images
of dummy head built for binaural streaming and recording
NEW! samples of binaural
recording: dj.wav grace.wav
(must be heard through headphones!)
Introduction
When performing with
a spatialized accompaniment (whether it is live or prerecorded), a
musician can receive only stereo information at best—most have never
heard what their spatialized performance “really” sounds like to someone
sitting in the audience. This poses a problem when one wishes to devise
a system where the performer is control of their own spatialization
patterns during performance; a performer needs feedback to ensure full
artistic control. My Masters thesis work at CCRMA involves the
development of a system that makes a binaural recording in the audience
space, and streams the signal directly to the performer, providing what
I call "live spatialized feedback."
Background
As a composer I am interested in space as a parameter of sound. As a
researcher I want to enable musicians to control the location of their
sound in the concert hall just like they can control pitch, intensity,
and timbre. This interest comes partly from
musicological research
that I conducted in Spring 2003, in which determined the importance
of distinct, implied sound sources for clearness of aesthetic statement
in electroacoustic music.
My group's project
for Music 250a in Fall 2003 involved the development of a device
that converts a flutist's unconscious physical gestures into
spatialization mappings. The resulting spatialization of the
performance wasn't as convincing as we expected. Physical gestures vary
greatly between performers, and it is hard to determine whether or not
a seasoned performer was using deliberate or unconscious motion. Thus,
I have decided to make future systems deliberately controlled by the
performer.
My experience in performing spatialized works dictates that the direct
sound from the instrument often overpowers any reverberent sound (or
sound heard through headphones) because it is louder and is the first
sound to reach the performer’s ears. Thus, the proper sound-delivering
device attenuates as much of the direct sound as possible while feeding
the signal from the dummy head with the lowest latency. In research I
conducted in Spring 2004 I found that a mixture of passive attenuation
(such as that found in traditional ear protectors) and active
attenuation (found in sound-cancelling devices) provided the most
attentuation of the direct signal.
SPF 8 Schematic
Schematic of SPF 8. Note that the
performer and audience can be separated between Room A and Room B, the
audience receiving the performer's signal through a networked audio
connection. (Please visit my homepage to see my past work in networked audio.)
The dummy head
Shell
I built a dummy head for binaural recording using alginate molds of a
real human's head, rather than relying on measurements. Alginate is a
non-toxic substance similar to latex which can be poured on human skin
to create a highly detailed negative. I created four negatives, for the
face, back of head, and two ears. I filled the negatives with casting
wax, and melted the negatives together with various heating implements
to create a delicate wax shell. I coated the inside of the shell with
several layers of plaster. Ear canals were created by placing drinking
straws where the ear canals would be and covering them with plaster.
interior of the dummy head
Microphone Circuits
A schematic of the microphone circuits is shown below. Each mic is
individually powered, and sends a mono signal to a preamp; the two
signals are combined to make a stereo signal which can be listened to
through headphones. The resistance and capacitance values were chosen
to create a high-pass filter with a cutoff frequency of approximately
1/RC = 30 Hz.
circuit diagram for each microphone
Binaural Recording
The dummy head is
needed for binaural recording to provide filtering of the microphone
signal similar to the filtering provided by the human head for the
eardrum. A recording made with a pair of microphones placed eight
inches apart will have limited spatial information. Certain geometries
present in the head, such as the pinnae, are essential for
differentiating between signals coming from in front and behind the
head.
...
Still to come
I designed and built the dummy and microphone circuits during the
Summer and Autumn Quarters of 2004. In Winter Quarter I plan to conduct
psychoacoustic tests to determine the effectiveness of the dummy head
in providing spatialized signals, and then modify the setup according
to these results. Then I should be able to stream the binaural signal
to a performer's sound-cancelling headset, providing an image of their
sound as heard by the audience. A report summarizing the design,
testing, and results will be written in Spring Quarter.
Planned Applications of SPF 8
