Some time ago I provide conclusively I knew nothing about audio when I got some help designing and redesigning my living room setup; my thanks again to all who helped, I use it regularly and it all just works.
I have a new and quite different problem today, and could use some acoustic advice. The subject is NOT music, it is voice, specifically voice assistants, the "Hey Computer, turn on the lights" type for Home Assistant. I want to replace my google speakers (at least in the sense of integration with my home) before Gemini makes a mess of them, and have most of the system side done, but the actual mic and speaker you interact with is giving me a fit.
The ready made ones have several flaws (subject for another day).
Seeed Studio has one where the mic array is fairly sophisticated (based on XVF3800, four mic's), but it comes without a speaker. It has a 5w (max) speaker output. It's about twice the size of a hockey puck but very light and thinner (about 30% thicker than the black part you can see in the first photo).
(See photo): The white thing is a 1st draft of a 3d print for a case. I need to figure out cabling routes, a slot for a mute button on one side (that right now is holding it a bit cockeyed), and color (I thought the white/black contrast might be good but having 2nd and 3rd thoughts).
My question is about the speaker, and its compartment.
This speaker is an amazon $6 special. I also have ordered a $16 fancy version to see if it makes more difference (bigger driver, much larger magnets). Remember, limited to 5 watts, but it's not music, it's voice quality I'm looking for, so a much more limited frequency range. (Yes, it can play music, I just won't -- I have speakers for that elsewhere).
You can see the shape. This speaker actually has a gasket around the edge and it mates to the whole, and I'll melt the little posts to hold it down, so it will be air tight at the bottom. The top is just a snug fit of the mic assembly against the sides, it is not air tight but it is a pretty snug fit.
The material is 6mm thick PLA with typical 3D print infill.
The cone on the bottom is just an invention to try to spread the sound (remember, no acoustic knowledge so there's no 'design" here, just a cone that fit the space).
It doesn't sound bad in a brief try, it's far louder than it was sitting loose, and so far no buzzing or vibration though I have not done enough to have exercised the whole frequency range.
General thoughts on what I might do better?
Should the hollow area behind the speaker be filled with foam or something, or open?
I read in some postings to make the surface irregular. Well, 3D prints at the micro level are quite irregular, and I left the posts exposed to break it up a bit, should I print some kind of baffles on the side? (it's all a single print from bottom up so vertical structures are easy, horizontal ones not so much).
Cone good or bad idea? I want the sound more or less omnidirectional as it will often be on the side of a room and may be spoken to from at least 180 degrees, maybe a bit more.
Does how thick the walls are really matter, it only needs about half that thickness for strength, I was trying to dampen vibration with that sort-of spongy infill pattern over 6mm.
First draft so it's a good time to change, still working on the hardware itself a bit, need to pick a specific driver and decide on rPi vs esp32 based computer inside (if rPi will need a bigger cavity, or separate cavity for the pi zero, but expect the esp32 to work if it ever gets delivered).
And I guess anything to look for in a speaker. Remember -- under-powered (5w), relatively small, voice only. There's no point in a super quality speaker with computer generated voice and very under-powered electronics. And it's not for music, if it was I'd take the audio out to a powered speaker (or 2).
So... suggestions?
View attachment 80100
View attachment 80101
View attachment 80102
