Listening to music doesn't drive a constant 22 kHz tone.
[Edit] I really don't understand -- why was I downvoted here? Can someone explain why they think I'm wrong? Isn't a constant, high-frequency tone going to draw more power than variable output with significant larger gaps (and a likely lower peak voltage difference?)
"To estimate the impact of energy harvesting on the phone, we modifying an iPod Touch to measure battery current and voltage. We play an audio file that generates the required 22 kHz tone and find that the iPod draws 91 mA at 4.04 V when the harvester is attached and 37 mA at 4.04 V when the harvester is not attached.
These figures illustrate the substantial inefficiency of this approach compared with directly supplying power to a peripheral. We note, however, that the power output of an iPod is higher than an iPhone, making the numbers appear worse than they are on the iPhone."
Speaking of figures, I wonder how this would compare to what can be gathered from ambient radiowaves with a device of a similar size.
I remember a posting from somewhere where Nokia claimed to reach in the miliwatt range only using ambient radiowaves. Can't seem to find it again though.
Currently, I am a bit doubtful concerning whether drawing extra power from an already very battery lifetime-constrained smartphone is a good idea.
On the other hand, the batteries shold improve in the future, so with a bit of luck the idea may just as well come to maturity at the right time!