This picture shows our SMA loudspeaker, which consists of a light plastic cup suspended between a 0.1mm SMA (Flexinol) wire, attached to the back of the cup (and almost invisible in the photo), and a pair of elastic bands, a piece of string and a small weight, attached to the front. The weight ensures an approximately constant tension on the SMA wire; and the cup does not touch the wooden frame below it.
The SMA wire is fed a heating current consisting of an audio signal and a DC bias that is just sufficient to ensure that the current is always positive. The result is a quiet but plainly audible sound. In particular, frequencies well above 1 kHz can be heard.
Here is an example audio output.
The point of this experiment is to demonstrate that SMA is capable of responding to a change in heating current in a fraction of a millisecond. Note that the SMA wire is suspended in approximately still air. There is no air movement near the wire other than that caused by normal office air conditioning.
How do we know it really is the shape memory effect causing the sound? See the evidence.