Zoom out Search Issue

ManualsBrandsContents Manualsaudio & home theatreZoom in

111

112

113

114

115

116

117

118

119

120

IEEE SIGNAL PROCESSING MAGAZINE [110] MARCH 2015

experiments [4], [38] showed that listeners prefer other alter-

native target responses more than the conventional FF and

DF equalizations. Examples of these preferred target curves

include RR_G and RR1_G proposed by Olive et al. [4] based

on the impulse response of the loudspeaker system in the

Harman Reference rooms.

Ideally, the best reference field that preserves the true quality

of the recording would be the field where the recording is carried

out. Furthermore, the choice of

headphones can also greatly affect

the transparency of the binaural ren-

dering even with the correct head-

phone equalization. The external ear

is unhindered in the natural listen-

ing conditions, where the sound

pressures at the ear are governed by

free-air characteristics. With head-

phones placed over the ear, the pres-

sure characteristics of the sound arriving at the eardrum are

greatly affected compared to the free-air characteristics due to the

interaction between the external ear and the headphone enclo-

sure. The closer the coupling characteristic of the headphones

with that of the free-air, the more accurate and transparent is the

reproduced sound. Møller [35] defined the effect of the headphone

for a binaural recording at the blocked ear canal in terms of the

electrical transmission gain, G:

MPTF HPTF

PDR

(7)

where MPTF is the transfer function of the recording microphone,

and PDR is the pressure division ratio. PDR is defined as the ratio of

the equivalent Thévenin impedances when the ear is in free-air to the

case when the headphone is placed on the ear, and is given as [35]:

PDR

earcanalradiation

earcanalheadphones

(8)

where

earcanal

and Z

headphones

are the input impedances of the

ear canal and the impedance of the headphone, respectively;

radiation

is the free-air radiation impedance as seen from the ear

canal. The PDR reduces to unity when the pressures in the free-

air and with headphones become equal. Such headphones are

defined as FEC (free-air equivalent coupling) headphones,

which are also sometimes called open headphones [35]. Open

headphones are different from the commercially available

“open–back headphones.” Most of

the commercially available head-

phones have less than ideal FEC

characteristics [35]. It is important

to note that the FEC condition for

the headphone is necessary only for

binaural recordings made at the

blocked ear canal, which is also the

most common technique for indi-

vidualized binaural recording [35].

In such a case, headphone equalization alone is sufficient to

achieve auralization transparency. To summarize, equalization

(both recording and playback) and individualization play a criti-

cal role in the natural rendering of sound of any formats (binau-

ral or stereo) over headphones.

INTEGRATION OF NATURAL SOUND

RENDERING TECHNIQUES

An integration of these signal processing techniques for natural

sound rendering reviewed in this article is depicted in Figure 5.

The original sound sources along with their environmental

information are represented as a sound mixture after the mix-

ing process. The sound scenes from the mix are then decom-

posed into primary components (sources) and/or ambient

components (environment) using BSS and/or PAE. The

extracted primary components, which are basically directional

sound sources as perceived by the listener, can be rendered

using (individualized) HRTFs [1]. Ambient components

[FIG5] The natural sound rendering system for headphones: an integration of all the signal processing techniques reviewed in

this article.

Head

Movement

Sound

Sources

Individual

Parameters

Sound

Environment

Sound

Mixing

Decomposition

Using BSS/PAE

Source

Rendering

Environment

Rendering

Equalization

Individualization

Head

Tracking

Virtualization

Headphone

Playback

Rendering of Natural Sound

Sound

Mixture

BY PROJECTING THE SOUND

FROM THE FRONT, THE

IDIOSYNCRATIC FRONTAL

PINNA SPECTRAL CUES OF

THE LISTENER ARE

CAPTURED INHERENTLY

DURING THE PLAYBACK.

THE WORLD’S NEWSSTAND

THE WORLD’S NEWSSTAND