What is the meaning of digital noise reduction for hearing aids?

What is the meaning of digital noise reduction for hearing aids?

On the previous high-quality hearing instrument, the speech and noise can be distinguished, and then the useful speech is amplified to suppress the bad noise in the signal. Recent developments in digital signal processing, including intelligent signal detection and noise attenuation, have helped this realization.

The effects of the noise attenuation algorithm mainly depend on the design of the signal detection and classification system of the hearing instrument. Some early noise attenuation algorithms only test the amplitude of the signal to determine whether the sound is a useful signal or noise. For example, an algorithm design classifies sound events by amplitude or intensity over each frequency range. Those with the lowest amplitude are assumed to be noise. The gain of each band is calculated separately to minimize the effects of the identified noise. The calculated gain is then applied to each band, after which the output signal consists of reconstructed signals for each band.

However, based on amplitude alone, the input signal cannot be correctly classified as speech or noise, and the sound cannot be accurately classified based only on their magnitude. Moreover, changes in conditions have not been taken into account, and noise may be more or less than statistical values ​​at different times. For example, listening to the radio while riding a car, when the speed of the car changes, the amplitude of the noise from the car also rises or falls appropriately, but the speech from the car radio remains unchanged. The result is stronger noise, such as car-accelerated engine noise will not be classified as noise, and many non-noise sounds, such as car radios, will be classified as noise.

The modulation index is another metric that can be used to classify incoming signals as speech or noise. The modulation index is defined as the rate of change of the signal amplitude at each band. Because of the fast and frequent amplitude changes, speech has a high modulation index. On the other hand, stable noise has a low modulation index. If noise is detected in a single band, the band is reduced compared to other noise-free bands. Therefore, in order to facilitate the noise-free band, the noisy band is suppressed.

A more efficient way to signal characterization and noise attenuation is to examine several elements of the signal simultaneously. This system can be applied to signals mixed with noise of different content, so it can be designed to detect and suppress many different types of noise.

According to the intensity variation, modulation frequency and time characteristics, the sound signals can generally be classified as:

1. Steady state noise( example: Air conditioner or motor/ engine);

2. False steady state noise( example: Traffic or crowd);

3. Instantaneous noise( example: Knocking or closing the door).

4. Useful signal( example: Speech or music).

An effective signal detection and noise attenuation system must be able to accurately classify signals for this3 Noise-like1Kind. Here:

· Intensity change : Defined as the change in intensity of the audio signal over a selected period of time.

· Modulation frequency : Defined as the rate of change of signal strength over a selected period of time.

· 时间 : Defined as the duration of the signal.

4Signal type(3 Noise-like and useful signals) Can be as per the form1The placement is shown on a continuous range.

Steady state noise False steady state noise Useful signal Instantaneous noise

Intensity change Minimum <————————-> maximum

Modulation frequency 最低 <————————-> highest

duration 最长 <————————-> 最短

The analysis of intensity variations, modulation frequency and time is performed simultaneously. In each narrowband band, these features are combined to produce a signal index through which each band’s signal passes.3The meta-signal index is classified as4One of the signal types. This signal index determines how much the output of this band is amplified or suppressed. Thus, most of the gain is applied to the band containing the useful signal, and almost no gain is used for the band containing the noise. The result of this design is a reliable, accurate, and adaptive signal detection system.