SYNTHETIC VOWELS
served’’ amplitudes of the several components of the curve.
The set of pipes is now placed in front of the phonodeik
in the position occupied by the original voice, and each
pipe corresponding to a component tone is separately ad¬
justed, till it shows in the phonodeik the amplitude required
by the analysis for this partial; the adjustment is readily
verified as the amplitude is directly measurable on the
ground glass of the camera. The reproduction is wholly
independent of the peculiarities of the phonodeik, for it is
made with the same instrument and under the same condi¬
tions as was the original record.
The fundamental pitch is set from a piano or tuning fork,
and the other partials can then be tuned to the exact har¬
monic ratios by means of the phonodeik. The pipe for the
second partial, already in approximate tune, is sounded
simultaneously with the fundamental, the resulting curve
is observed on the ground glass of the phonodeik, and the
pipe is tuned until the wave form remains constant. The
tuning is now necessarily perfect, since inexact relationship
produces a slowly changing curve. The third and other
partials are then successively tuned in the same manner.
This method of tuning is perhaps the best possible for two
or more frequencies which are in exact ratios, since it pos¬
sesses the advantages of Lissajous’s optical method and is
more generally applicable. While it is not difficult to ad¬
just a small number of pipes to practically perfect harmonic
frequencies, it is hardly possible to tune sixteen pipes so that
the resultant wave form remains unchanged, and conse¬
quently the synthesized tones do not blend as perfectly as
do the partials from a single source. With care, however,
the tuning is sufficiently exact to secure success in the ex¬
periments.
249