You are on page 1of 9

Available online at www.sciencedirect.

com
Hearing
Hearing Research 235 (2008) 125–133
Research
www.elsevier.com/locate/heares

Research paper

Binaural perception of the modulation depth of AM signals


Edward Ozimek a,*, Jacek Konieczny a, Toshio Sone b
a
Institute of Acoustics, A. Mickiewicz University, Umultowska 85, 61-614 Poznań, Poland
b
Akita Prefectural University, Faculty of Electronic and Information Systems, 84-4 Ebinokuchi, Tsuchiya, Honjo City 015-0055, Japan

Received 5 May 2006; received in revised form 26 October 2007; accepted 29 October 2007
Available online 13 November 2007

Abstract

The purpose of this study was to determine the binaurally perceived modulation depth (m) of the low rate amplitude modulated (AM)
signals, under conditions of their dichotic presentation, i.e., when the AM signals presented to the left and the right ear had different
modulation depths. The modulation depth was determined as a point of subjective equality between the sensations of the modulation
depth of AM signals presented to the left and right ear, using a one-up, one-down adaptive procedure. Measurements were made for
the carrier frequencies (fc) of 250, 1000, and 4000 Hz, and the modulation frequency (fm) of 4 Hz. Experimental data showed that,
for sufficiently small interaural difference in modulation depth (Dm), the perceived modulation approximated the mean of the modulation
depths presented to the left and the right ear. However, for moderate and large Dm, the binaurally perceived modulation was lower than
the mean of ml and mr and the steepness of the function m = f(Dm) gradually decreased with an increase of Dm. Results of the calculation
of the binaurally perceived modulation depth, obtained on the assumption of binaural loudness summation, were found to be consistent
within the limit of standard deviation, with the experimental data for relatively wide range of Dm.
Ó 2007 Elsevier B.V. All rights reserved.

Keywords: Binaural perception; AM signals; Loudness summation

1. Introduction ments of the modulation depth of AM signals (Houtgast


and Steeneken, 1985; Ozimek and Se˛k, 1996). The Hout-
Among different aspects of binaural processing, an gast and Steneken’s method consists in determination of
important problem is the binaural interaction of amplitude the so-called modulation transfer function, which charac-
modulated (AM) signals and its effect on their perception. terizes the decrease in the AM index in a room. However,
The problem is relevant among others to the investigation such a method does not take into account the influence of
of the intelligibility of speech in a room based on measure- the interaural differences in the modulation of AM signals.
This may be a source of discrepancy between objectively
measured and subjectively estimated changes in the modu-
Abbreviations: A, amplitude; AM, amplitude modulation; c, constant; lation index. The problem is also relevant for the loudness
fc, fm, carrier and modulation frequencies; L, loudness; LSO, lateral su- meter design (Zwicker and Zwicker, 1991), simulation of
perior olive; m = f(Dm), modulation-matching function; ml, mr, modula- auditory virtual environments, an auditory scene analysis,
tion indexes on the left and on the right ear; mms, modulation depth of the
matching stimulus; mw, resultant modulation depth; MSO, medial supe-
a space sound reproduction, etc. (Blauert, 1983; Bregman,
rior olive; p, exponent; r, correlation coefficient; S, slope of the modu- 1990). While there is quite a lot of data on the monaural
lation-matching function; SAM, sinusoidal amplitude modulation; SD, processing of AM signals little is known about the binaural
standard deviation; SOC, superior olivary complex; Dm, the interaural processing of these signals. Deeper experimental insight
differences in modulation into how the binaural system process sounds with tempo-
*
Corresponding author. Tel.: +48 61 8295133; fax: +48 61 8295123.
E-mail addresses: ozimaku@amu.edu.pl (E. Ozimek), koniaku@ rally varied intensity and transforms it to psychoacoustical
amu.edu.pl (J. Konieczny), tsone@akita-pu.ac.jp (T. Sone). sensations is of particular importance. The present study

0378-5955/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.heares.2007.10.008
126 E. Ozimek et al. / Hearing Research 235 (2008) 125–133

addresses the questions related to the binaurally perceived reported by Ozimek and Sone (1997) and aims at determi-
modulation depth of the AM signals with a low modula- nation of the perceived modulation depth when two AM
tion frequency (4 Hz), when the modulation indexes on signals with different modulation depths are presented bin-
the left (ml) and on the right ear (mr) are different. More aurally. The second aspect of the study is related to the cal-
specifically, the question is how the binaurally perceived culation of the binaurally perceived modulation depth
modulation depth depends on the interaural difference in assuming the binaural loudness summation and supposi-
modulation depth or is the perceived modulation depth tion that the peak loudness, proportional to AM peak
equal to the mean of the modulation indexes presented to amplitudes, is a basic cue for evaluation of the perceived
the left and the right ear? These questions, which have modulation depth. Comparison of the calculated and
not received much attention in binaural investigations to experimental data have shown their consistency for wide
date, are the basic ones posed in this study. The answers range of the interaural difference in modulation depth.
to these questions would extend the knowledge on the pro-
cessing of dynamic interaural intensity differences by the 2. Experiment
binaural system.
The characteristic feature of the AM signal perception, 2.1. Procedure
with a low rate of modulation, is a dynamic change in loud-
ness. The sensory quality associated with amplitude modu- The experimental procedure consisted in adjusting by
lation at a low modulation rate is often referred as the the subject the binaurally perceived modulation depth of
‘‘fluctuation strength” (Zwicker and Fastl, 1999). The basic the matching stimulus (mms) to that of the test stimulus,
issue of the present study is the subjective estimation of the according to a one-up, one-down adaptive procedure.
modulation depth of AM signal based on the perception of Two successive AM stimuli were presented on each trial:
loudness changes resulting from changes in the modulation the matching stimulus with equal modulation depths at
index of AM signal. In our study, this perception took the two ears and the test stimulus (target) with the same
place in the binaural scenario which resulted in complex (diotic conditions) or different (dichotic conditions) modu-
interaural interaction of AM signals. One of the basic ele- lation depths at the left and the right ear. The first trials
ment of this interaction was the binaural loudness summa- started with the modulation index mms of 10% above the
tion of the signals presented to the left and right ear. So far average modulation depth of the target so that the subject
some aspects of the binaural loudness summation have could easily notice the difference in AM modulation
been studied but for steady sounds. Hellman and Zwislocki between the matching and the target signals. The modula-
(1963), and also Marks (1978) using pure tones, reported tion depth of the matching stimulus was varied by changing
perfect binaural summation of loudness, whereas others the amplitude ratio of the modulator to the carrier at a con-
applying pure tones and bands of noise have reported less stant power of the SAM stimuli. It was varied in 1-dB steps
than perfect summation (Scharf and Fishken, 1970) or until four turnpoints were reached, and then in 0.5 dB steps
have questioned such a summation. Reynolds and Stevens for the rest of the run. A single run was made of a set of 40
(1960) showed that at low levels the monaural signal has to trials. For each run, at least eight turnpoints were obtained.
be increased by only about 3 dB to match the loudness of The average of the eight turnpoints determined the single
the diotic signal, however, at high levels the required estimate (threshold) of the binaural modulation depth.
increase is about 12 dB. Moore and Glasberg (2007) devel- The final threshold value was calculated as an arithmetic
oped the binaural loudness model, based on the idea of mean of five single estimates taken from 5 runs. In a few
inhibitory interactions between the outputs of the two ears, cases when the standard deviation of the final threshold
which predicted that a sound presented binaurally (dioti- exceeded 20% of the mean value, three additional runs were
cally) was 1.5 times as loud as the same sound presented performed and all estimates were averaged. A software pre-
monaurally. Some authors have shown that the amount caution was taken against possible overmodulation of the
of binaural summation of loudness depends on the mode matching stimulus (mms 6 1) and to compensate for varia-
of presentation (diotic or dichotic), interaural level differ- tions in overall intensity level of AM signals as a function
ences, spectral and temporal distribution, etc. (Algom of modulation (Viemeister, 1979). To prevent distraction
et al., 1989; Zwicker and Zwicker, 1991). These data show of the subject’s attention during the run, no feedback indi-
that the mechanism of the binaural loudness summation is cating the correct/wrong responses was provided.
still uncertain and its explanation needs further study. To check for possible bias effects, two pilot experiments
Up to now, the data on the processing of AM signals in were performed, in which the subject started with the mod-
the scenario of their binaural perception have been very ulation index mms of 10% below the average modulation
limited, while binaural loudness summation or binaurally depth of the target and in which the modulation depth of
perceived modulation depth have not been studied at all. the target was varied (both for diotic and dichotic presen-
The dynamic coding of the envelope of AM signals is tations) to adjust to the constant modulation depth of the
important for better understanding of the signal processing matching stimulus. No systematic or significant differences
at higher levels of the auditory system. The investigation were found between the data obtained in the pilot experi-
undertaken at the present study substantially extends that ments and those in the basic study.
E. Ozimek et al. / Hearing Research 235 (2008) 125–133 127

2.2. Stimuli (modulation depths at the right ear): 0%, 10%, 20%, and
60% were chosen. Measurements of the binaural modula-
The stimuli were the AM signals presented binaurally. tion depth were carried out for the interaural differences
They are generally described by the formula in modulation (Dm = ml  mr) of 0%, 20%, 40%, 60%,
s(t) = [1 + mcos(2pfmt)]cos(2pfct), where m (0 6 m 6 1) is 80%, 90% and 100% (ml and mr – modulation depths pre-
the modulation index (modulation depth), fm and fc are sented to the left and right ear, respectively). The modula-
the modulation and carrier frequencies, respectively. The tion depth values presented to the left ear (ml) for different
factor of [1 + mcos(2pfmt)] is called the instantaneous combination of mr and Dm are given in Table 1.
amplitude of s(t). The AM signal is characterized by a spec-
trum with three components representing the carrier fre- 2.3. Subjects
quency and the two symmetrically located side bands
corresponding to the sum and the difference of the carrier Five paid subjects (two women and three men, students
and the modulation frequencies. At a low modulation fre- from outside of our institute), aged 20–23, with normal-
quency considered in this paper, all components of the AM hearing but without psychophysical experience were tested
stimulus fell within the narrow frequency band, yielding individually in a soundproof chamber. The pure-tone
the slow intensity fluctuation, which was perceived as a sen- audiograms for the left and right ear of each subject did
sation of fluctuating loudness. The low modulation fre- not exceed 5–8 dB HL over the 6 octaves from 250 to
quency chosen coincided well with the prominent 8000 Hz. None of them was an author of this paper. The
envelope modulation (4 Hz) in speech (Houtgast et al., subject’s task was to indicate the stimulus with a larger
1980). modulation depth perceived as fluctuating loudness. In this
The stimuli were generated digitally and converted to way the point of subjective equalization between the sensa-
analog signals with a two-channel, 16-bit DA converter tions of the modulation depth of the target and that of the
at a sampling rate of 40 kHz and then low-pass filtered at matching stimuli was achieved.
10 kHz (Tucker-Davis Technologies III, TDT). The overall Differences in modulation depth were accompanied by
duration of each stimulus was 1000 ms, including 20 ms certain lateralization effects, mainly in the form of the
raised-cosine rise/fall ramps. The stimuli in each trial were movement of perceived location (from side to side between
separated by a 600-ms interval and presented in a random ears) of a sound image having a spatially broad form. The
order in a run. They were presented binaurally to the sub- subjects were informed about these effects and were asked
jects through Sennheiser HDA 200 headphones. Special to pay attention to the differences between the modulation
care was taken to accurately calibrate the two binaural depth of the target and the matching stimuli. Initially, sub-
channels in order to ensure that their parameters would jects experienced some difficulty comparing the modulation
be the same over the wide range of frequencies tested. depth of the target to that of the matching stimulus due to
Measurements of estimates of the modulation depth per- the presence of lateralization effects. Similar difficulty in
ceived by the subject were made for the modulation fre- measuring the perceived modulation depth in monaural
quency of 4 Hz, and sinusoidal carrier frequencies of 250, scenario was encountered by Moore et al. (1996) who
1000, and 4000 Hz, at an equal overall loudness level of investigated the effect of loudness recruitment on the per-
70 phons (based on the rms of the whole signal) for all ception of amplitude modulation. After 4 h of practice
SAM stimuli tested. The overall loudness level corre- their three subjects were able to produce consistent results.
sponded to 72, 70 and 67 dB SPL (an average across sub- In our experiment, prior to data collection, the subjects
jects) for carrier frequencies 250, 1000 and 4000 Hz, received a minimum of 5 h of practice to learn to focus
respectively. These levels were determined on the basis of their attention on the difference in the modulation depths.
binaural loudness matching procedure (Schlauch, 1994). After such practice, in which all stimuli were used, the sub-
The starting phases of the sines of the carrier and modula- jects declared no difficulty in indicating the stimulus in a
tor, for both the target and the matching stimuli, were set pair with the larger modulation depth and their perfor-
at zero. The modulation rate and carrier frequencies of mance had stabilized.
the matching stimuli were equal to those of the target ones.
In dichotic listening conditions, four different mr values 2.4. Results

Table 1 2.4.1. Diotic listening conditions


The modulation depth values presented to the left ear – ml (bold-italic At the first stage of the experiment, binaurally perceived
data) for different combination of mr and Dm
modulation depth were determined for diotic listening con-
Interaural difference in modulation depth Dm (%) ditions i.e., when the modulation depths of the target at the
0 20 40 60 80 90 100 left and right ear were the same (Dm = 0). The modulation-
mr values (%) 0 – 20 40 60 80 90 100 matching test aimed at defining the subject’s ability and
10 10 30 50 70 90 100 – accuracy in evaluating the binaurally perceived modulation
20 20 40 60 80 100 – – depth within the range of parameters measured. Fig. 1
60 60 80 100 – – – –
shows the modulation depth value of the matching
128 E. Ozimek et al. / Hearing Research 235 (2008) 125–133

100 fc = 250 Hz fc = 1000 Hz fc = 4000 Hz


Matched modulation depth [%]

80

60

40

20

0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100

Modulation depth of target ml = mr [%]

Fig. 1. Dependence of the binaurally matched modulation depth (m) on the modulation presented to the left (ml) and right (mr) ear, under diotic
conditions of perception (ml = mr), for three carrier frequencies fc equal 250, 1000 and 4000 Hz. Data averaged across five subjects. Dotted lines (diagonals
of the panels) show theoretical dependencies according to formula (13) (see p. 3). The error bars indicate inter-subject standard deviation.

stimulus adjusted to that of the target, expressed in per- accuracy despite some lateralization artefacts caused by
centage, as a function of the modulation depth of the tar- modulation. This statement is important for reliable mea-
get, averaged across subjects, for the carrier frequencies surements of the binaurally perceived modulation depth
of 250, 1000 and 4000 Hz. The error bars indicate ±1 stan- in the dichotic listening conditions.
dard deviation across subjects. The bars are not shown
when smaller than the symbol used to represent the point, 2.4.2. Dichotic listening conditions
as was the case for a low modulation depth of the target. In dichotic listening conditions the modulation depths
According to Fig. 1, the experimental results obtained for the target were different at both ears, whereas the mod-
for the diotic condition lay on the straight line (diagonal ulation depths for the matching stimulus were the same.
of the plot) and are characterized by small values of SD. The data for these conditions (ml 6¼ mr) are presented in
This indicates a good agreement between the modulation Fig. 2. The plots (solid lines) show the measured modula-
depths of the matching and test stimuli for all subjects tion depth of the matched stimulus as a function of the
tested, regardless of the modulation value. The relation- interaural difference in modulation depth Dm = ml  mr,
ships obtained are independent of the carrier frequency. for fc = 250, 1000 and 4000 Hz. The dotted lines indicate
A regression analysis of the data revealed that relationships the relation m = f(Dm) calculated according to formula
are highly linear (r = 0.99). The data resulting from this (12) (see p. 3). The parameter of the curves is mr (modula-
experiment have confirmed the subjects ability to perform tion depth presented to the right ear) equals to 0%, 10%,
matching of modulation depths of stimuli used to a high 20%, and 60%. The error bars indicate standard deviation

100 fc = 250 Hz mr = 0 % fc = 1000 Hz fc = 4000 Hz


Matched modulation depth [%]

mr = 10 %
mr = 20 %
80 mr = 60 %

60

40

20

0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100


Interaural difference in modulation depth Δ m [%]

Fig. 2. Dependence of the binaurally matched modulation depth (m) on the interaural difference in modulation depth (Dm = ml  mr), under dichotic
conditions of perception (ml 6¼ mr), for three carrier frequencies (250, 1000 and 4000 Hz). Results for four different mr values are indicated by the four
plotting symbols. Data averaged across five subjects. Solid lines show experimental data, dotted lines theoretical dependencies according to formula (12)
(see p. 3). The error bars indicate inter-subject standard deviation.
E. Ozimek et al. / Hearing Research 235 (2008) 125–133 129

across the subjects. The bars are not seen when they are the modulation depth mr was significant [F(1,
smaller than the symbol used to represent a given point. 216) = 881.08, p < 0.05]. No statistically significant effect
Similar dependencies (not displayed here) were obtained of fc and subjects on the perceived modulation depth was
for the series of additional measurements in which the same evident, nor was there any signifi cant interaction of sub-
four values of modulation depth (0%, 10%, 20%, and 60%.) jects with fc.
were presented to the left not the right ear. In these addi-
tional measurements, the matched modulation depth val- 3. Estimation of binaurally perceived modulation depth
ues were similar to those obtained in the basic
experiment within the limit of error (across the subjects) In our estimation it was assumed that the peak loudness
less than 6% for 0%, 10% and 20% of the modulation (not overall one), proportional to the AM peak amplitudes,
and 11% for 60% of the modulation. These data indicated is a basic cue for evaluation of the binaurally perceived
reasonably good symmetry of binaural perception of the modulation depth, at least for low rate of modulation rate.
subjects tested. Such an assumption is supported by Bauch (1956) data
As shown in Fig. 2, the matched modulation depth m, who reported that for a low modulation rate, loudness
presented on the ordinate axis, increases with increasing was determined by the peak level. Zwicker and Fastl
interaural difference in modulation depth Dm, and the rate (1999) showed that for a narrow frequency separation of
of the increase is Dm dependent. To quantify the rate of the two-tone complexes (up to about 10 Hz) the loudness level
increase in the perceived modulation depth versus Dm, the was determined by the peak value of the sound pressure
approximate slopes S = mms/Dm of the modulation-match- reached within the beating period.
ing functions were calculated in two ranges of Dm, between It is worth mentioning that the hitherto literature on the
0–40% and above 40%. It was found that for the range of loudness of modulated sounds reported different findings
0–40%, the mms value, for all carrier frequencies tested, on the effect of modulation on the perceived loudness.
exhibited approximately an increase of about 14%, which For example, Moore et al. (1999) data indicated that equal
corresponded to the slope S0–40 of about 0.35. For the sec- rms levels led to approximately equal loudness of steady
ond range of Dm (above 40%), and for consecutive mr, the and modulated sounds. Grimm et al. (2002), investigating
slopes S>40 amounted to 0.29; 0.27; and 0.25 (an average the perception of loudness of fluctuating sounds have
0.27) for fc = 250 Hz; 0.25; 0.28; 0.23 (an average 0.25) reported that the loudness was influenced by temporal
for fc = 1000 Hz, and 0.22; 0.24; 0.22 (an average 0.23) envelope i.e. a high crest factor led to increased loudness.
for fc = 4000 Hz, respectively. The obtained data suggest For the loudness of the modulated sounds they have found
that for the dichotic listening condition, the slope of the small level differences (of the order of 1 dB) between
perceived modulation depth versus Dm is steeper for the equally loud modulated and unmodulated signals. Zhang
small interaural modulation depth than for large Dm. It and Zeng (1997) have suggested that loudness was deter-
is worth noticing that the corresponding slope of the mod- mined more by the peak level at low sensation levels and
ulation-matching functions, for the diotic condition (see more by the rms level at high sensation levels. Other papers
Fig. 1), was equal to 1. Generally, one can say that the suggested that the loudness of AM sounds may be also
matched m is almost linearly related to Dm for small Dm related to the compressive property of the basilar mem-
(less than 40%), however, for large Dm, the relation brane (Florentine et al., 1996; Buus et al., 1997). AM sig-
becomes more complex and the perceived modulation devi- nals with the same rms level (i.e., the same overall
ates from linearity with increasing Dm. Besides, with loudness) but with different peak factors caused by different
increasing interaural differences Dm, the variability in the modulation indices may have different loudness, since their
matched m also increases. For small Dm, the SD calculated excitation of the basilar membrane differs when the com-
across the subjects does not exceed 5%, but for large Dm, pression mechanism is included (Carlyon and Datta,
the SD increases to about 10–15%. 1997). One may assume that the effect of the compression
The data presented in Fig. 2 do not indicate any partic- changes with increasing modulation index since such an
ular effect of the carrier frequencies tested on the perceived increase results in greater amplitude changes.
modulation depth. As Fig. 2 shows, none of the specific From the physiological point of view, at some stage of
changes in the m(Dm) relations are found at 250; 1000 binaural perception, changes in the modulation depth of
and 4000 Hz frequencies. Similar subject’s performance AM signals, are reproduced by interactions between the
across the three carrier frequencies imply a central process- neural firing patterns of the two ears. The superior olivary
ing of the binaural modulation perception. complex (SOC) is the site at which the neural firings from
The experimental data were subjected to a within-sub- the two cochlear nuclei first interact. This interaction
jects analysis of variance (ANOVA) with factors: interaural occurs in two nuclei of the SOC i.e. in the medial superior
difference in modulation (Dm), modulation depth (mr) and olive (MSO) and lateral superior olive (LSO) (Irvin, 1992;
carrier frequency (fc). This analysis indicated that the effect Schwartz, 1992; Kuwada et al., 1997). However, the type
of the interaural differences in modulation on the binau- of neural firing interaction in SOC is still uncertain. The
rally perceived modulation depth was statistically signifi- physiological data imply that the MSO receives excitatory
cant [F(3, 216) = 550.72, p < 0.05]. It also showed that inputs and the LSO inhibitory inputs primarily from the
130 E. Ozimek et al. / Hearing Research 235 (2008) 125–133

ipsilateral ear (Kuwada et al., 1997). Further and more It was assumed that in case of binaural presentation, the
complex neural interactions occur at higher levels of the resultant loudness perceived by the subject, can be
hearing pathway mainly in the auditory brainstem struc- expressed as
tures and in the auditory area of cerebral cortex (Phillips Lw ¼ cðLl þ Lr Þ ð6Þ
et al., 1991; Altschuler et al., 1991; Wang and Sachs,
1992; Kuwada et al., 1997). where Ll and Lr – peak loudness in the left and right ear, c –
Let us estimate the modulation depth of the binaurally a coefficient with a value changing within a certain range. If
perceived signals, on the basis of the interaction of their we assume a linear summation of loudness during binaural
loudness. It is generally known that the loudness, L, is a perception of signals (Hellman and Zwislocki, 1963;
power function of the signal intensity, I (for sensation level Marks, 1978), the value of the coefficient c = 1. If it is as-
above 30 dB SL). Thus, one can write sumed that the summation is not fully linear (Scharf and
Fishken, 1970), the value of coefficient c should be taken
L ¼ kI p ð1Þ
as smaller than 1.
where the exponent, p, is 0.3 (Hellman and Hellman, 1990; The change in loudness DLw for AM signal can be gen-
Green, 1993; Plack and Carlyon, 1995) and the constant k erally written as the difference between maximal and mini-
is 1/15.849 (for a sone scale (Hartmann, 1997)). For sensa- mal loudness, i.e.,
tion levels lower than 30 dB SL, no consensus exists on the DLw ¼ Lw max  Lw min ð7Þ
form of the loudness function, but generally it is steeper
than at moderate level (Buus et al., 1998). We assume that Changes in the left and right ear can be expressed in a sim-
the modulators are in phase in both ears (as was done in ilar way:
the experiment). DLl ¼ Ll max  Ll min
The increment of loudness, dL, depending on the incre-
DLr ¼ Lr max  Lr min
ment of intensity changes, dI, can be obtained by differen-
tiating expression (1). In this way one can get Using (6), the maximal and minimal loudness can be ex-
ðp1Þ pressed as
dL ¼ pkI dI ð2Þ
Lw max ¼ cðLl max þ Lr max Þ ð8Þ
For limited increments of intensity, the formula (2) can be
approximated by (2a) Lw min ¼ cðLl min þ Lr min Þ ð9Þ

DL ¼ pkI ðp1Þ DI ¼ pkI p DI=I ¼ pLDI=I ð2aÞ Subtracting (9) from (8) we get
Intensity increment DI depends on the difference between DLw ¼ cðDLl þ DLr Þ ð10Þ
its maximal and minimal value, i.e., Putting (5) (which applies to the dichotic and monaural
DI ¼ I max  I min case) into (6) and then into (10) we get
where 4p½cðLl þ Lr Þmw ¼ 4pcLl ml þ 4pcLr mr ð11Þ
where mw means the resultant modulation depth produced
I max  A2max and I min  A2min
during the binaural perception of AM signals. After simpli-
Amax and Amin – maximal and minimal value of signal fication of (11) one can get the final equation for mw, for
amplitude. Since for SAM signals the case of binaural perception of amplitude modulated
Amax ¼ A0 ð1 þ mÞ and Amin ¼ A0 ð1  mÞ signals with modulation indexes at the left and right ears
ml and mr and modulators in phase in both ears.
where A0 is the amplitude of a non-modulated signal, hence
mw ¼ ðLl ml þ Lr mr Þ=ðLl þ Lr Þ ð12Þ
2 2
I max  A20 ð1 þ mÞ and I min  A20 ð1  mÞ
Eq. (12) indicates that the resultant modulation depth, mw
and formed during binaural perception of AM signals, depends
DI  A20 ð1 þ 2m þ m2 Þ  A20 ð1  2m þ m2 Þ  4A20 m ð3Þ both on their modulation indexes ml and mr, and peak
loudness Ll and Lr.
Assuming for changing intensity (DI) and for the average In our calculation it was assumed that Ll and Lr were
intensity ðI  A20 Þ the same proportional constants, one proportional to AM peak amplitudes raised to 0.6 power.
can transform expression (3) to the form: The numerical estimates of the Ll and Lr, were determined
0:6
DI=I ¼ 4m ð4Þ 1
according to the formula: Ll ¼ 15:85 A0:6 1
max l ¼ 15:85 ½A0l ð1 þ ml Þ
0:6
Substituting (4) to (2a) we get (5).
1
and Lr ¼ 15:85 A0:6 1
max r ¼ 15:85 ½A0r ð1 þ mr Þ , A0 is the ampli-
tude of a non-modulated signal. The peak amplitude of
DL ¼ 4pLm ð5Þ AM signal was equal Amax = A0(1 + m). For m = 0,
where L is the loudness according to (1). Eq. (5) gives the Amax = A0 where A0 amplitude of the non-modulated sig-
description of the loudness increase DL related to the mod- nal. In dichotic presentation that is when ml 6¼ mr, and
ulation depth m. Ll 6¼ Lr, the peak amplitudes were different (since modula-
E. Ozimek et al. / Hearing Research 235 (2008) 125–133 131

tion indexes ml and mr were different), and the binaural AM signals by CANS than on that at the peripheral level.
modulation depth was calculated on the basis of Eq. (12). CANS comprises the cochlear nucleus complexes, a few
For the diotic presentation of AM signals, Ll = Lr since structures of the brainstem and the auditory cortical fields.
ml = mr. Thus, Eq. (12) can be simplified to the form An initial convergence of the information from the two
mw ¼ ðml þ mr Þ=2 ð13Þ cochlear nuclei takes place in the superior olivary complex
(Atkin et al., 1984; Yin and Kuwada, 1984; Phillips et al.,
Fig. 2 shows the curves of the binaurally perceived modu- 1991). Responses of brainstem to the interaural changes
lation depth, determined in accordance with Eq. (12) (thin in intensity show that the cortex processes many of the bin-
dotted lines) as a function of the interaural depth differ- aural interactions [Kuwada and Yin, 1983]. Band-pass
ence, for three carrier frequencies. As can be seen, a good characteristics of central auditory neurons responding to
agreement between the predictions and the experimental re- AM signals exhibit considerable selectivity and multiple
sults is obtained for small and medium interaural differ- levels of transformation arising from neural mechanisms
ences in modulation depth, Dm. However, for large Dm, in the central auditory system. Specific cortical neurons
the matched m starts to depart systematically from the data can sum spike discharges received from the two inputs
predicted by Eq. (12). (Kitzes et al., 1980). This supports the assumption related
to the binaural loudness summation taken in p. 3. The neu-
4. Discussion rons from central parts of the auditory system may exhibit
a decrease in the discharge rate with increasing stimulation
The experimental data resulting from the diotic presen- (Suga, 1992), limiting the accurate processing of the large
tation of AM signals, i.e., when mr = ml, have shown the interaural differences in modulation depth. This to some
subjects’ high accuracy in adjusting the modulation depth extent accounts for the flattening of m = f(Dm) function
of the matching stimulus to that of the test stimulus. This seen for large Dm.
holds true for all carrier frequencies tested. Our data do not show statistically significant changes in
In dichotic presentation of AM signals (ml 6¼ mr), the the perceived modulation depth versus the carrier frequen-
binaurally perceived modulation depth was approximately cies tested. Since the sensation of the modulation depth for
the mean of ml and mr, for interaural difference, Dm less low fm is based on the perception of the peak loudness (as it
than 40%. However, for a larger Dm the perceived modula- was assumed) one can expect that binaural loudness sum-
tion was generally less that the mean of ml and mr. In this mation is independent of frequency. This agrees with the
case significantly different values of the modulation depths experimental finding showing that binaural loudness is
at the two ears produce more spatially broadened independent of frequency (Zwicker and Zwicker, 1991). It
(smeared) image than it takes place for a smaller Dm. Par- is also consistent with Bauch data (1956) who found, study-
ticular situation is observed for the largest interaural differ- ing the loudness of amplitude modulated signals, that the
ences Dm = ml  mr, close to 100%. In such condition, an loudness of those signals is independent of carrier
unmodulated signal (pure tone) was presented to one frequencies.
(right) ear (mr = 0%), and an amplitude modulated signal Besides the experiments which allowed measurements of
to the other (left) ear (ml = 100%). When the AM signal the binaurally perceived modulation depth, the theoretical
reached its minimal amplitude, the non-modulated (pure) approach to estimate its value and to indicate a role of bin-
tone masked the AM signal, reducing its modulation aural loudness interaction was also made and results have
depth. The result of this effect is seen in Fig. 2 in which been compared. It was found that the experimental and cal-
the perceived modulation depth, at ml = 100% and culated data are in a good agreement within the range of
mr = 0 is within the range 30–35%. The more complex Dm up to 40% and in the majority of cases up to about
trend of m = f(Dm) functions for large Dm may be caused 70%. However, some differences between these results
by the fact that temporal changes in amplitude of the appear for Dm above 70%. This implies that for larger
SAM signal are not linearly projected into the fibre dis- Dm values the perceived modulation may be influenced
charge rate. Physiological findings indicate that the fibres additionally by some other phenomena. One of them is
show an increasing discharge rate only in a limited dynamic the peripheral nonlinearity caused by the compressive
range (Joris and Yin, 1992). The pattern of amplitude property of the basilar membrane (Florentine et al., 1996;
changes in the fibres is also related to the level of their Buus et al., 1997; Zhang and Zeng, 1997; Carlyon and Dat-
spontaneous activity. The fibres with high levels of this ta, 1997; Moore et al., 1999). For AM signals, with the
activity reproduce amplitude modulation much worse than same average level but with large interaural difference in
those with low levels of spontaneous activity (Joris and modulation depths (large Dm) the compressive (nonlinear)
Yin, 1992). property, affects much more the peak amplitude of the
One has to bear in mind that the binaural signal process- stronger modulated signal than the weaker one since the
ing is not only peripheral but mainly of the central origin, larger the peak amplitude the larger the compression. In
based on the activity of the central auditory nervous system this way, the interaural difference in intensity diminishes
(CANS). Since CANS is less accessible to a direct study, so leading to a discrepancy between binaurally perceived
far less is known on the coding and binaural processing of modulation depth and calculated one for which the
132 E. Ozimek et al. / Hearing Research 235 (2008) 125–133

compression effect was not considered. It is also possible orders reaching the observation point with different time
that for large Dm the processing that takes place is more delays and the room‘s frequency response Taking into
complex than the simply loudness summation of AM sig- account dichotic listening conditions it was also reported
nals assumed in the calculations. Such a complex process- that the speech intelligibility scores in a room were related
ing has been implied by Wojtczak and Viemeister (1999). to the so-called modulation transfer function (Miyata
They investigated the thresholds for detection of low-rate et al., 1991). Our studies (Ozimek and Se˛k, 1996; Ozimek
sinusoidal amplitude modulation and for detection of and Rutkowski, 2001) showed that the AM index (m), mea-
intensity increments over a wide range of levels and found sured by use of a dummy head, differed in the left and right
a complex relationship between the psychometric functions ear i.e., changed over the interaural distance. The m
for modulation and increment detection. Perhaps a similar change, for low fm, reached sometimes even 30% and
relationship is also involved in our suprathreshold depen- depended on the modulation frequency and room reverber-
dence between the modulation and loudness. Some effect ation time. Such changes were greater than the perception
on the difference between the resultant modulation depth thresholds of the interaural differences in the modulation
calculated on the basis of Eq. (12) and that perceived in depth (Ozimek and Se˛k, 1996). They should be perceived
the experiment for large Dm probably results from the fact by the subjects and consequently may effect the speech
that Eq. (12) has been derived assuming the limited incre- intelligibility or the quality of music in a room. However,
ment of intensity (cf. 2a). However, for high Dm values, further study of a much wider range of stimuli and their
corresponding to considerable increments of intensity, this physical parameters, in different acoustic enclosures, is
assumption is not fully satisfied so the data obtained from needed to get practical application of the effects related
Eq. (12) can be charged with some error. It would be inter- to binaural perception of modulation depth in room
esting to take into account the quantity describing the neu- acoustics.
ral count rate proposed by McGill and Goldberg (1968)
instead of a limited increment of intensity. They assumed 5. Conclusions
that the discrimination of a change in intensity was deter-
mined by a statistically reliable change in the neural count – In dichotic presentation of AM stimuli, the perceived
rate. In this way the number of neural counts generated by modulation depth, for small interaural difference
the intensity increment could be directly related to the Dm < 40%, approximates the arithmetic mean of the
loudness increment. Hellman and Hellman (1990) using modulation presented to the left and right ear. For
such an approach found a good agreement between the higher values of Dm, the perceived modulation tends
measured and calculated loudness for different increments to be lower than the mean of ml and mr.
of intensity and for a wide range of sound levels. – As regards the calculation of the modulation depth
One should stress that the above mentioned calculations based on the assumption of loudness summation, the
were not aimed at modelling of the AM binaural percep- experimental and calculated value have been found in
tion but an analysis of the role of the loudness summation a good agreement for the interaural differences in mod-
in this perception. The calculations applied do not include ulation depth Dm less than about 70%. However, for lar-
the functions usually used in modelling of the binaural sys- ger Dm, some discrepancies between these values,
tem, such as: filtering in the auditory periphery, neural cod- roughly within the range of SD, appeared.
ing of the stimuli, comparison of interaural temporal – In the case of slow amplitude modulation, the binaural
information based on a correlation or coincidence mecha- loudness summation of the input AM signals plays the
nism, and a decision making mechanism (Colburn, 1995; basic role in the binaural perception of the modulation
Stern and Trahiotis, 1995). Surprisingly, under such a sim- depth.
plification of the processing applied, the perceived binaural
modulation depth seems to be in a reasonable agreement Acknowledgements
for measurements and calculations for low fm (except for
large Dm). This finding suggests that in the case of the slow This study was supported by the State Committee for
amplitude modulation, the binaural interaction in terms of Scientific Research (KBN), Grant No. 2 P05D 122 26
loudness summation of the input AM signals is one of and by Japanese Ministry of Education, Science and Cul-
important mechanisms in the perception of the modulation ture which financed the sabbatical year of the first author
depth. at the Research Institute of Electrical Communication, To-
It is worth stressing the practical aspect of the data hoku University, Sendai. We thank the anonymous review-
resulting from the present study on the binaural perception ers for helpful remarks on the earlier version of this paper.
of AM signals. This aspect is related to the speech intelligi-
bility in a room. Some authors showed that the binaural References
effects strongly depended on the positions of the speech
and the noise source in a room (Moncur and Dirk, 1967; Algom, D., Ben-Aharon, B., Cohen-Raz, L., 1989. Dichotic, diotic and
Nabelek and Robinson, 1982). These effects were caused monaural summation of loudness. A comprehensive analysis and
mainly by the superposition of reflections of different psychophysical functions. Percept. Psychophys. 46, 567–578.
E. Ozimek et al. / Hearing Research 235 (2008) 125–133 133

Altschuler, R.A., Bobbin, R.P., Clopton, B.C., Hoffman, D.W. Moncur, J.P., Dirk, D., 1967. Binaural and monaural speech intelligibility
(Eds.), 1991. Neurobiology of Hearing: The Central Auditory of connected discourse in reverberation as a function of azimuth of a
System. Raven Press, Ltd., New York. single competing sound source (speech or noise). Acustica 34, 200–208.
Atkin, L.M., Irvine, D.R., Webster, W.R., 1984. Central neural mecha- Moore, B.C.J., Wojtczak, M., Vickers, D.A., 1996. Effect of loudness
nism of hearing. In: Darian-Smith, I. (Ed.), Handbook of Physiology – recruitment on the perception of amplitude modulation. J. Acoust.
The Nervous System III. Am. Physiol. Soc., Washington, DC, pp. Soc. Am. 100, 481–489.
675–737. Moore, B.C.J., Vickers, A., Baer, T., Launer, S., 1999. Factors affecting
Bauch, H., 1956. Die Bedeutung der Frequenzgruppe für die Lautheit von the loudness of modulated sounds. J. Acoust. Soc. Am. 105, 2757–
Klängen. Acustica 6, 40–45. 2772.
Blauert, J., 1983. Spatial Hearing. MIT Press, Cambridge, MA. Moore, B.C.J., Glasberg, B.R., 2007. Modeling binaural loudness. J.
Bregman, A.S., 1990. Auditory Scene Analysis: The Perceptual Organi- Acoust. Soc. Am. 121, 1604–1612.
zation of Sound. MIT Press, Cambridge, MA. Nabelek, A.K., Robinson, P.K., 1982. Monaural and binaural speech
Buus, S., Florentine, M., Poulsen, T., 1997. Temporal integration of perception in reverberation for listeners of various ages. J. Acoust. Soc.
loudness, loudness discrimination, and the form of the loudness Am. 71, 1242–1248.
function. J. Acoust. Soc. Am. 101, 669–680. Ozimek, E., Rutkowski, L., 2001. Changes in frequency of FM signals
Buus, S., Müsch, H., Florentine, M., 1998. On loudness at threshold. J. propagating in a room. In: 17th International Congress on Acoustics,
Acoust. Soc. Am. 104, 399–410. Rome, September 2–7, 3P.49, p. 53.
Carlyon, R.P., Datta, A.J., 1997. Excitation produced by Schroeder phase Ozimek, E., Se˛k, A., 1996. AM and FM difference limens and their
complexes: evidence for fast-acting compression in the auditory reference to amplitude-frequency changes of a sound in a room. Acta
system. J. Acoust. Soc. Am. 101, 3636–3647. Acust. 82, 114–122.
Colburn, H.S., 1995. Computational Models of Binaural Processing. In: Ozimek, E., Sone, T., 1997. Binaural perception of the modulation depth
Hawkins, H.L., McMullen, T.A., Popper, A.N., Fay, R. (Eds.), of AM signals. In: Preis, A., Hornowski, T. (Eds.), Proceedings of the
Auditory Computation. Springer Verlag, pp. 332–400. XIII Annual Meeting of the Inter. Soc. for Psychop. Poznań, Poland,
Florentine, M., Buus, S., Poulsen, T., 1996. Temporal integration of pp. 73–78.
loudness as a function of level. J. Acoust. Soc. Am. 99, 1633–1644. Phillips, D.P., Reale, R.A., Brugge, J.F., 1991. Stimulus processing in the
Green, D., 1993. Auditory intensity discrimination. In: Yost, W., Popper, auditory cortex. In: Altschuler, R.A. et al. (Eds.), Neurobiology of
A., Fay, R. (Eds.), Human Psychophysics. Springer Verlag, pp. 13–55. Hearing: The Central Auditory System. Raven Press, Ltd., New York,
Grimm, G., Hohmann, V., Verhey, J., 2002. Loudness of fluctuating pp. 335–365.
sounds. Acta Acustica united with Acustica 88, 359–368. Plack, C.J., Carlyon, R.P., 1995. Loudness perception and intensity
Hartmann, W.M., 1997. Signals, Sound, and Sensation. American coding. In: Moore, B.C. (Ed.), Hearing. Academic Press, pp. 123–160.
Institute of Physics, Woodbury, New York. Reynolds, G.S., Stevens, S.S., 1960. Binaural summation of loudness. J.
Hellman, W.S., Hellman, R.P., 1990. Intensity discrimination as the Acoust. Soc. Am. 32, 1337–1344.
driving force for loudness. Application to pure tones in quiet. J. Scharf, B., Fishken, D., 1970. Binaural summation of loudness: Recon-
Acoust. Soc. Am. 87, 1255–1265. sidered. J. Exp. Psychol. 86, 374–379.
Hellman, R.P., Zwislocki, J.J., 1963. Monaural loudness function at Schlauch, R.S., 1994. Intensity resolution and loudness in high-pass noise.
1000 cps, and interaural summation. J. Acoust. Soc. Am. 35, 856–865. J. Acoust. Soc. Am. 95, 2171–2179.
Houtgast, T., Steeneken, H.J., Plomp, R., 1980. Predicting speech Schwartz, I.R., 1992. The superior olivary complex and lateral lemniscal
intelligibility in rooms from the modulation transfer function. I. nuclei. In: Webster, D.B., Popper, A.N., Fay, R.R. (Eds.), The
General room acoustics. Acustica 46, 60–72. Mammalian Auditory Pathway: Neuroanatomy. Springer-Verlag,
Houtgast, T., Steeneken, H.J., 1985. Review of the MTF concept in room New York, pp. 117–167.
acoustics and its use for estimating speech intelligibility in auditoria. J. Stern, R.M., Trahiotis, C., 1995. Models of binaural interaction. In:
Acoust. Soc. Am. 77, 1061–1077. Moore, B.C. (Ed.), Hearing. Academic Press, pp. 347–386.
Irvin, D.R.F., 1992. Physiology of the auditory brainstem. In: Popper, Suga, N., 1992. Philosophy and stimulus design for neuroethology of
A.N., Fay, R.R. (Eds.), The Mammalian Auditory Pathway: Neuro- complex-sound processing. In: Viemeister N.F. (Ed.), 1979. Temporal
physiology. Springer-Verlag, New York, pp. 153–231. Modulation Transfer Functions Based Upon Modulation Thresholds.
Joris, P.X., Yin, C.T., 1992. Responses to amplitude modulated tones in J. Acoust. Soc. Am. 66, pp. 1364–1380.
the auditory nerve of the cat. J. Acoust. Soc. Am. 91, 215–232. Viemeister, N.F., 1979. Temporal modulation transfer functions based
Kitzes, L.M., Wrege, K.S., Cassady, J.M., 1980. Patterns of responses of upon modulation thresholds. J. Acoust. Soc. Am. 66, 1364–1380.
cortical cells to binaural stimulation. J. Comp. Neurol. 192, 455–472. Wang, X., Sachs, M.B., 1992. Coding of envelope modulation in the
Kuwada, S., Yin, T.C., 1983. Binaural interaction in low frequency auditory nerve and anteroventral cochlear nucleus. In: Carlyon, R.P.,
neurons in inferior colliculus of the cat. I. Effects of long interaural Darwin, C.J., Russell, I.J. (Eds.), Processing of Complex Sounds by
delays, intensity and repetition rate on interaural delay function. J. the Auditory System. Clarendon Press, Oxford, pp. 105–108.
Neurophysiol. 50, 981–999. Wojtczak, M., Viemeister, N.F., 1999. Intensity discrimination and
Kuwada, S., Batra, R., Fitzpatrick, D.C., 1997. Neural processing of detection of amplitude modulation. J. Acoust. Soc. Am. 106, 1917–
binaural temporal cues. In: Gilkey, R.H., Anderson, T.R. (Eds.), 1924.
Binaural and Spatial Hearing in Real and Virtual Environments. LAE Yin, T.C., Kuwada, S., 1984. Neuronal mechanisms of binaural interac-
Publ., Mahwah, New Jersey, pp. 399–426. tional. In: Edelman, G.M., Gal, W.E., Cowan, W.M. (Eds.), Dynamic
Marks, L.E., 1978. Binaural summation of the loudness of pure tones. J. Aspects of Neocortical Function. Wiley, New York, pp. 263–313.
Acoust. Soc. Am. 64, 107–113. Zhang, C., Zeng, F., 1997. Loudness of dynamic stimulus in acoustic and
McGill, W.J., Goldberg, J.P., 1968. A study of the near-miss involving electric hearing. J. Acoust. Soc. Am. 102, 2925–2934.
Weber’s Law and pure-tone intensity discrimination. Percept. Psycho- Zwicker, E., Fastl, H., 1999. Psychoacoustics Facts and Models. Springer.
phys. 4, 105–109. Zwicker, E., Zwicker, U.T., 1991. Dependence of binaural loudness
Miyata, H., Nomura, H., Houtgast, T., 1991. Speech intelligibility and summation on interaural level differences, spectral distribution, and
subjective MTF under diotic and dichotic listening conditions in temporal distribution. J. Acoust. Soc. Am. 89, 756–764.
reverberant sound fields. Acustica 73, 200–207.

You might also like