Conference PaperPDF Available

Modeling the Harmonic Exciter

Conference Paper

Modeling the Harmonic Exciter

Abstract and Figures

A harmonic exciter is an audio effects signal processor applied to enhance the brightness and clarity of a sound, particularly used for vocals. This is achieved by inducing a measured amount of high-frequency distortion. In this paper, an exciter is digitally modeled and implemented as a standalone application (or plugin) using the FAUST (Functional AUdio STream) programming language for real-time audio. The model is based on the Aural Exciter by Aphex Ltd., an analog hardware unit. Technical specifications of the Aural Exciter are drawn from the original 1979 patent. The digital model performs as expected, recreating a "vintage" style audio effect.
Content may be subject to copyright.
Audio Engineering Society
Convention e-Brief 104
Presented at the 135th Convention
2013 October 17–20 New York, USA
This Engineering Brief was selected on the basis of a submitted synopsis. The author is solely responsible for its presentation,
and the AES takes no responsibility for the contents. All rights reserved. Reproduction of this paper, or any portion thereof, is not
permitted without direct permission from the Audio Engineering Society.
Modeling the Harmonic Exciter
Priyanka Shekar1, Julius Smith1
1Center for Computer Research in Music and Acoustics, Stanford University, Stanford, CA, 94305, USA
pshekar—jos@ccrma.stanford.edu
ABSTRACT
A harmonic exciter is an audio effects signal processor applied to enhance the brightness and clarity of a
sound, particularly used for vocals. This is achieved by inducing a measured amount of high-frequency
distortion. In this paper, an exciter is digitally modeled and implemented as a standalone application (or
plugin) using the FAUST (Functional AUdio STream) programming language for real-time audio. The model
is based on the Aural Exciter by Aphex Ltd., an analog hardware unit. Technical specifications of the Aural
Exciter are drawn from the original 1979 patent. The digital model performs as expected, recreating a
“vintage” style audio effect.
1.INTRODUCTION
A harmonic exciter is an audio effects processor that em-
phasizes certain frequencies in order to change the timbre
of a sound. The timbre is described as having a bright,
airy or sparkly feel, free of the harshness that can some-
times result from equalizer treble boost. The exciter in-
troduces a measured amount of high-frequency distortion
to achieve this, occasionally employing phase shifting. It
has also been termed a “sharpness maximizer” (Chalup-
per 2000), where small non-linear effects yield marked
psychoacoustic effects. This technology was patented
in 1979 by Aphex Ltd., and trademarked as the Aural
Exciter (Knoppel 1979), (Z¨
olzer et al. 2002). Competi-
tors have created similar effects, both prior to, and af-
ter the advent of the APHEX Aural Exciter (Dietz et al.
2002). Today, digital exciter effects exist in audio edit-
ing software such as Avid Pro Tools and Apple Logic
Pro. Exciters also find use in the communications and
entertainment markets to improve speech intelligibility
(Aphex 2013).
2.DIGITAL MODELING
FAUST is a functional programming language for real-
time audio signal processing, enabling generation of
standalone applications and audio plugins.1In FAUST,
a digital model of an exciter was implemented based on
the original APHEX Aural Exciter.2
2.1.Synthesis of Model
The technical information disclosed in the Aural Ex-
citer’s patent was closely referenced for modeling (refer
to Fig. 1).
The source (No. 11 in Fig. 1) is a monaural signal from
a microphone, radio tuner or amplifier combination. The
output transducer (No. 25) may be a loudspeaker or mag-
netic tape. Leaving aside the pre-amplifiers (13 and 23),
the processing for excitation occurs in the exciter (No.
19), amplitude attentuator (No. 21) and mixer (No. 20).
1http://faust.grame.fr/
2see exciter() and exciter demo() in FAUST’s effect.lib
Shekar et al. Modeling the Harmonic Exciter
The excited path passes through the exciter block. This
consists of a 2nd-order Butterworth high-pass filter, fol-
lowed by an asymmetric soft clipper. The high-frequency
components are extracted and transformed non-linearly
by the clipper. Soft clipping ensures only low-order har-
monics are generated. By clipping asymmetrically, both
odd and even harmonics are generated. The signal is then
passed through the amplitude attenuator, attenuating be-
tween 20-70% for vocals and speech. The two paths are
then mixed back before output.
Fig. 1: APHEX Aural Exciter Block Diagram (Knoppel 1979)
2.1.1.Assumptions and Modifications
Fig. 2: Digital Model Mono Block Diagram
The digital model aligns well with the Aural Exciter (re-
fer to Fig. 2). However, the following assumptions and
diversions are made:
1. The 2nd-order Butterworth highpass filter (highpass
block) has a user-selectable cutoff frequency across
a range of 1000–10,000 Hz, with a default setting
of 5000 Hz. This matches the Aural Exciter, which
has a fixed cutoff of 5000 Hz.
2. A compressor block is placed prior to the pre-
distortion gain. This was also implemented in later
models of the Aural Exciter. The compressor en-
sures that the harmonic creator is supplied with a
reasonably constant voltage level, so that similar
harmonic distortion spectra are achieved for vari-
able input levels.
3. A pre-distortion gain (pre-gain block) is placed
prior to the clipper (harmonic creator block) in the
digital model. This selectable gain takes the func-
tion of a variable gain amplifier within the exciter
block of the Aural Exciter. The pre-distortion gain
is proportional to the number of harmonics gener-
ated, as this gain determines the boosted peak of
the signal to be clipped. The digital model has a
user-selectable number of harmonics; this directly
controls pre-distortion gain. The Aural Exciter has
a user control for clipper threshold, which accom-
plishes the same clipping behavior.
4. An asymmetric cubic soft clipper (in the harmonic
creator block) is implemented in the digital model
to approximate the asymmetric soft clipper in the
Aural Exciter. This generates a gentle non-linearity
on the high-frequency components.
5. A post-distortion gain (post-gain block) reverses the
amplification of the pre-distortion gain, attenuating
the signal by the same amount. This was imple-
mented in the digital model in place of the ampli-
tude attenuator; it was assumed to carry the same
functionality.
6. Instead of allowing only a monaural input, the digi-
tal model was designed for stereo signals by apply-
ing the same exciter effect on both channels inde-
pendently.
2.1.2.User-selectable Controls
The following table details the user-selectable parame-
ters of the digital model:
Control Parameter Min Max Step
Compressor
Ratio 1.0 20.0 0.1
Threshold -100.dB 10dB 0.1dB
Attack 0.0ms 500.0ms 0.1ms
Release 0.0ms 1000.0ms 0.1ms
Exciter
Cutoff Frequency (highpass) 1000 Hz 10000 Hz 100 Hz
Harmonics (pre-gain) 0% 200% 1%
Mix (balance) 0.00 1.00 0.01
Table 1: User-selectable Parameters
2.1.3.Graphical User Interface
Fig. 3 illustrates the Graphical User Interface (GUI) of
the digital model generated in Qt by the FAUST compiler:
2.2.Analysis of Model
Due to the unavailability of the APHEX Aural Exciter for
testing the digital model against, the exciter effect in the
Apple Logic Pro editing software was used, providing a
’ballpark’ check on response.
AES 135th Convention, New York, USA, 2013 October 17–20
Page 2 of 4
Shekar et al. Modeling the Harmonic Exciter
Fig. 3: Digital Model GUI
2.2.1.Frequency Response
Frequency response was recorded by sending an im-
pulse through the digital model. The compressor was
bypassed, cutoff frequency was set to 5000 Hz, percent-
age of harmonics was set to 10.6% (the default for Logic
Pro), and mix to 0.5. It is evident that the digital model
provides a similar high-frequency boost to the Logic Pro
Exciter (refer to Fig. 4). However, the digital model has a
clear highpass characteristic, whereas the Logic Pro Ex-
citer appears to have a bandpass characteristic, or high-
pass followed by a lowpass. The phase response of the
digital model is inverted with respect to the Logic Pro
Exciter, but otherwise consistent (refer to Fig. 5).
Fig. 4: Comparison of Amplitude Responses
2.2.2.Sine Wave Inputs
Sine wave inputs at low-frequency (e.g. 100 Hz) were
examined to observe the widened spectrum due to gen-
erated harmonics. The compressor was bypassed, cutoff
frequency was set to 5000 Hz, percentage of harmonics
was set to 10.6%, and mix to 0.5. Both amplitude and
Fig. 5: Comparison of Phase Responses
phase spectra are very well matched with the Logic Pro
Exciter (refer to Figs. 6 & 7). The amplitude spectrum
exhibits a smooth rolloff of harmonics with a small rip-
ple.
Fig. 6: Comparison of Amplitude Spectra
2.2.3.Plosive Vocal Inputs
It was interesting to inspect the output of the digital
model on vocal consonants that are plosive. Plosives
are formed in the mouth by blocking air flow with the
tongue, lips or glottis, giving a percussive timbre. Sam-
ples of the plosives \k\and \p\were self-recorded. The
compressor was used in this test: threshold was set to -
30dB, ratio to 4:1, attack to 50ms (in order to preserve
the transient of the plosive), and release to 500ms. For
the exciter, cutoff frequency was set to 5000 Hz, percent-
AES 135th Convention, New York, USA, 2013 October 17–20
Page 3 of 4
Shekar et al. Modeling the Harmonic Exciter
Fig. 7: Comparison of Phase Spectra
age of harmonics to 20%, and mix to 0.5. As is expected
with the APHEX Aural Exciter, significant low-order har-
monic energy is contributed to the signal; consonants are
enhanced due to their percussive leading edges.
Fig. 8: Amplitude Spectra for Plosive \k\
3.DISCUSSION
The digital model exhibits performance that is closely
matched with the Logic Pro Exciter.
As future work, it is pertinent to take signal responses
from the original APHEX Aural Exciter in order to imi-
tate its response better. It may be desirable to have a user-
selectable even/odd harmonics ratio, as in the present-
day APHEX Aural Exciter (Chalupper 2000). This can
Fig. 9: Amplitude Spectra for Plosive \p\
be achieved by designing a smooth transition between
asymmetric clipping, which generates more even har-
monics, and symmetric clipping, which generates more
odd harmonics. The Logic Pro Exciter’s evident lowpass
characteristic could be due to the use of oversampling in
its implementation. Ideally a cubic nonlinearity should
be oversampled by at least a factor of 2 to avoid aliasing.
References
Aphex (2013). Exciter. http://www.aphex.com/the-
benefits-of-aphex/.
Chalupper, J. (2000). Aural exciter and loudness max-
imizer: What’s psychoacoustic about-psychoacoustic
processors?-. In Audio Engineering Society Conven-
tion 109. Audio Engineering Society.
Dietz, M., Liljeryd, L., Kjorling, K., and Kunz, O.
(2002). Spectral band replication, a novel approach
in audio coding. In Audio Engineering Society Con-
vention 112. Audio Engineering Society.
Knoppel, C. (1979). Signal distortion circuit and method
of use. Patent No. US4150253 A.
Z¨
olzer, U., Amatriain, X., and Wiley, J. (2002). DAFX:
digital audio effects, volume 1. Wiley New York.
AES 135th Convention, New York, USA, 2013 October 17–20
Page 4 of 4
... The exciting algorithm improves the assertiveness of the backing track. The digital signal processing methods are implemented following the APHEX Aural Exciter described in [18]. The audibility of the mixed signal is enhanced by adding harmonic distortions in the upper frequency range. ...
Conference Paper
The audio mixing process is an art that has proven to be extremely hard to model: What makes a certain mix better than another one? How can the mixing processing chain be automatically optimized to obtain better results in a more efficient manner? Over the last years, the scientific community has exploited methods from signal processing, music information retrieval, machine learning, and more recently, deep learning techniques to address these issues. In this work, a novel system based on deep neural networks (DNNs) is presented. It replaces the previously proposed steps of pitch-informed source separation and panorama-based remixing by an ensemble of trained DNNs.
Article
Full-text available
This paper explores the behavior of an algorithm called Audio Dynamics Automatic Equalization (ADAE). This algorithm has been inspired by research carried out in the context of image restoration: it is the adaptation of a contrast and color unsupervised equalizer for images, called Automatic Color Equalization (ACE), into the audio domain. Beside testing if the domain shift from image to audio processing can bring some interesting result, this work also investigates if ADAE behaves like already-known technologies for audio manipulation and restoration. To this end, after a description of the original and the derived algorithms, quantitative test are carried out using typical analyses from the Sound and Music Computing literature, such as frequency response, transfer function, and harmonic distortion. Finally, the paper discusses how the algorithm introduces dynamic range adjustments and non-linear distortions, thus behaving like a dynamics processor, a harmonic exciter, and a waveshaper.
Article
Spectral Band Replication (SBR) is a novel technology which significantly improves the compression efficiency of perceptual audio codecs. SBR reconstructs the high frequency components of an audio signal on the receiver side. Thus, it takes the burden of encoding and transmitting high frequency components off the encoder, allowing for a much higher audio quality at low datarates. In December 2001, SBR has been chosen as Reference Model for the MPEG standardization process for bandwidth extension. The paper will highlight the underlying technical ideas and the achievable efficiency improvements. A second focus will be a description of current and future applications of SBR. INTRODUCTION Back in 1997, Coding Technologies started research on a novel approach to improve the efficiency of digital audio coding algorithms. A team of Swedish researchers led by Lars Liljeryd had the idea to recreate (replicate) the high-frequencies of a decoded audio signal in an accurate way, mainly by the re-use of signal information obtained from the decoded base-band signal. Thus a new concept of redundancy coding in the frequency domain was developed. This concept led to what is today known as Spectral Band Replication technology (SBR). SBR is able to significantly improve the compression efficiency of audio codecs over a wide range of bitrates. As SBR is an enhancement technology, it always needs an audio codec to hook upon. Two obvious candidates were identified: mp3 [1,2,6], as the worlds most popular coding format, and MPEG AAC [3,7], so far the most powerful waveform audio coding algorithm. After a year of successful cooperation with the Fraunhofer Institute for Integrated Circuits (FhG-IIS), several mp3 and AAC experts from FhG joined Coding Technologies in autumn 2000 to work on the SBR enhanced audio coding systems. Since then, several applications have adopted SBR. mp3PRO is the result of combining mp3 with SBR in a back-and forwards compatible way. mp3PRO is available since June 2001 and has in the meantime gained significant market acceptance. In January 2001, AAC+SBR has been adopted as audio codec for the Digital Radio Mondiale System. In December 2001, SBR has been successfully submitted to MPEG and became reference model of the MPEG-4 version 3 audio standardization process. The following chapters will describe in more detail the DIETZ ET AL. SPECTRAL BAND REPLICATION AES 112 TH CONVENTION, MUNICH, GERMANY, 2002 MAY 10–13 2 background of SBR, the technology behind SBR and applications for this novel technology.
Article
A small dimension low frequency loudspeaker has a folded exponential horn which provides a unitary curved sound path from an electroacoustic transducer at the throat of the horn to a volume into which sound is radiated at the mouth of the horn. The length of the horn is such that, at an exponential rate of expansion between the throat and the mouth, the mouth, when it is bounded by at least one planar surface, such as a floor, a ceiling, and/or walls of a room, has adequate area to enable reproduction of low audible frequencies. An illustrative embodiment of the low frequency loudspeaker has an effective low end cut-off frequency of 55 Hz. A loudspeaker system, including a low frequency loudspeaker as well as midrange and high frequency loudspeakers and an LC crossover network, is also disclosed. The LC crossover network includes an autotransformer which not only serves as a component to determine a crossover frequency but which also boosts the electrical signal that is input to the electroacoustic transducer of a less efficient loudspeaker. The autotransformer increases the output of the less efficient loudspeaker and accommodates its use with more efficient loudspeakers so that the overall loudspeaker system operates at optimum efficiency. An illustrative embodiment of the loudspeaker system affords 108 dB SPL output at 1 meter with 1 watt input which corresponds to about 20% overall efficiency. The smoothness of amplitude response over the range of audible frequencies that is necessary for high fidelity sound reproduction is improved by inclusion of peaking circuits in the LC crossover network of the loudspeaker system to enhance amplitude response in the regions of crossover frequencies. Side wings are additionally provided to eliminate cavities at the sides of the loudspeaker system which would otherwise cause deterioration of smoothness of amplitude response.
Article
In this study two so-called "psychoacoustic processors" are examined exemplarily by applying concepts, models and methods of scientific psychoacoustics. Physical measurements of processed sounds and results of hearing experiments on speech intelligibility and sound quality (Aural Exciter) and loudness (Loudness Maximizer) are presented and discussed with regard to classic psychoacoustic models and potential new applications. Therefore, relevant psychoacoustic facts, in particular on perception of nonlinear distortion, are reviewed.
Chapter
Scitation is the online home of leading journals and conference proceedings from AIP Publishing and AIP Member Societies
Exciter. http://www.aphex.com/the- benefits-of-aphex
  • Aphex
Aphex (2013). Exciter. http://www.aphex.com/the- benefits-of-aphex/.