Proceedings of

INTERNATIONAL CONFERENCE ON COMPUTING, COMMUNICATION AND ENERGY SYSTEMS

(ICCCES-16)
In Association with IET, UK & Sponsored by TEQIP-II
29th -30th, Jan. 2016

Paper ID: E&TC17

REVIEW OF LOUDSPEAKER EQUALIZATION

Ms. Geeta S. Kadam
Assistant Professor, ETC Dept, AIT, Vita
Ms. Ashwini S. Shinde
Assistant Professor, ETC Dept, AIT, Vita

Abstract— Different methods to design digital filters this will not work at higher frequencies over a
aimed for equalization of loudspeaker/room responses reasonable listening area. Several musical sounds contain
are considered. Design of inverse filters is based on only few spectral components(with substantial separation),
measured loudspeaker/room impulse responses and we have only a low density of room modes below the
combined with room- and psychoacoustic knowledge. Schroeder frequency (typical around 100 - 200 Hz),
Frequency dependent smoothing and nonlinear therefore the need for equalization in this low frequency
equalization effort is applied, and a new iterative region is very pronounced.
method has been proposed. B. Room- and psychoacoustic criteria
As mentioned, it is not possible to make a complete
Index Terms— Deconvolution, MIMO, FIR, IIR, deconvolution across the listening area. Fortunately, the
Kautz and Wrapped filter target for the equalization task is not an anechoic chamber.
Perception of distance is poor in anechoic surroundings,
I. INTRODUCTION
Loudspeaker equalization is an essential technique in the perceived distance will depend directly on the
play-back level. Further, the spatial impression is poor in a
audio system design. As loudspeaker arrangement driven
through a compensating network. The compensating standard two channel stereo configuration, without side
wall reflections. In fact, it has been proposed to improve
network has an amplitude-frequency response
the listening conditions in a typical living room by adding
characteristic which is substantially flat over the range of
frequencies for which the loudspeaker characteristic is flat, more lateral energy, see Griesinger [1]. Therefore we have
to decide to what degree the reverberant sound field shall
and which rises at higher and lower frequencies to
compensate for the decline in the loudspeaker be reduced and how we shall equalize the direct sound and
characteristic at those frequencies. The compensating the early reflections. Shall we define different target rooms
(application specific) described by the common room
network described however, does not compensate for the
periodic variations in response exhibited by many acoustic parameters like reverberation time, clarity etc? In
this case the optimal target room will depend on the type of
loudspeakers at the higher frequencies.
music. A reverberation time around 0.4 sec. seems to be an
It is an object of the present invention to extend the
appropriate target since much CD material is mixed to
frequency range over which a loudspeaker can be used to
sound good in such surroundings. What levels of early
provide high fidelity reproduction. It is a more specific
reflections are acceptable? If these levels cannot be
object of the present invention to provide a loudspeaker
achieved, can we then improve the sound quality by adding
arrangement in which variations in the
some early reflections combined with a suppression of the
amplitude-frequency response characteristic resulting
most dominant early reflections? Other fundamental
from reflections and/or mechanical resonances (i.e.
issues, related to psychoacoustic criteria are, how flat a
regardless of origin) are reduced or substantially
frequency response do we need, and what is the optimal
eliminated. To these ends, the present invention provides
equalization of peaks/dips.
speaker equalization (i.e. correction of speaker response to
C. Phase equalization
obtain a substantially flat frequency-amplitude response
A fundamental issue is: Can we ignore equalization of the
characteristic) through incorporation of discrete time filter
excess phase part in loudspeaker/room transfer functions?
means in the input to the speaker.
At the moment there is no clear answer, in an earlier
II NEED OF EQUALIZATION investigation we have shown that the excess phase, under
A. Low frequency room modes certain circumstances, is audible, see Johansen & Rubak
At low frequencies it is recommended to have a [2]. It is not clear how important it will be to separate the
complete correction of the room modes (caused by two parts of the compound impulse response for
standing waves) including high-Q room resonances; but loudspeaker/room. Craven & Gerzon claim that it is
important to compensate the phase response of the woofer
unit, and propose a linear phase response. A recent review

K.E. Society's
RAJARAMBAPU INSTITUTE OF TECHNOLOGY
 Proceedings of
INTERNATIONAL CONFERENCE ON COMPUTING, COMMUNICATION AND ENERGY SYSTEMS
(ICCCES-16)
In Association with IET, UK & Sponsored by TEQIP-II
29th -30th, Jan. 2016

Paper ID: E&TC17

concerning equalization of loudspeakers is given by tail. There is some evidence, according to Craven &
Karjalainen et al. [11] Gerzon [3], that the low-frequency part of the room
impulse- response also is close to minimum phase. A
III. EQUALIZATION OF LOUDSPEAKER AND ROOM minimum phase equalizer seems to be appropriate for
A. Introduction equalization of both the early part of the room
For small rooms (including normal living rooms) we impulse-response and the low-frequency high Q
have two very important issues, how to correct the early resonances.
reflections (maybe combined with the direct sound) and D. Frequency resolution
high Q low-frequency room resonances. The reverberation Equalization of low-frequency high Q room resonances
time is usually in the range 0.4 - 0.8 sec therefore the need requires a very high frequency resolution, about 1-2 Hz.
for a reduction is often limited. The value (in this range) of Implementation of this resolution, using an FIR filter
the reverberation time is less important than the temporal requires an unrealistic number of filter coefficients, in the
distribution and the levels of the early reflections. order of 40,000 taps. Craven & Gerzon have solved this
B. Loudspeaker equalization problem by using down-sampling. An alternative method
It is difficult to separate loudspeaker/room impulse is application of “Warped-filters”, Johansen & Rubak [7].
responses. It is possible to derive an impulse response
which is close to the effect caused by the loudspeaker IV. EXISTING WORK
alone (position independent part). The simplest way is to The existing works on loudspeaker equalization can be
use an anechoic measurement e.g. in 30 degree of the classified into several categories: Deconvolution method,
loudspeakers impulse-response. The directional properties Use of wrapping filters, MIMO feed forward control. This
are of cause not included, but the loudspeakers basic are listed below.
frequency response is accounted for. Craven & Gerzon [3] A. Convolution Method:
propose an equalization of the loudspeaker including Zhang Ping [8] proposed loudspeaker equalization is an
non-minimum phase correction. In their opinion it is essential technique in audio system design. A well-known
important to achieve a linear phase characteristic for the equalization scheme is based on the deconvolution of the
woofer high pass response. A high quality loudspeaker is desired equalized response with the measured impulse
mainly a minimum phase system, but the crossover response of the loudspeaker. In this paper, a
network can include a non-minimum phase part (all pass). post-processing scheme is combined with the
The impulse response for that part is short (a few ms), deconvolution-based algorithm to provide a better
therefore it is possible to correct the all pass part using a equalization effect. Computer simulation results are given
reasonable short delay necessary to obtain a causal impulse to demonstrate the significant improvement that can be
response. As pointed out above, it seems to be the most achieved using this method.
appropriate for small rooms to equalize the combined B.Wraped filter Design:
impulse response for loudspeaker/room. Matti Karjalainen [9] proposed a technique based on
C. Room equalization wraped filters. They allow for the design of equalizers on
Room impulse responses are generally non-minimum non uniform frequency resolution that is characteristic to
phase systems see Neely &Allan [4], Johansen & Rubak auditory perception, which enables also to use lower filter
[2]. Rooms with very short reverberation times seem to be orders which compensates for inherently more complex
closer to minimum phase. But what about the very structures of wrapped filters.
important early reflections, are they minimum phase? The proposed wrapped structure requires less precision,
Genereux [5] discusses this issue using a simple model. He avoid excessive emphasis on equalization of high
consider a direct sound combined with one broad-band frequency resonances and antiresonaces which easily
reflection (frequency independent reflection coefficient). happens with uniform frequency.
The transfer function is given by: C.MIMO feed forward control:
H(z) = 1 + a z -m Adrian Bahne. [10] In this work we presented a method for
In this case all m zeros are placed on a circle with radius loudspeaker-room equalization by means of combining a
a 1/m, and therefore we have a minimum phase system for general MIMO equalization method presented by the
a < 1. In other words, if the amplitude of the reflection is authors earlier together with a novel pair wise channel
less than the direct sound (this condition is fulfilled for similarity criterion. The similarity criterion is motivated by
ordinary rum) we have a minimum phase system. This is the requirements of multichannel standards like
not a general proof of the hypothesis that all early stereophonic or 5.1 surround sound reproductions, where
reflections represent a simple minimum phase system, but phantom images are created based on amplitude and phase
data presented by Mourjopoulos [6] show that differences between symmetric channels. Correct playback
non-minimum phase components in a measured room of recordings using these techniques, basically all
impulse response are predominantly in the reverberation multichannel recordings, thus requires symmetrical and

K.E. Society's
RAJARAMBAPU INSTITUTE OF TECHNOLOGY
 Proceedings of
INTERNATIONAL CONFERENCE ON COMPUTING, COMMUNICATION AND ENERGY SYSTEMS
(ICCCES-16)
In Association with IET, UK & Sponsored by TEQIP-II
29th -30th, Jan. 2016

Paper ID: E&TC17

therefore similar RTFs. To assess the proposed method a targets for the equalization task. Focus is put on the
measure of RTF similarity is required. To this end we problematic position sensitivity, which is very severe at
introduced the cross-correlation between two channels in mid to high frequencies. Averaging across the listening
narrow frequency bands corresponding to the critical area is one approach, but we have chosen a alternative
bandwidth of the auditory filter. The proposed method was method based on decreasing frequency resolution at higher
then investigated by means of measurements of two frequencies. Different preprocessing techniques are
multichannel audio systems considered. Optimization is based on MATLAB
The problem of loudspeaker response equalization is simulations, and evaluation of the corrected impulse
simpler than the correction of a full acoustic path including responses is based on a new software toolbox. The
room acoustics. Loudspeaker impulse responses are equalizer is based on measurement in one or 4 listening
relatively short and the magnitude response is regular in a points of the compound transfer function for
well designed speaker. EQ filter techniques proposed for loudspeaker/room/listener, and minimum phase EQ
the purpose include FIR filters, warped FIR and IIR filters design. A new method is under investigation. The
[11], and Kautz filters [12]. distribution of early reflections is modified by adding
While flattening of the magnitude response also in this reflections, to obtain a more random distribution and a
case is relatively easy to carry out, difficult problems are better balance in relation to the reverberant part. This
found particularly in reducing excessive reverberation, procedure is combined with the previously used frequency
reflections from room surfaces, and sharp resonances due domain techniques.
to low-frequency room modes. Reduction of the effect of
perceived room reverberation, in order to improve clarity,
is a very hard task because of the highly complex modal REFERENCES
behavior of rooms at mid to high frequencies. By proper [1] Griesinger, “Theory and Design of a Digital Audio
shaping of the temporal envelope of the response, for Signal Processor for Home Use”, JAES, Vol. 37, No. 1/2
example, by complex smoothing technique in EQ FIR 1989.
filter design [13, 14], this can be achieved to some degree. [2] L. G. Johansen and P. Rubak, “The Excess Phase in
This requires necessarily high-order equalization filters. Loudspeaker/Room Transfer Functions: Can it be Ignored
Counteracting room surface reflections is only possible to in Equalization Tasks?” , preprint 4181, Audio Eng. Soc.
a specified point in the space, from where the receiver is 100th Conv., Copenhagen 1996.
allowed to move less than a fraction of wavelength of the [3] P. G. Craven and M. A. Gerzon, “Practical Adaptive
highest frequency in question. At lowest frequencies, Room and Loudspeaker Equalizer for Hi-Fi Use”, AES
UK DSP Conference September 1992.
modal equalization [15] has been developed to control the
[4] S. T. Neely and J. B. Allen, “Invertibility of a room
temporal decay characteristics of modal resonances that
impulse response” , J. Acoust. Soc. Am. 66 (1), July 1979
have too high Q-values.
[5] R. Genereux, “Signal Processing Considerations for
Acoustic Environment Correction”, AES UK DSP
V OBJECTIVES OF THE STUDY Conference September 1992.
To control excessively long decays is problematic or [6] J. Mourjopoulos, “Digital Equalization of Room
difficult with conventional passive means. Also develop Acoustics”, J. Audio Eng. Soc., 42(11), November 1994
the model decay behavior of a loudspeaker-room system. [7] L. G. Johansen and P. Rubak, “Equalising and Control
To design, development and analysis FIR filter for of Sound Fields: A Novel Approach by Warped FIR
equalization and implement and performance evaluation of Filters”, proc. of Internoise, Christchurch 1998
reconfigurable FIR filter using FPGA. [8] Zhang Wing, Wang Peng, “Loudspeaker Equalization
By fulfilling the above mentioned objectives, develop a with Post-Processing”,EURASIP journal on advances in
model for loudspeaker and room equalization. signal processing (Impact Factor: 0.89). 01/2002;2002(1).
DOI: 10.1155/S111086570220801X
VI CONCLUSION [9] M. Karjalainen and et al., “Model equalizations by
Design criteria for equalization of loudspeaker/room warped FIR filter design”, in Pro. AES 23rd Int. Conf.
impulse responses are developed. The criteria are based on Copenhagen, 2003.
a systematic analysis of the equalization scenario including [10] Adrian Bahne, Lars-Johan Brännmark , Anders Ahlén
loudspeaker, room and listener. The analysis includes “Symmetric Loudspeaker-Room Equalization Utilizing a
room acoustic and psychoacoustic factors, as well as Pairwise Channel Similarity Criterion”, IEEE Trans.,
Signal Processing pp. 6276-6290.
theoretical aspects of time- and frequency domain
[11] M. Karjalainen, E. Piirila, A. J ¨ arvinen, and J.
analysis. The inherent problems considering equalization
Huopaniemi, ¨ “Comparison of loudspeaker equalization
of non-minimum phase systems are discussed. Because a
methods based on DSP techniques,” Journal of the Audio
complete deconvolution is impossible to achieve across a Engineering Society, vol. 47, no. 1-2, pp. 15–31, 1999.
reasonable listening area we discuss different possible [12] T. Paatero and M. Karjalainen, “Kautz filters and

K.E. Society's
RAJARAMBAPU INSTITUTE OF TECHNOLOGY
 Proceedings of
INTERNATIONAL CONFERENCE ON COMPUTING, COMMUNICATION AND ENERGY SYSTEMS
(ICCCES-16)
In Association with IET, UK & Sponsored by TEQIP-II
29th -30th, Jan. 2016

Paper ID: E&TC17

generalized frequency resolution: theory and audio
applications,” Journal of the Audio Engineering Society,
vol. 51, no. 1-2, pp. 27–44, 2003.
[13] P. D. Hatziantoniou and J. N. Mourjopoulos,
“Generalized fractional-octave smoothing of audio and
acoustic responses,” Journal of the Audio Engineering
Society, vol. 48, no. 4, pp. 259– 280, 2000.
[14] J. W. Worley, P. D. Hatziantoniou, and J. N.
Mourjopoulos, “Subjective assessments of real-time room
dereverberation and loudspeaker equalization,” in
Proceedings of 118th Audio Engineering Society
Convention, Barcelona, Spain, May 2005, Paper 6461.
[15] Makivirta, P. Antsalo, M. Karjalainen, and V. V ¨ alim¨
aki, “Modal equalization of loudspeaker-room responses at
low frequencies,” Journal of the Audio Engineering
Society, vol. 51, no. 5, pp. 324–343, 2003. B. D. Radlovic
and R. A. Kennedy, “Non-minimum-phase equalization
and its subjective importance in room acoustics,” IEEE
Transactions on Speech and Audio Processing, vol. 8, no.
6, pp. 728–737, 2000.

K.E. Society's
RAJARAMBAPU INSTITUTE OF TECHNOLOGY