Module: De-Esser - Yellowtec Digital Voice Processor Manual

8.2.5 Module: De-esser

Application: Suppression of noisy hissing sounds. The De-esser is a compressor especially
configured for a reduction of "S"- or sibilant sounds. This compressor operates on frequency
selections. This means that only certain parts of the frequency spectrum initiate control activities in the
input signal and the other signal components pass the module unchanged. This module is an
important tool for editing speech. For example, it is used to set an especially "present" sound
character of a voice with an equalizer used at the same time without loosing control of sibilant sounds.
Theoretical background: Psycho-acoustics introduced the term of "sharpness" (S) as a measure for
the portion of high frequency and complex signal components within an audio signal. The more high-
frequency signal portions referred to the power contained are within a signal, the sharper it will be
received by people. A sinusoidal sound with a frequency of 1 kHz is perceived as a reference for a
sharpness of = 1. Higher tones and complex sound such as sibilant sounds have a sharpness of S>1.
Technical implementation: The De-esser contained in VIPdigital analyzes the sharpness of the input
signal, using a special algorithm. At the same time the spectrum is analyzed, and filter parameters
such as filter center frequency, attenuation and bandwidth are evaluated. The filter operates as a
function of the input signal within the frequency range of 2 kHz and 10 kHz so that no manual
adjustment of the center frequency is required by the user. This type of adaptive filtering makes sure,
that only these frequencies are attenuated, which are over-represented in the frequency band. The
second reason for the neutrality of sound is that its control parameters are rather depending on the
spectrum-analysis than on the level-amplitude of the audio signal. The sharpness is largely
independent of the amplitude.
Basic setting: Start your settings with a basic setting of 1.2 for the Sharpness Threshold and 2:1 for
ratio. Now determine the optimum values for the three parameters by hearing tests.
Operator controls:
Sharpness Threshold
Ratio
Bandwidth
If you are not sure whether or not to activate this function, switch off the module.
VIPdigital Manual 4.0 Yellowtec 2004 Rev. 26.07.2004
This is the threshold for the psycho-acoustic sharpness parameter. If the
sharpness of the input signal exceeds the value set here, the De-esser will be
activated and reduce the sharpness of the output signal as a function of the
values for Ratio and Bandwidth (see below). The setting range of the
parameter is between 0.8 and 1.8. Values of 1.0 or lower values result in a
strongly attenuated sound of speech signals. However, only a few and
particularly sharp hissing sounds will be detected in the maximum position.
The higher the set ratio, the higher the attenuation of the adaptive filter. The
setting range of the parameter is between 1:1 and 4:1. The De-esser is
deactivated in its 1:1 position.
This parameter controls the bandwidth of the adaptive filter. The higher the set
value, the broader will be the frequency range attenuated by the filter.
Page 45