Apple ambitions, Shenzhen budget: the speaker that forced a redesign
On a consumer AI voice toy, getting enough loudness without distortion, without blowing the budget, and within an acceptable size, has no easy answer. We ended up changing the amplifier IC, changing the mechanical design, changing the speaker, and going to an external-mounted speaker. The detour in the middle is the part worth reading.
The trap starts with a nice enclosure
The first product's industrial design took its cue from an AirPods charging case. Small, refined, good in the hand.
Then we took it outdoors and you had to hold it against your ear to make out what it said.
On a voice toy, speech is the core function. If it is not loud enough, half the product is already broken. That was obvious quickly. Fixing it was not.
Loudness against size is physics, not effort
Speaker acoustics set the ceiling directly. A larger diaphragm moves more air, which produces higher sound pressure, which is louder. A physically smaller speaker is therefore harder to make loud, and no amount of engineering removes that relationship.
This is not about capability. Apple applies better engineering and a far higher cost density than we can, and an AirPods case still does not produce speaker-grade volume.
Can the amplifier fix it
The first instinct was to keep the enclosure and push the volume up with the amplifier IC. The direction is not wrong, and two things get in the way.
First, a small speaker will not absorb brute-force power. Drive too much into a small diaphragm and it distorts and clips. Louder, worse, and the user experience drops further than it did from being too quiet.
Second, good amplifier ICs are expensive and not small. Phone-class parts perform, and both the price and the PCB area are a burden on a consumer toy. A respin is not cheap either, and chasing loudness plus low distortion inside the same enclosure keeps pushing you toward more expensive parts.
The phrase I used at the time was that we had Apple's ambitions and a small Shenzhen company's resources. That is what it felt like.
How we decided
We ranked "refined enclosure" against "audible in real use". The first requirement of a voice toy is that the user can hear what the AI is saying. If they have to bring it to their ear every time, the product has failed on its own terms.
So: new amplifier IC, new mechanical design, new speaker, external-mounted speaker solution. The device got larger. It also became audible, clean and undistorted at two metres outdoors.
The pattern this belongs to
Enclosure versus core function is a common fight in consumer electronics, and there is only one way to settle it. Work out what the user is buying the product for, then decide which of the two you will not compromise.
In a toy, whether it is loud enough decides whether the child keeps using it. That is the one you do not trade. Industrial design can improve across versions, and when resources are tight, get the core right first.
Related
Want to try the hardware behind these notes?
Six AI hardware platforms are in mass production. Tell us your market and channel and we send an evaluation sample.
Request a sample