Although the overall displacement of a magnetostrictive core is very small (roughly 30 parts per million of its length, for nickel), the force of this displacement is very large, being limited only by the tensile strength of the core material. This is a poor combination for generating sound, since moving air requires a weak force but a large displacement, and as a result my earphone was somewhat inefficient. I decided to correct this by designing a lever mechanism to amplify the displacement of the core, and the result can be seen above.
The mechanism is simple; the nickel core contracts in the presence of a magnetic field (generated by the coil), which causes it to pull on the wooden tabs on the back of the diaphragm. These bend the center strip of the diaphragm outward, and since the diaphragm is much longer than the tabs, the displacement is amplified as a result. With the help of a small transistor amplifier (shown below), it can produce enough volume to be audible in a quiet room, and the sound quality is acceptable, comparable to that of a portable radio.
In powering this speaker, the one key factor is the field strength of the coil. To achieve the loudest possible volume without introducing distortion, the field should be strong enough that the audio modulation does not cause clipping, but not so strong as to saturate the core (at which point the audio would become quiet). For a nickel core, I found the optimal field strength to be around 70 amp-turns per inch. Since each inch of coil windings contains 140 turns (five layers of 25-30 turns per inch), this means that the coil requires half an amp of current (before modulation) to achieve the widest range of volume. This comes at the expense of audio quality however, since stronger fields produce a more non-linear magnetostrictive response.
Overall this speaker performs adequately, and its compact shape makes it a worthwhile design; however, it is by no means high-fidelity. The bass response is limited by the small diaphragm displacement, and at higher frequencies the resonance of the wood panel begins to interfere. Of all other types of speakers it is most similar to a piezoelectric disc, both in its merits and its faults. I have recorded a demonstration of the speaker in action, and the video of this can be seen below: