increases the jaw accuracy
by using a simple five LED
VU meter board. Dialog
audio is fed into the LED
VU meter board and the
Arduino is wired to monitor
the middle three LEDs.
By tapping into these
middle three LEDs, the
Arduino is able to “read”
the volume of the signal
and then tell the servo to
move to a correlating
position (Figure 7).
This allows the
Jawduino to react more
accurately to audio volume
levels and is a definite
improvement over the
Scary Terry design.
However, all of the simple
“volume to position”
solutions miss an important
aspect of recorded speech — one that has a direct bearing
on jaw sync accuracy.
What’s the Frequency,
Kenneth?
Spoken audio consists of a spectrum of frequencies.
There are differing amounts of energy devoted to different
bands depending on the phonemes that compose a given
word. A good example of the differences in energy is
shown in the spectrogram of the phrase “She sells sea
shells” (Figure 8).
You can see that if we limit ourselves to using
amplitude as the only criteria to determine jaw position,
The ideal solution would be to sample the audio
signal across its spectrum and collect information about
each of the frequencies that have a significant impact on
jaw position; then, use this information to direct jaw
motions.
Can Wee Talk?
Steve Bjork from Haunt Hackers (see Resources) and
the creator of the Wee Little Talker board, decided to
tackle the jaw sync problem using a combination of
hardware and software. On the hardware side, he started
with the MSGEQ7 seven band digitally controlled
spectrum analyzer chip (Figure 9). This small eight-pin
September 2017 41
■ FIGURE 8. Sounds such as "sh" and "s" (sometimes
called unvoiced sounds) contain a broader range of
higher frequencies, while other (voiced) sounds —
such as "e" or "ell" — contain mostly lower
frequencies (image source
www.ele.uri.edu/~hansenj/projects).
■ FIGURE 9. The MSGEQ7
chip spectrum analyzer
chip.
Talking Skull Control Board Feature Comparison Matrix
Key: “;” = yes, “ ” = no, “ “ = N/A
Scary Terry
Board Jawduino Wee Little Talker
Uses multi-frequency audio sampling for jaw sync ✗✗ ✓
Setup can be programmed without an external computer ✗ ✓
Can be triggered by PIR or pushbutton switch ✓
Adjustable “hold off” time to prevent retriggering ✓
Embedded voice feedback menu system ✓
Embedded audio player ✓
Embedded audio amplifier ✓
Plays “ambient” sound file between shows ✓
Non-volatile AGC settings storage ✓
Haunt
Hackers
Steve Bjork started
Haunt Hackers as a
site to promote
DIY boards for the
hobbyist. This
differentiates these
boards from his
professional line
that he sells to
commercial haunts
and theme parks.