Zoom out Search Issue

ManualsBrandsContents Manualsaudio & home theatreZoom in

111

112

113

114

115

116

117

118

119

120

IEEE SIGNAL PROCESSING MAGAZINE [112] MARCH 2015

pie chart suggests that 61% of the subjects preferred the natu-

ral sound rendering, while only 33% preferred the conven-

tional stereo rendering.

To sum up the subjective test results, we found that the

natural sound rendering system using the various signal pro-

cessing techniques explained in this article enhances the lis-

tening experience compared to a conventional stereo system.

Additionally, the presence of head tracking in the system will

only improve the natural sound

rendering as observed in several

studies [10].

CONCLUSIONS AND

FUTURE TRENDS

With the advent of low cost, low

power, small form factor, and high-

speed multicore embedded proces-

sor, we can now implement the

aforementioned signal processing

techniques in real time and embed

processors into the headphone design. However, various imple-

mentation issues regarding the computation cost of sound

scene decomposition, HRTF/BRIR filtering in virtualization,

and equalization as well as the latency in head tracking should

be carefully considered. One example of such a natural sound

rendering system is the four-emitter 3-D audio headphone [39]

developed at the Digital Signal Processing Lab at NTU. This

system has been psychophysically validated and found to per-

form much better than the conventional stereo headphone

playback system.

Besides the five types of techniques discussed in this article,

there have been other efforts to enhance the natural experience

of headphone listening. To enable the natural pass through of

the sound from outside world without coloration, headphones

can be designed with suitable acoustically transparent materi-

als. When this is not effective, microphones integrated into

headphones and associated signal processing techniques, such

as equalization, and active noise control, are employed. The

headphones with built-in microphones open a new dimension

to augment the listening experience with the physical world.

The future of headphones for assistive listening applications

would be where listeners cannot dif-

ferentiate between the virtual acous-

tic space created from headphone

playback and the real acoustic space.

This would require a combined effort

from the whole audio community—

from the headphone manufacturers

and sound engineers to audio scien-

tists. More information about the

content production has to be distrib-

uted from the content developers to

the end user to enhance the extrac-

tion process. Moreover, obtaining and exploiting every individual’s

anthropometrical feature or hearing profile is crucial for a natural

listening experience. Finally, with more sensors, such as global

positioning systems, gyroscopes, and microphones that can be

integrated into headphones, future headphones are becoming

more content-, location-, and listener-aware, and hence more

intelligent and assistive.

ACKNOWLEDGMENTS

This work is supported by the Singapore National Research

Foundation Proof-of-Concept program under grant NRF 2011

NRF-POC001-033. We thank the guest editors and reviewers for

their constructive comments and suggestions.

1 2 3 4

100

MOS for Four Measures

Measure

Mean Opinion Score

Natural Sound Rendering

Conventional Stereo

0 50 100

100

Scatter Plot for All Scores

Conventional Stereo

(a)

(b) (c)

Natural Sound Rendering

61%

33%

Preference of the Tracks

Prefer Natural Sound Rendering

Not Sure

Prefer Conventional Stereo

[FIG6] Results of the subjective experiments: (a) MOS, (b) scatter plot, and (c) overall preference.

THE FUTURE OF HEADPHONES

FOR ASSISTIVE LISTENING

APPLICATIONS WOULD BE WHERE

LISTENERS CANNOT DIFFERENTIATE

BETWEEN THE VIRTUAL

ACOUSTIC SPACE CREATED

FROM HEADPHONE PLAYBACK

AND THE REAL ACOUSTIC SPACE.

THE WORLD’S NEWSSTAND

THE WORLD’S NEWSSTAND