It would be interesting to see the gain curve of this speaker.
My (limited) experience with piezoelectric speakers is that they resonate at a single frequency very loudly, and are practically silent outside the resonant peak. Perfect for a microwave beeper, but never going to produce audible speech.
The "high-quality" descriptor makes it sound like they have produced a reasonably flat gain curve, which seems really significant! But without any explanation I'm skeptical.
It may even be as silly as using an unintuitive technical definition of "quality" - in a second order linear system, the "quality" of the gain curve is the ratio of the amplitude at the peak to the input amplitude... The exact opposite of what a reasonable person would consider high-quality sound.
Horn loaded piezos are common in cheap PA systems (examples: https://www.parts-express.com/search?order=relevance:desc&ke...). They typically work from about 3khz on up, which does involve the upper end of voice. Fidelity isn't great but the things are loud and hard to break for how cheap they are.
No one in acoustics calls Q "quality" really. It's just "q" or people talk about underdamped vs overdamped, etc. If quality comes up, it's usually in the context of lower q designs being higher fidelity (eg, a subwoofer that's ~0.707 vs one that's say 1.2).
What's most interesting about this new transducer is that it's physically thin, but acts as a monopole driver. That's cool and unique.
They'll have to make some variations to dial in just how the geometry affects the response.
Also, this driver as is would beam significant when made as large as some of the examples they talk about. But that's probably ok as the output is low enough you'd want to be in the near field anyhow.
In the article, they mention that a 25V signal at 1kHz produced 66dBA, while the same 25V signal at 10kHz produced 86dBA. That suggests the curve is not very flat.
We've seen plenty of examples of people using interesting materials to visualize harmonics on surfaces. Everything from rice to gloop.
A given surface can have many resonant frequencies, where volume pitches upward dramatically. That's a lot more than just beeps, but a good deal less than human speech or music. With enough separate speakers you might be able to manage, but old hi-fi sets had 2-3 speakers per channel at most, and you'd probably need many more than that. At some point you'd start to wonder if a phased array were a better option.
In speaker transducers, normally you're trying to repress any bending modes of the piston. These produce diffractive interference effects, or in some materials "ringing" modes like a bell that are very hard to filter away. The industry standard way for measuring and optimizing this stuff is a laser measuring rig from Kippel. You can find Kippel data for most drivers on the market.
My (limited) experience with piezoelectric speakers is that they resonate at a single frequency very loudly, and are practically silent outside the resonant peak. Perfect for a microwave beeper, but never going to produce audible speech.
The "high-quality" descriptor makes it sound like they have produced a reasonably flat gain curve, which seems really significant! But without any explanation I'm skeptical.
It may even be as silly as using an unintuitive technical definition of "quality" - in a second order linear system, the "quality" of the gain curve is the ratio of the amplitude at the peak to the input amplitude... The exact opposite of what a reasonable person would consider high-quality sound.