Junger Audio d02 - Digital Dynamics Processor - Manual (Page 29)

5. APPLICATION NOTES
When mixing together a delayed signal and a direct signal there may
be cancellation of the signal waveform at some frequencies and re-
inforcement of the waveform at other frequencies (comb filter effect).
Corresponding 2ms delay of direct signals should therefore be
carried out before mixing them with delayed processed signals.
Signal compression and the loudness enhancement of the digital audio
signal resulting from it can be achieved by combining two dynamic
range control processes: firstly, the compression achieved by
increasing small and medium signal levels and secondly, linear
amplification combined with the inaudible limitation of individual,
remaining peak levels with the limiter.
In the gain change mode the operation of compressor and limiter can be
observed on the display. For smaller signal levels the compressor
causes additional amplification which however decreases the higher the
signal level is . With full scale levels the compressor is practically
ineffective so that even an increase of the RATIO will have no effect.
If you now increase the linear amplification GAIN, individual peak levels
are raised above the limiter threshold and limited inaudibly. All other
signal components can however be increased. If the gain is too large
also medium levels are treated by the limiter, which means that the
limiter then reduces the signals continually and again reduces the
additionally applied amplification. 5.5.
Selection of
parameters to
increase loudness
The display for Limiter-Gain-Reduction should be in the region of 0...-
6...-8 dB and should not light up red continuously, so that a dynamic
limitation only applies to signal peaks. Then the signal compression and
therefore also the increase of loudness is at its most effective.
5.6.
Redithering -
Reduction of word
length of digital
output signal
At the end of postproduction the material must be prepared for copy on
COMPACT DISC. The information of a 24 bits signal is not more
storable linear. One must shorten 24 bits data word to 16 bit word
length. The practice offers several procedures for it.
In the simplest case, the last bits are cut off - truncation. One requires
no further processing to this, it is enough to record a 20or 24 bits signal
direct on a 16 bits storage medium. In this case, a not unimportant
quantization mistake however results, the part of the harmonious
distortions increases. Single numeric roundoff of the signal to 16 bits
reduces this mistake. Nevertheless, the result will normally be worse
than the data by the same original analog signal converted with a good
16 bit ADC.
truncation
In order to receive a better quality during cut down the data to 16 bit one
must redithering the material with corresponding devices. Here the
device is calculating random numbers (dither signal) and add a different
random number to every sample. Then it will be cut off to 16 bit. As a
result, the bit with least weight (LSB) is put in such a way that it
corresponds best to the information of the last bits following available
ones no more and makes less distortions as hissing in the signal. redithering
5-3