DOLBY C-TYPE NOISE REDUCTION
Introduced to the public in 1980, Dolby C-type provides 20 dB of noise
reduction above about 1,000 Hz, fully doubling the amount given by
B-type. Additional features were incorporated to deal specifically with
the difficulties of recording on slow-speed consumer tape formats.
Increased Noise-Reduction Bandwidth
Cassette tape gives a noise spectrum, without Noise Reduction, which
appears to be concentrated in the high frequencies. That's why we refer
to hiss rather than roar or rumble. Noise is less noticeable when no one
part of the audible spectrum is apparently dominant. At the other
extreme, noise with a distinct pitch, due to dominant energy in a narrow
region of the spectrum, is especially irritating. Listening to tape hiss
with B-type noise reduction reveals that the noise is not only lower in
level, but that it sounds more evenly balanced across the frequency
spectrum, because the concentrated noise at high frequencies has been
reduced. However, as you continue to reduce high frequency noise, as in
Dolby C-type, obviously the middle and low-frequency components of the
tape noise become relatively more significant. If you were to reduce the
high frequency noise by twice as much in the region of the spectrum
where B-type noise reduction operates, middle frequencies would dominate
in the resultant noise, and you might well describe it as a roar rather
than a hiss. As you apply more and more high-frequency noise reduction
to cassette tape, you need to extend the region of that noise reduction
lower and lower in frequency so that no one area of the spectrum becomes
(apparently) dominant.
Referring to Figure 6, you see that the Dolby B-type system begins to
take effect in the 300-Hz region and increases its action until a
maximum of 10 dB of noise reduction is achieved in the 4,000Hz and above
region. What you hear is an overall reduction of noise. The
remaining noise is not noticeable because it appears to be spread evenly
over the frequency range.
Figure 6. Low-level encoding characteristics of Dolby B and Dolby C.
More boost is provided by Dolby C, and it extends about two octaves
lower. The playback decoding, and thus the maximum noise reduction
effect, is the reciprocal of the curves shown.
Again, referring to Figure 6, you see that Dolby C-type noise reduction
begins to take effect in the 100 Hz region and provides about 15 dB of
noise reduction around 400 Hz and 20 dB in the critical 2,000 to 10,000
Hz hiss area. By starting two octaves lower than Dolby B-type, what
little noise that is left again appears to be spread evenly over the
frequency range and is not noticeable.
Dual-Level Processing
Covering a full 20 dB of processing range with a single sliding filter
leads to problems such as excess encoder gain during musical transients
(overshoots), and overstressed manufacturing tolerances. These problems
were solved by using TWO SLIDING COMPRESSION-EXPANSION BANDS in a
special way: both bands cover the same frequency range but are sensitive
to signals at different levels. One, of the two sliding
compression-expansion bands, is sensitive to signals at almost the same
level as a Dolby B-type processor, while the other sliding
compression-expansion band is sensitive to signals at a lower level. As
one filter reaches the end of its sliding range, the other one gradually
takes over. Each one provides 10 dB of compansion. They are connected in
series, so their effects add together to provide 20 dB of
compression-expansion and so 20 dB of noise reduction (see Figure 7).
Figure 7. Dolby C transfer characteristics, showing how the effects of
the two stages combine to produce 20 dB of compansion.
TECHNICALLY SPEAKING
Figure 8 shows how the dual-level processors work together. You can see
that the compression-expansion process is only operating at middle-level
signals (not loud or soft passages). At high signal levels (loud
passages), there is no dynamic action and the system
acts as a unity-gain amplifier. At low signal levels (soft passages) the
system acts only as a fixed-gain amplifier. By restricting the system's
compression-expansion action to only the middle-level signals,
distortion caused by overshoots is minimized.
Figure 8. The Dolby S-type NR system prevents multiplication of
compression ratios by staggering the signal levels at which gain changes
occur, as shown for the high-level stage (A) and low-level stage (B).
However, at low levels, the boost of the two stages add (C) to
provide more noise reduction.
Spectral Skewing and Anti saturation
While B-type noise reduction was first developed for use in an open-reel
format, C-type noise reduction was able to focus on the cassette's
limitations more directly. It was evident that modern music was pushing
the levels of high frequencies as time progressed. Additionally, in
noise-reduction systems the encoded signal must be precisely decoded.
But the decoder does its work AFTER the encoded signal has been recorded
on the tape. If the tape recorder and the tape don't work together well,
or differ significantly between Record and Playback performance, they
can change the encoded signal before the Dolby decoder sees it. This can
cause mistracking and possible side effects that you can hear. The
larger the amount of noise reduction that you try to do, the easier it
is to hear these side effects. With Dolby C-type, with 20 dB of noise
reduction, two additional developments were added to prevent
mistracking, particularly at higher frequencies, and to improve high
frequency quality:
Spectral Skewing: In the very first step of the encoding mode, just
before the signal is boosted, the high frequencies (above 10,000 Hz) are
precisely lowered in volume (filtered). This high frequency roll-off
causes the encoder to ignore what's happening above 10,000 Hz.
The purpose of this "spectral skewing" process is two-fold. First, the
noise reduction circuits will be much less sensitive to errors in
record-play frequency response because they can ignore
what's happening above 10,000 Hz. Second, the levels of high frequencies
actually being recorded on the tape are reduced significantly between
10,000 and 20,000 Hz, making it much easier for the tape to accurately
handle these signals. A mirror image increase in volume of the
frequencies above 10,000 Hz is provided (to restore a flat frequency
response) right after the signal is lowered in the decode mode. (See
figure 9 for location within Block Diagram of Dolby C-type) Because the
filter shape is very selective, the fact that it is boosting gain (and
noise) during playback does not degrade the quality of the noise
reduction at all.
Anti saturation networks reduce the high-frequency losses and distortion
caused by tape saturation, further reducing decoder mistracking. They
also increase recording headroom. Anti saturation networks start their
action at a lower frequency (about 1,500 Hz) than spectral skewing, so
they are only used on loud (high level) levels. Otherwise, they might
interfere with the noise reduction process. Just like spectral skewing,
during the encode mode, just before the signal is boosted, the affected
frequencies are lowered in volume. Also, just like spectral skewing, a
mirror image increase in volume of the affected frequencies is provided
in the decoder to maintain a flat frequency response.
The encoder curves shown in Figure 10 not only show the overall
characteristic of Dolby C-type noise reduction, but also show the
effects of spectral skewing and anti saturation networks. The
roll-off above 10,000 Hz at all levels is the result of spectral
skewing; the gentler downward slope beginning at about 1,500 Hz on the
higher-level curves is the result of anti saturation.
The development of the acclaimed Dolby SR Spectral Recording process for
professional audio applications brought forth a number of powerful new
audio processing techniques. Experiments with these clearly demonstrated
the potential benefits to the cassette format. After several years of
intense design work on the first integrated circuits, the first Dolby
S-type cassette decks were introduced to the public in late 1990.
Copyright © 1996 Dolby Laboratories, Inc.
--- QScan/PCB v1.18b / 01-0249
---------------