed of Gain Reduction

  1. The system moves faster to speech and adapt slower to noise

    1. When the hearing aid is in Noise Only mode, the system takes 0.2–0.9 sec to move to the Speech Only or Speech-in-Noise modes

    2. When the hearing aid is in the Speech modes, it takes the system 1.8–7 sec to change to Noise Only mode

  1. Less than 5 ms for the maximum follower

  2. 4 sec for the minimum follower

  3. Speed of gain reduction: 10 sec from 0 dB gain reduction to the maximum gain reduction setting

 The noise detector is a sliding 1.2-sec calculation. It changes gain based on estimated SNR.
Speed of gain reduction: equals to the attack time of the compression system, (i.e., between 2 and 50 ms across frequency channels)		 Speed of gain reduction: Initial gain reduction within 2 sec, maximum gain reduction is achieved within 6–8-sec Speed of gain reduction: 5-sec for a 10 dB gain change
Release Speed/Speed of Gain Recovery

  1. The exact switching times depend on the Identity setting for the hearing aid user. In general, a more active Identity setting corresponds to faster time constants.

  2. Speed of gain reduction and speech of gain recovery: Proprietary

  1. Less than 5 ms for the maximum follower

  2. 2 ms for the minimum follower

  3. Speed of gain recovery: 10 ms for the noise reduction system to recover to 0 dB gain reduction if the modulation depth is higher than the modulation threshold

 Speed of gain recovery: equals to release time of compression system (i.e., between 2 and 50 ms across frequency channels) Speed of gain recovery: Less than 1 sec Speed of gain recovery: 0.5 sec
Information Source(s) Oticon, 2004; Flynn, 2004, personal communication Groth, 2004, personal communication; Smriga and Groth, 1999 Nilsson, personal communication. US Patent 06757395; Johns et al., 2002 Powers, 2004, personal communication Kuk et al., 2002b; Kuk, 2004, personal communication
Clinical Verification Unavailable Unavailable Bray and Nilsson, 2000; Bray et al., 2002; Johns et al., 2002; Galster and Ricketts, 2004: improvement of SNR for 1–1.8 dB Unavailable Unavailable
*

These hearing aids are selected to demonstrate the range and the differences in implementation methods of noise reduction algorithms in commercially available hearing aids. ACG = automatic gain control. SNR = signal-to-noise ratio.