04

Proc. of
Proc. of the
the 77th
th
Int.Conference
Int. Conferenceon
onDigital
DigitalAudio
AudioEffects
Effects(DAFx'04),
(DAFX-04),Naples,
Naples,Italy,
Italy, October
October 5-8,
5-8, 2004
2004 DAFx
DIGITAL AUDIO EFFECTS APPLIED DIRECTLY ON A DSD BITSTREAM

Josh Reiss and Mark Sandler

Centre for Digital Music

Queen Mary, University of London
Mile End Road, London E14NS, U.K.
{josh.reiss|mark.sandler}@elec.qmul.ac.uk

ABSTRACT PCM

Interpol-
Digital audio effects are typically implemented on 16 or 24 bit 1 bit ADC
Decim-
PCM ation
Analog
ation SDM Low-pass
signals sampled at 44.1 kHz. Yet high quality audio is often en- Front End
Filter
Recorder Digital
Filter
Filter
coded in a one-bit, highly oversampled format, such as DSD.
Processing of a bitstream, and the application of audio effects on a DSD
bitstream, requires special care and modification of existing meth- Analog
1 bit ADC DSD
ods. However, it has strong advantages due to the high quality Front End Recorder
Low-pass
Filter
phase information and the elimination of multiple decimators and
interpolators in the recording and playback process. We present
several methods by which audio effects can be applied directly on Figure 1: The standard multibit PCM recording amd playback
a bitstream. We also discuss the modifications that need to be chain, (a), requires a decimation filter on the recording side and
made to existing methods for them to be properly applied to DSD an oversampling filter on the playback side, whereas Direct
audio. Methods are presented through the use of block diagrams, Stream Digital, (b), enables sound to be recorded directly in the 1-
and results are reported. bit signal format and eliminates the need for these filters.
Keywords: Sigma Delta Modulation, SACD, DSD, Digital
Audio Effects, Bitstream Signal Processing Digital (see Figure 1). As in conventional PCM systems, the ana-
log signal is first converted to digital by 64x oversampling sigma
delta modulation. The result is a 1-bit digital representation of the
1. INTRODUCTION audio signal. Where conventional systems immediately decimate
the 1-bit signal into a multibit PCM code, Direct Stream Digital
One-bit signals are used throughout the audio recording, editing records the 1-bit pulses directly.
and playback process. Most analog to digital and digital to analog The resulting pulse train has some remarkable properties. The
converters employ a sigma delta modulator that converts a signal bandwidth now extends over more than 1.4 MHz. Through the use
to a bitstream. Digital audio is often stored during production in a of high order sigma delta modulators (SDMs), the noise can be
single bit format. In addition, the high-end audio distribution for- shifted up to inaudible frequencies. And the digital-to-analog con-
mat, SuperAudio CD, employs the single bit recording format version is now as simple as running the pulse train through an
known as Direct Stream Digital, or DSD. analog low-pass filter.
The benefits of the DSD format are numerous. Improvements in Ultra-high signal-to-noise ratios as required for DSD in the audio
the traditional pulse code modulation (PCM) format from higher band are achieved through 5th-order noise shaping filters. Thus
bit rates and higher sampling rates have experienced diminishing DSD can represent signals with a frequency response from DC to
returns. This is partly due to the difficulties in implementing accu- 100 kHz. The residual noise power is held at -120 dB through the
rate high bit quantisers, but primarily due to the losses incurred audio band [2].
from filtering. PCM systems require steep filters at the input to Although single bit, oversampled formats have been found to be
block any signal at or above half the sampling frequency. Ideally, excellent for archiving, A/D and D/A conversion, and re-
a brick wall filter should be used; passing all frequencies below cording [3], they suffer from a serious drawback in the editing and
the Nyquist frequency, and rejecting all above. Yet an ideal brick mastering phase. Few tools have been developed which allow
wall filter does not exist. effective processing of audio bitstreams. To apply audio effects
In addition, requantization noise is added by the multi-stage or directly on the bitstream, it is vital that requantisation, decimation
cascaded decimation (downsampling) digital filters used in re- and interpolation be kept to a minimum.
cording and the multi-stage interpolation (oversampling) digital However, processing and audio effect creation in the 1 bit domain
filters used in playback. Increasing the sample rate, as with DVD- is appealing for many reasons. The oversampled signal has very
Audio, eases the difficulty of the brick wall filter, but does not high quality phase information, making phase vocoder-based ef-
correct the problems introduced by multi-stage decimation and fects easier and more accurate. Effects using variable delays, such
interpolation. as chorus and flange, also benefit from oversampling since inter-
This was the inspiration for a 1 bit audio format, as first proposed polation of the delay is far more precise. Furthermore, 1-bit audio
by Angus [1], and independently implemented as Direct Stream effects can be applied on the DSD signal directly before or after

DAFX-1
372 — DAFx'04 Proceedings — 372
Proc. of
Proc. of the
the 77th
th
Int.Conference
Int. Conferenceon
onDigital
DigitalAudio
AudioEffects
Effects(DAFx'04),
(DAFX-04),Naples,
Naples,Italy,
Italy, October
October 5-8,
5-8, 2004
2004

encoding, thus maintaining the simplified production chain as in 3.1. Bitstream addition
Figure 1.
The goal of this paper is to describe how to develop standard au- Perhaps the most fundamental signal processing is the addition of
dio effects on the DSD bitstream, while minimizing intermediate two signals. O’Leary and Maloberti [15] demonstrated an elegant
conversions to multibit format (thus destroying all benefits of bitstream adder (Figure 2). The oversampled nature of the bit-
DSD). Previous work [4-12] has already established that suitable stream allows one to use a simple feedback loop whereby two
IIR and FIR filters can be created, as well as some mixing tools. bitstreams are added along with the sum bit from the previous
However, common audio effects, such as compandors, expandors, iteration. When the bandwidth of the input signals is far below the
reverb, modulation, and so on, have not yet been developed. In the sampling frequency, as is the case with DSD, the output carry bits
following Sections we will demonstrate how these effects can be are an excellent representation of the average of the two signals.
applied directly on a bitstream without introducing unwanted arti- This bitstream adder is remarkable because it requires no requanti-
facts, or significant degradation of audio quality. sation, and it has been shown to be highly effective for oversam-
pled signals. The alternative, bitstream addition via the interleav-
ing of bitstreams [16], suffers degradation of audio quality due to
2. PROPERTIES OF THE DSD BITSTREAM downsampling, phase shift and possible introduction of low-
frequency noise.
There are several features of DSD which distinguish it from PCM. However, although this bitstream adder does not explicitly per-
At its heart, DSD is specified as being a 1-bit format, with a sam- form requantisation, it amounts to the same effect. Thus it acts as a
pling rate of 64*44.1 kHz, or 2.8224 MHz [13]. Little else is speci- first order sigma delta modulator and introduces some noise and
fied regarding the format, although constraints are imposed for the distortion into the audible band. The bitstream adder is suitable
archiving of DSD on SuperAudioCDs and the playback of those either for a limited duration, or when increased noise is accept-
CDs (notably, restrictions on noise levels, frequency response, able. An alternative would involve summing the signals and then
peak levels and DC offsets). However, the specifications of DSD performing high order noise shaping.
also note the following properties
A
1. The 1-bit format is such that the 1 represents a positive output sum
(+1) and the 0 a negative output (-1).
B + z-1
carry A+B
2. The 0 dB reference level has been set to 50% of the maximum 2
theoretically possible modulation depth. At least 4 out of any
28 consecutive bits must be set to 1 (and similarly for 0). This Figure 2: A bitstream adder.
maximum setting corresponds to 3.10 dB.
3.2. Delay based effects
3. Silence patterns are defined as repeating bytes where each byte
contains an equal number of 1s and 0s. By using the bitstream adder in conjunction with multiple delays,
it is possible to create a flanger or chorus effect entirely through
Unlike PCM, the DSD signal always has a power of 1 (the bits simple logic operations on the bitstream. This is indicated in
representing +1 and -1 levels). Thus any instantaneous measure- Figure 3, where BSA represents the bitstream adder from Figure 2.
ment of signal level is meaningless. Furthermore, whereas PCM This implementation is very elegant and appealing because it re-
has a strict 0dB maximum, the 0 dB limit for DSD has been im- quires no filtering, decimation, interpolation or requantisation. It
posed as a safety measure. In practice, this means that a DSD sig- deals solely with bit operations and delays. Furthermore, the de-
nal, when put through a sigma delta modulator, is unlikely to re- lays can be set to any length, and due to the high sampling rate of
sult in instability or severe clipping since its peak levels have DSD, there are far more options over the number of voices and
already been restricted to within safe margins. their placement. To weight the delayed signals, a given delay time
Silence patterns do not make sense in 44.1 kHz PCM since any may be repeated in the inputs to the bitstream adders.
repeating pattern would be ≤ 22.05 kHz and hence potentially au-
dible. A constant DC level represents silence in PCM. But for a
DSD BSA
DSD signal, constant levels (i.e., all zeroes or all ones) are not Input z-n1
allowed. A repeating pattern of 8 bits or less, on the other hand,
only has frequency components above 176 kHz, i.e., far outside the BSA
DSD
range of human hearing. Thus whenever inaudible output is re- z-n2 Output
quired, a silence pattern should be used. This is important in the BSA ... BSA
construction of many audio effects, such as noise-gating. z-n3
BSA
...

3. TIME-DOMAIN AUDIO EFFECTS z-nk-1

BSA
Most time-domain based audio effects have well-established im- z-nk
plementations [14]. The general design of these effects, when im-
plemented on a DSD signal, can follow the design used for PCM
signals. In this Section we describe those design modifications Figure 3: Implementation of a basic flanger or chorus using the
which are necessary for DSD. bitstream adder (BSA) of Figure 2.

DAFX-2
373 — DAFx'04 Proceedings — 373
Proc. of
Proc. of the
the 77th
th
Int.Conference
Int. Conferenceon
onDigital
DigitalAudio
AudioEffects
Effects(DAFx'04),
(DAFX-04),Naples,
Naples,Italy,
Italy, October
October 5-8,
5-8, 2004
2004

However, it suffers serious limitations in that it allows for no mix-

ing of signals other than additively. Furthermore, the number of DSD DSD
signals mixed in this way must be a power of 2. Successive use of s Bitstream
Output
the bitstream adder in parallel and series may mimic the effect of a X SDM
multiplier, but significant noise might then accumulate in the au-
dio band, and it still does not allow for easy implementation of a
gain control. A bit stream multiplier is essential for volume ad-
justment, or for versatile mixing of signals. Therefore, most ef-
fects will be implemented using conversion to a multibit domain, Figure 4: Modulation of a DSD bitstream.
and then a sigma delta modulator in the final stage is used for
requantisation to DSD. As shown in Section 4, this SDM can
sometimes be incorporated into the effect processing stage. Currently, the only alternative is to perform multiplication of DSD
signals via decimation to a multibit domain, and then reconverting
to DSD via upsampling and requantisation. This suffers severe
3.3. Level detector drawbacks because of the introduction of low frequency noise.
In order to implement many effects, such as noise gating, expan- However, since, the carrier signal is intended to be an internally
sion, limiting and compression, a level detector is required. In generated waveform, it need not be in DSD format. This allows
PCM, this is trivial, since the instantaneous level is given by the for mixed domain processing. The carrier signal can be generated
quantised signal at any given time. For a bitstream, however, the multibit, at the DSD sampling rate. The DSD bitstream can then
instantaneous value is either 0 or 1, corresponding to a 1 or -1, for be multiplied by this multibit signal, and converted back to single
input over the range [-Max, Max] where the maximum absolute bit output. Filtering of the output should be kept minimal since the
value of the input is some value Max < 1. purpose of most modulators, such as ring modulation, is to intro-
Nevertheless, PCM level detection usually employs a time average duce new frequencies. This system is depicted in Figure 4.
d power of the signal and bitstream level detection can do the
same. It is important however, that the time average be over
3.5. Noise gating
roughly the same amount of time but not over the same amount of
samples. The high oversampling rate demands this. An extreme noise gate operates simply as a threshold below which
Time average level detection becomes even simpler for DSD sig- there should be no signal. A noise gate operating on a DSD signal
nals. RMS estimation of power is unnecessary. One can simply has several important distinguishing characteristics which require
count the bits. Over a window of size N, where M is the number of modifications of the standard PCM noise gate in order to function.
ones in the window, P=|N-2M|/N gives an estimate of the power. First, the level detector or envelope follower requires modifica-
A value between 0 and 1 for P can set the threshold. For most tion, as mentioned in Section 3.3.
dynamic processing, standard techniques can then be applied. A Noise gating however, requires further modification. When the
variable gain can multiply the signal, with the additional require- signal has been faded to zero, the output must correspond to DSD
ment that the output is processed through a sigma delta modulator silence. It is conceivably possible that traditional techniques will
(and optionally, a low pass filter), to return the signal to DSD produce a signal that, although representing the output of an SDM
format. acting on zero input, will not be silent [17, 18]. This could occur
For an accurate envelope detector, a simple moving average filter due to small DC offsets or initial conditions of the SDM. This
should not be used. A decimation filter is preferred since it more problem is especially serious because, rather than this signal being
accurately represents the multibit level of the signal at any in- a very high frequency pattern, as DSD silence is defined, it may be
stance. It is important to note that under such a situation, decima- a very low frequency pattern and hence audible.
tion need only be used for level detection, and no additional deci- For these reasons, when silence is required at the output, as may
mation/interpolation is applied to the bitstream. be the case in a noise gate, the output bitstream is replaced with a
DSD silence pattern. If smooth transitioning between silence and
low-level signal is required, then one of the switching techniques
3.4. Modulators
described in Section 3.6 can be applied during the fade-in and
Modulation involves the multiplication of an audio signal by some fade-out stages.
carrier signal, typically a sinusoid. To do this using entirely DSD
signals would involve the multiplication of two bitstreams. Unfor-
3.6. Smooth mixing and switching of bitstreams
tunately, this is not as simple as the addition of bitstreams as in
Figure 2. The product of two single-bit signals can be obtained It is well-known that switching of PCM signals can result in audi-
with just one logical gate, an XNOR (or an AND if the signals ble artefacts due to discontinuities in the output signal. This is
were restricted to [0;Max]). However, this approach affects the avoided by strictly requiring that the PCM samples from the initial
noise-shaping characteristics. Multiplication in time domain corre- and replacement streams be identical at the point at which the
sponds to convolution in z-domain. Therefore, the resulting bit- switch is made. Samples around the switch should also be roughly
stream has four components: one from the convolution of the two identical to prevent abrupt changes in signal slope (and instanta-
signals, two from convolutions between one signal and the shaped neous frequency) as well.
noise of the other bitstream, and the last from the convolution of But the DSD signal contains historical information. That is, the
the two shaped noises. Since the last term has a flat frequency current signal is determined by a sequence of bits, and the next bit
spectrum, the result of a multiplication of two noise-shaped bit- is a function of prior states as well as current input. Thus, sample
streams is a non noise-shaped waveform, whose in-band noise matching is not sufficient. Smooth switching requires that the
limits the accuracy of processing. switch happen when the two bitstreams are synchronised.

DAFX-3
374 — DAFx'04 Proceedings — 374
Proc. of
Proc. of the
the 77th
th
Int.Conference
Int. Conferenceon
onDigital
DigitalAudio
AudioEffects
Effects(DAFx'04),
(DAFX-04),Naples,
Naples,Italy,
Italy, October
October 5-8,
5-8, 2004
2004

Level
A G

DSD
Detector
+ z-1 Q
Bitstream
B 1-G
+ -
DSD
Silence +
Figure 5: A hard noise gate implemented on a DSD bitstream. A
DSD silence signal must be used since constant DC levels are not Figure 7: Smooth switching between DSD bitstreams using a
possible. slowly changing gain and a first order SDM.

In [19], Reefman and Nuitjen described an approach to synchroni- The result of this switching scheme on input signals of frequency
sation of bitstreams which allows for seamless switching. This 1 and 2 kHz, is depicted in Figure 8. A switch is desired at 2 milli-
approach involves the use of a sigma delta modulator acting on the seconds. The example is particularly pernicious (and somewhat
mix of the two input bitstreams. However, this SDM must be syn- unrealistic) since the waveforms are very different; out-of-phase
chronised such that it produces the bitstream A when acting just and with peak amplitudes of 0.2 and 0.9. The gain is changed
on A, and the bitstream B when acting just on B. linearly from 1 to 0 over 1,600 samples, or just over half a milli-
In order to produce synchronisation, the integrator states, or initial second. Depicted are the analog input signals before conversion to
conditions of the SDM, must match those integrator states. This bitstreams, and the output signal after decimation to multibit,
synchroniser can be implemented by using a least squares ap- 44.1 kHz using a sinc2 filter. The resulting transition at 2 msecs is
proach to find integrator states which minimise the difference smooth without abrupt changes in amplitude or slope. There is a
between a DSD input signal and the resulting DSD output signal. slight and temporary increase in frequency, but this effect can be
Thus editing is done as depicted in Figure 6. When synchronisa- minimised through the use of a slower gain change or eliminated
tion is ready, the switch is changed to the central position, and G completely by using a detection scheme to find a more appropriate
is set to 1. G is slowly decreased to 0, then the output stream is time to perform the edit.
resynchronised to input stream B, and the switch is set to the Improvements to this method could also be achieved by using a
downwards position. more effective noise shaper (higher order SDM) instead of the first
An alternative switching method is proposed in Figure 7. We note order SDM in Figure 7. However, with gain equal to 1, the output
first that both input and output streams are low-pass filtered, and bitstream would not be identical to the input bitstream. To phase
the application of a slowly changing gain and a first order SDM out the effects of requantisation, and resynchronize the output
should not significantly change the bandwidth of the signal. Im- bitstream with the input stream A, we can slowly reduce the feed-
portantly, a first order SDM will have no effect on a DSD bit- back coefficients of the modulator. As feedback coefficients ap-
stream. The difference between quantization of a bit and the origi- proach zero, the modulator becomes lower order until it ap-
nal bit is zero. Thus, when G is set to 1 in Figure 7, the output proaches a first order SDM, and as before, has no effect on the
bitstream is A. As G is decreased, a cumulative error based on the bitstream.
difference between the 2 input signals is added to the quantiser
input. As G approaches 0, the difference between the output and
input bitstream B also approaches 0. Eventually, the feedback 1.0
term approaches a constant (typically non-zero) and the output
0.5
Analog Input

bitstream is identical to B. The only significant introduction of

noise is the non-shaped noise due to the first order SDM acting on 0.0
the sum of two bitstreams when the gain is in the region
0<<G<<1. However, this occurs over a relatively short period and -0.5
is minimized since both inputs are already low-pass filtered. -1.0
0 1 2 3 4
-n 1.0
z
Decimated Output

0.5
Sync1
A G 0.0

+ SDM -0.5

B 1-G -1.0
Sync2 0 1 2 3 4
Time (msec)
-n
z
Figure 8: Smooth switching between DSD bitstreams using the
Figure 6: Smooth switching between bitstreams using synchronisa- circuit from Figure 7. This is the worst case scenario, where the
tion. input bitstreams have differing amplitudes and opposing phases.

DAFX-4
375 — DAFx'04 Proceedings — 375
Proc. of
Proc. of the
the 77th
th
Int.Conference
Int. Conferenceon
onDigital
DigitalAudio
AudioEffects
Effects(DAFx'04),
(DAFX-04),Naples,
Naples,Italy,
Italy, October
October 5-8,
5-8, 2004
2004

4. FREQUENCY-DOMAIN EFFECTS DSD

DSD
Input
Virtually all frequency-domain based audio effects, such as equal- Output
isers, wah-wah, and phasers, require the construction of FIR or IIR z-1 z-1 z-1 z-1 z-1 Q

filters. A significant body of research exists on 1-bit filters. A full

discussion of 1-bit filter designs is beyond the scope of this work.
Here, we note the main research and how 1-bit designs differ from
their PCM-based equivalents. Figure 9: Configuration of a combined IIR filter and remodulator.
Angus [4] provided a means of implementing FIR and IIR filters
on the DSD bitstream. This was based partly on prior work on FIR
filters by Wong and Gray [5, 6] and Kershaw, et. al. [7] and IIR 5. CONCLUSIONS
filters by Johns and Lewis [8, 9], and on his own work concerning
the processing of one bit digital audio signals [10]. This work concerned how to apply audio effects directly on a
Equalisation is usually implemented by shelving filter design us- DSD bitstream. The general architecture of many effects is ap-
ing first order filters. In [4], Angus demonstrated a bass cut/boost proximately the same. However, major modifications need to be
control filter which acts directly on the DSD bitstream. He re- made to level detection, noise gating, and switching methods.
ported roughly equivalent performance to PCM equalizers. Conversions to the multibit domain, quantisations and filtering
should be minimized. Thus, wherever possible, processing stages
should be combined and a single requantisation step should be
4.1. FIR Filters placed at the end.
One subject which has not yet been adequately investigated is an
Filters for DSD input and output signals have several design con- empirical comparison of audio effects implemented on PCM and
siderations which distinguish them from their PCM equivalents. DSD signals. All the effects methods discussed within were ana-
The main alterations are not the same for IIR filters and FIR fil- lysed via the use of simple SDMs for requantisation and a decima-
ters [11]. For a one-bit FIR filter acting on a 64 times oversampled tion filter allowing comparison with PCM effects. However, this
DSD signal, the delay line consists of z-64 delays rather than z-1 introduces further noise and thus direct comparison is not easy.
delays. In effect, the taps are subsampled. This has the effect of Development of sophisticated decimation filters and implementa-
zero-interleaving the impulse response by a factor of 64. The fre- tion of high order SDMs would allow for a more rigorous analy-
quency response is replicated throughout the entire frequency sis. Also, proper analysis of audio effects on DSD signals requires
range. This would thus demand a high order filter, except for the listening tests comparing the signal before and after the effect is
fact that this replicated response is outside the audible range. In applied. However, DSD signals are hard to come by. A new audio
general, the out-of-band frequency response is irrelevant. Whether format, DSDIFF, has been proposed for the exchange and storage
the signal needs additional filtering is then dependent on the use of of DSD-encoded audio [20]. As the format gains acceptance, DSD
the filter and on the requirements for the high frequency content of sample files will become available and direct comparison of audio
the signal. Alternatively, one could redesign the filter using single effects on DSD and PCM signals will become possible.
delays and take into account the high sample rate and single bit
input. This approach involves a combination of cascaded integra-
tors and a sparse tap filter [4]. It is efficient, removes the high fre- 6. ACKNOWLEDGMENTS
quency noise and can achieve the desired frequency response.
The authors gratefully acknowledge the contribution of Prof.
James Angus for his comments and criticism concerning this
work.
4.2. IIR Filters

IIR filtering of a DSD signal, on the other hand, does not change 7. REFERENCES
the delays but changes the coefficients. The coefficients of the
filter can be calculated in the same way as for PCM, but the over- [1] J. A. S. Angus, “The One Bit Alternative for Audio Process-
sampling implies that their values will be very different. ing and Mastering,” Proceedings of the Audio Engineering
As has been mentioned, requantisations should be kept to a mini- Society Conference on Managing the Bit Budget, London, pp.
mum. Thus, if the filtering consists of IIR/FIR filters, a noise 34–40, 1994.
shaping filter and a low pass filter, then these stages should be
[2] D. Reefman and E. Janssen, “Signal processing for Direct
combined in such a way that there is only one requantisation in the
Stream Digital: A tutorial for digital Sigma Delta modulation
final stage. Figure 9 depicts an IIR filter which incorporates an
and 1-bit digital audio processing,” Philips Research, Eind-
SDM-based requantiser. Although such a design is efficient and
hoven, White Paper 18 December 2002
eliminates the multi-bit stage, it does not differ greatly from a
cascade of one bit filters followed by a remodulator. [3] D. Reefman and P. Nuijten, “Why Direct Stream Digital
Minimising decimation, interpolation, and requantisation is not a (DSD) is the best choice as a digital audio format,” Proceed-
drawback. These filters add to system complexity and degrade ings of the Audio Engineering Society 110th Convention,
performance. In addition, filtering in the oversampled domain is Amsterdam, Holland, 2001.
advantageous because it relaxes specifications on anti-alias and [4] J. A. S. Angus, “Direct Digital Processing of 'Super Audio
reconstruction filters at the analog interfaces, thus improving CD' Signals,” Proceedings of the Audio Engineering Society
phase linearity [12]. 108th Convention, Paris, France, 2000.

DAFX-5
376 — DAFx'04 Proceedings — 376
Proc. of
Proc. of the
the 77th
th
Int.Conference
Int. Conferenceon
onDigital
DigitalAudio
AudioEffects
Effects(DAFx'04),
(DAFX-04),Naples,
Naples,Italy,
Italy, October
October 5-8,
5-8, 2004
2004

[5] P. W. Wong and R. M. Gray, “FIR filters with sigma-delta

modulation encoding,” IEEE Transactions on Acoustics,
Speech, and Signal Processing, vol. 38, pp. 979–990, 1990.
[6] P. W. Wong, “Fully sigma-delta modulation encoded FIR
filter,” IEEE Transactions on Signal Processing, vol. 40, pp.
1605–1610, 1992.
[7] S. Kershaw, et al., “Realization and Implementation of
Sigma Delta Bitstream FIR Filter,” IEE Proc. Circuits De-
vices Syst., vol. 143, pp. 267–273, 1996.
[8] D. A. Johns and D. M. Lewis, “Design and analysis of delta-
sigma based IIR filters,” IEEE Trans. on Circuits and Sys-
tems - II: Analog and Digital Signal Processing, vol. 40, pp.
233–240, 1993.
[9] D. A. Johns and D. M. Lewis, “IIR filtering on sigma-delta
modulated signals,” Electronic Letters, vol. 27, pp. 307–308,
1991.
[10] N. M. Casey and J. A. S. Angus, “One Bit Digital Processing
of Audio Signals,” Proceedings of the Audio Engineering So-
ciety 95th Convention, New York, 1993.
[11] S. Kershaw, “Sigma delta bitstream processors - analysis and
design,” Electronic Engineering. London: Kings College,
1996, pp. 325.
[12] J. A. S. Angus and S. Draper, “An Improved Method For
Directly Filtering S-D Audio Signals,” Proceedings of the
Audio Engineering Society 104th Convention, Amsterdam,
1998.
[13] Sony and Philips, “Super Audio CD Audio Signal Properties
(SACD Scarlet Book),” March 2003.
www.superaudiocd.philips.com/Assets/Down
loadablefile/SACD_Au_DE_130(scarlet-book)
-2325.pdf
[14] Dafx:Digital Audio Effects: Wiley, John & Sons, 2002.
[15] P. O'Leary and F. Maloberti, “Bit stream adder for oversam-
pling coded data,” Electronic Letters, pp. 1708–1709, 1990.
[16] A. K. Lu and G. W. Roberts, “An analog multi-tone signal
generator for built-in self-test applications,” Proceedings of
the IEEE International Test Conference, Washington, pp.
650–659, 1994.
[17] D. Reefman, et al., “Stability Analysis of Limit Cycles in
High Order Sigma Delta Modulators,” Proceedings of the
Audio Engineering Society 115th Convention, New York,
New York, 2003.
[18] J. D. Reiss and M. B. Sandler, “They Exist: Limit Cycles in
High Order Sigma Delta Modulators,” Proceedings of the
114th Convention of the Audio Engineering Society, Amster-
dam, The Netherlands, 2003.
[19] D. Reefman and P. Nuijten, “Editing and switching in 1-bit
audio streams,” Proceedings of the 110th Convention of the
AES, Amsterdam, Netherlands, 2001.
[20] Philips, “Specification DSD Interchange File Format,” Ver-
sion 1.4 ed., 2003.
www.superaudiocd.philips.com/Information
Center/NO/FArticleSummary.asp?lNodeId=3404

DAFX-6
377 — DAFx'04 Proceedings — 377