Noise reduction method

from Wikipedia, the free encyclopedia

Noise suppression method ( English noise reduction , noise cancellation or Squelch ) are technical methods in the field of signal processing , which have the aim of the most undesirable noise to reduce in a useful signal. For this purpose, certain frequency components are suppressed or amplified in the signal using filters that are tailored to the application . Areas of application include audio technology for speech and music, acoustics in general , or image processing to minimize the noise from image sensors. Corresponding filter functions can also be used in areas of mechanics such as B. found in hydraulics .

Depending on the area of ​​application, noise suppression methods are also referred to as noise filters or interference filters .

Areas of application

Front panel of a Denon tape recorder from the 1980s. On the right below the display the buttons for switching on the Dolby B or Dolby C noise reduction system

Audio technology

A distinction is made between dynamic and static processes. In digital technology , for example with audio CDs or music storage in MP3 format, such processes are only necessary in special cases, for example with strong compression, due to the fact that they are almost completely free from noise . The main areas of application for analog noise suppression methods were sound recording on records , tape recorders and cassette recorders . Such processes are still widespread in the areas of broadcasting and voice radio , some of which are still based on analog technology .


The hearing threshold is particularly low between 1 kHz and 5 kHz, which is why noise is particularly annoying in this area. The distance between the signal level and the noise can be increased if the upper part of the audio frequency spectrum is overemphasized during recording. During playback, the original frequency response of the audio signal is restored and the noise is also reduced. In FM broadcasting service working with this method.

Another example of this is the RIAA characteristic curve for analog records. This procedure significantly reduces high-frequency noise, and therefore the remaining, rather low-frequency noise is reminiscent of a rumble.

Compander method

With the compander method, certain frequencies are boosted to different degrees depending on their input level before they are stored or transmitted, and correspondingly equalized during playback. The noise suppression effect achieved is significant. This method is used by the well-known noise suppression methods Dolby NR, B, C and S (in the studio also Dolby A and SR) as well as the less well-known HighCom , HighCom II, Super D and dbx and gained worldwide popularity, especially with cassette recorders . The Wegener Panda-1 process also became very important in analog satellite TV transmission.


Not rushing thus ready to receive analog radios in pauses or for very weak signal, they have a noise reduction ( "Squelch") that mutes the audio portion as soon as the signal received by the antenna falls below an adjustable limit value.

The license-free handheld radios ( PMR radio - Private Mobile Radio) are also operated with noise suppression activated as standard, but this can be switched off.

Image processing

In image processing , noise suppression methods reduce unwanted interference, such as B. the noise of image sensors , and thus simplify the analysis or processing of the image.

For example, with image enhancement using digital signal processing, several recordings of the same object can be made and then low-pass filtered along the time axis. This removes noise, but also objects (e.g. people walking by) that only appear in a few images. This means that an acceptable result can still be achieved with underexposure and dark source images, which would lead to a high proportion of noise in the image.


One means of realizing the filter function is Fourier analysis with subsequent selective filtering and inverse transformation by means of Fourier synthesis . The actual function of the filter, it represents a convolution operation of the transfer function with the signal, is not carried out directly, but in the spectral range as a multiplication with a combination of fast convolution . The Fast Fourier Transformation (FFT) based on the Discrete Fourier Transformation (DFT) is used for implementation.

In the following example, the filtering in the spectral range is illustrated using a one-dimensional signal, such as occurs in audio technology:

Time course of the signal
Useful signal with noise
Transformation of the signal curve over time to the frequency curve
Frequency analysis of the useful signal with noise
← Filtering →
Frequency analysis of the useful signal
Frequency analysis of the noise
Reverse transformation from the frequency profile of the filtered signals to the time profile
Filtered useful signal
Filtered noise


  • Thomas Görne: Sound engineering . Fachbuchverlag Leipzig in Carl Hanser Verlag, 2006, ISBN 3-446-40198-9 .
  • Hubert Henle: The recording studio manual. Practical introduction to professional recording technology . 5th edition. Carstensen, Munich 2001, ISBN 3-910098-19-3 .
  • Hans Lobensommer: Handbook of modern radio technology. Principles, technology, systems and practical applications . Franzis Verlag GmbH, Poing 1995, ISBN 3-7723-4262-0 .
  • Bernd Jähne : Digital image processing . 6th edition. Springer, 2005, ISBN 978-3-540-24999-3 .

Web links

Commons : Noise reduction method  - collection of images, videos and audio files