DIY Headphone Adventure

Hi first time poster here so please excuse some roughness.

In the last ~1,5 years I have built 3 headphones so far with rising challenge each time and i thought it might be interesting for other people to see aswell as give some ideas and feedback for my projects going forward.
I will post all three with my build process in chronological order in this thread.

This journey started with me looking for a good closed back headphone for a fair price that was also readily available here in Germany, but I became frustrated with the lack of actually good options. Then after seeing a comment on Resolves review on the ADX5000 which said that the whole driver assembly was available for a fair price I thought why not just build a headphone yourself. Starting open back and then advancing to closed backs. And thus my DIY ADX5000 was born.

For this the Build was fairly straight forward. I bought the replacement drivers some connectors for the already built in A2DC ports and then a headband on Aliexpress. As Body for cups I used a piece of Walnut which I turned to a pair of rings via woodturning.
I began with a thick circular piece and screwed it into the metal plate holding it onto the lathe. Then i turned first the outer part to get a nice curve and a diameter of about 110mm. Then I hollowed the inside out to get to an inner diamater of a bit more than 100mm.


After some sanding on the lathe I cut out the inner part, drilled 2 3mm holes in the sides for the headband and added the 6mm hole for the connector at a slight angle.

After a quick testfit I predrilled with the drivers in the cups the holes for some 2,5mm woodscrews to hold the assembly into the rings.
To protect the wood and add flair I gave the rings 3 passes with wood oil and lastly added a honeycomb aluminium grid in the back. For added comfort I replaced the knock off HD600 headband padding with a strip of soft brown leather as a comfort strap.

The tuning was done solely via the pad choice. I experimented with slight front dampening but was always unhappy with how transients and details felt after that so ended up just buying about 10 pairs of random pads and just listening. I began with a pair of leather angled pads with inner perforations. With those it sounded too sharp and unpleasant. I then turned to Brainwavz XL Suede with the Cooling gel. The Gel ones have a slightly smoother material which I simply found more comfortable. Aswell due to the Gel they begin being more Firm but mold to your head after a while.


After adding catears for the obvious 25% plus in detail I found them to be really nice for my tastes. Nice a bit forward Vocals and plenty of air without being fatiguing. The bass seems a bit wooly which I account to the rise at about 150hz and the continuous fall towards the sub bass.

I made some measurements which I validated as best as i could by ear. This is done on an absolutely non valudated rig so only take it as an indication and comparison to the added measurement of the Sundara. The rig consists of a Yamaha surround receiver measurement microphone with a 3d Printed human ear on there, so take this really with a grain of salt and just as an added indication for what ive built.


I personally hear the bump at 900hz and 2,5khz stronger while the dip at 5khz is more at 5.2khz and a lot less strong.

To get these just a tad closer to what I like a added some eq wth a slight bass tilt towards the subbass and less mid/upper bass and then added a bit upper mids until 3.2k with a reduction of 4k to 5k after that. Each with at most 3dB difference and all in all not really a neccessary or striking change.

In the future i will try a few more pads to see if I can get away as good as I can from EQ for more flexibility.

I hope this was atleast an interesting read. Please feel free to ask questions or leave feedback or advice. I hope to add the posts for the other 2 in the coming hours to days

13 Likes

Lets talk about the second pair, my first DIY closed back. For them I took inspiration from the well known Fostex Th-x00 family or rather Forster Biocellulose Family.
For the build i wanted to also design my own headband and baffle structure. So almost all parts are handmade by me. They are built with 23 Ohm Biocellulose Drivers from Aliexpress.

The Physical Desing

I began again with the cups. Here I used a piece of amaranth and again cut out round disks to work on the woodlathe. Here screws in the middle were no option since that side would be the most seen part of the headphone in general, thus I added a sacrificial piece of beech wood which then got attached via woodglue to the amaranth and with screws to the lathe. I began with the Outside making sure to get a nice curve towards the back and leaving a bit of space towards the beech. This is necessary since the woodglue soaks a bit into the wood itself and interferes with the oil I intended to use. The outer diameter was about 105mm. Then the inside was carved out leaving a rather thick wall since i needed a slit between the baffle and cup for the pads while also being able to screw the baffle down into the cup.


I then sanded them first with an orbital sander to a grid 2000 and then polished per hand up until a grid 4000 until i achieved a mirror finish on the raw wood revealing each fiber and grain.

I then gave the wood about 3 passes with a very thin coat of wood oil but in the time after that I added a new coat once in a while.

The next task was building the baffle. For that I just cut out a circular piece of plywood with a thickness of about 4mm and then made the 53mm hole for the drivers. for testing I made a set of solid ones and a set of ones with 18 6mm holes drilled into it. On the backside of the holed one I later glued a piece of fabric to slightly reduce the treble that was added with the holes.
The Headband was the next challenge. Staying with the theme of the Fostex Headphones i first designed a Headband from Brass with an offset attachment of the yolk structure towards the back. The yolk itself was made from a 7mm wide strip of brass which i bent to form. The “blocks” holding the Beyerdynamic Spring steel and the yolk together was also from solid brass and designed to be DCA’esque able to fold the headphone for transport. With an added leather strip for comfort it looked a bit flashy but not bad per se.

This was a really bad design. The whole structire was too tall for a normal human head. The spring steel is not made to resist twisting so the pressure was far too low since the backside just twisted outwards. Next the hinge while not bad in execution is simply at the wrong spot. While “folded” the headband was just strongly bent outwand since the hinge point just doesnt leave enought length for it to “wrap around” the headphones cups.
My next design was more of the Audeze type. I attached a 6mm brass rod to the yolk strip with a screw and strong glue and replaced the folding block with a simple piece with a hole in the middle and a kinked attachment for the spring steel for a bit higher pressure. This also changed the extension from screwing trough a different hole in the leather to smply sliding the rod.

This was a LOT better. While still fairly heavy it sat solidly on my head to commute without thinking i’d loose it. But recently after acquiring a 3D printer i replaced that part with an even smaller and lighter piece that also gets the spring steel lower and further out to accomondate more heads. Another more obscure benefit is that the full metal headband had a tendency to “sing” or rather ring with winds which audibly resonated through the headphone. That was bypassed with some rubber but eliminated with the new plastic block

This also added a screw in the back to potentially lock the rod in place. (ignore the chipped black paint. Not only I am not good at painting metal these headphones also got used quite a bit)

The Tuning

I began with a simple foam insert from Beyerdynamic in the back of the cups. But with the cups sealed bass was simply non existant. To vent it I drilled 2 3mm holes next to the 3.5mm ports on each cup. That gave the bass back… big time.

The Pads I chose were a smaller variant of the ones I initially used on the ADX project. With solid leather outside and perforations inside it gave a decent treble and enough volume for a decent bass. As mentioned I made 2 baffles. The solid one gave in general a dull wonky sound so I didnt bother with them. The perforated ones added some more mids and treble while cutting a bit of the upper bass. All good improvements.


Lastly I added a foam piece in the cup that was part of the packaging of another headphone funnily enough. That gave EVEN MORE subbass while also evening and boosting the treble region.

Some Measurements

Firstly this was again made with the same NOT COMPARABLE OR NECESSARILY ACCURATE rig from the first post. So only take this as comparison and support for my personal impressions
The first graph will be just the Frequency response in comparison to the DIY ADX and my Sundara.


That phat subbass is just as audible but due to the big cut in the “normal bass” area is not always noticeable. Also The dip at 2.7k feels more like from 1.5k up until 3.5k where a big rise to 4k begins. i also hear the 9k peak stronger than indicated. That all gives the mids a bit of an artificial “hushness”. But pulling 4k and slightly boosting 1.5k to 3.5k fixes most of it. But that bass is with popstuff just right, with rock usually not apparant and with electronic just stupid fun.

Secondly heres the difference between the perforated and the solid baffle.

Here is the difference the added foam ring makes.


We see a big dip at 6.3k and a bit more upper bass with less subbass. The big dip is very likely due to a destructice interferance in the cup since the walls have no dampening.

The next graph shows the behaviour with all vents open, 1 vent blocke with my finger and then both vents closed.


You can see that the more closed it gets the 9k peak also rises. With one closed the dip in the bass goes lower and the subbass looses a bit of its “cleanliness”. With both vents blocked the dip dissapears but the bass rolles noticeably off. In person this effect is even stronger. But I may experiment in the future with semi porous material in the hole to get the best of both worlds, reduce the dip but keep my pocket subwoofer.

16 Likes

Holy hell brother. This is awesome and next level. I myself am attempting to put together a set of alpha opens …and hope to someday advance to this level. Love the woods and raw materials you used. I own a few ZMFs and Hylands and that aesthetic is beautiful…your creations are right on that level. Well done. Looking forward to reading about your 3rd set!

3 Likes

I am speechless! These are works of art and things of beauty!

2 Likes

The Third one will be a multiparter since I am still modifying it.

For this the inspiration was simply an inexpensive Hifiman Eggshape Pad on Aliexpress. I bought the pads and built the rest around that.

The drivers are the Peerless by Tymphany HPD-50N25PR00-32 which are as far as I know used in the Borealis and some Kennerton Headphones. I bought a few sets to also use in a future second closed back project.

The Physical

I again began with the “cups”. I used a piece of nice and thick plywood where i traced the outline of the pads on. I then cut the outside with a bandsaw and screwed the piece down towards a donor piece of wood. With a hand router and a concave curved bit I milled the outer profile. After then cutting a “lung profile” inside the piece with an electric jigsaw. With a convex curved bit the inner part got milled. This gave me a fairly reduced profile to reduce weight and give me room for modding.


For the grills I used the same honeycomb aluminium mesh from my ADX5000 project. I then also drilled the holes for the headband and a angled hole for the 3.5mm plugs on each side. After that I again gave the wood a few passes with wood oil to get the woodgrain out.

The innards I designed fully in 3D and then printed. In this post I will just show the Status Quo while I am still optimizing the design.
It consists of 3 parts. The first part is a flat piece with the correct holes for the cups and 8 3mm holes around the driver. The second big part is the backpiece that clamps the driver to the front piece. This also clamps the fabric down I will talk about later. The last part is a simple grill to keep me from sticking my fingers into the bare driver even more often. Due to the flat main plate the driver sits fairly close to the ear. The headband is the same design as the one for my Biocell closed back, more precisely the scraps from the prototype for that headband. With the new printed blocks it also holds the yolks very firmly. Since the pads are now slightly bigger or rather bulge out a bit the cups are a bit hard to tilt inwards.

The tuning

For tuning I ended here with a 3 part solution for now. Firstly I am using the full sheepskin pads with inner perforations. I shortly used the hybrid pads but didnt find them to fix the issue I was looking at, but I will revisit them with later designs for sure. Secondly I experimented with the backside venting or rather blocking of the openings from the “plate”. Here I am again using the fairly thin cotton fabric screwed down into the cup and between the plate and the back plastic. The densitiy does strongly impact all frequencys under about 250hz. After adding further fairly dense foams squished in the back openings I ended on a fairly linear bass response.

The foam does only impact the bass slightly and also did a slight change to the treble which is honestly not really audible to me but is measureble.
The pads are shipped with 3M adhesive but since I was not super happy with my results for the moment I only attached the pads with a strip of electrical tape. But random note be careful with that since the tape is appearantly sometimes strong enough to rip parts out of the leather. So no ripping it off every 20 minutes.

The Tuning I got sounds warm and thick to me. Not really what i was looking for. While measurements I will show later indicate that the bass and lower midrange are linear with just some dips in the upper mids it sounds as if after 2khz there are just 5+dB missing for what I usually enjoy. But that also leads to a sound that is nonfatiguing. It reminds me of my HD58X. Comfortable and listenable to hours but just missing in stage, clarity and just simply a wow factor.
With that being said lets see some measurements

Measurements

Here it is again with my dodgy rig compared to my Sundara. It has to be said that the measurements sometimes varied strongly but this is a result I got regularly and most reapeatably but again just with another grain of salt. As can be seen the bass extrends really well for a open back. The lack of air may be responsible for some of the perceived “lack” in clarity. Also just the imbalances in the upper mids / lower treble just give vocals a bit of a “dull shine” or rather just veiled with sudden moments of a kind of glare.
Its a bit hard to describe but in general I am just not content with my results and am currently printing new baffles for more air and space between the driver and ear. In very brief and dirty measurements the added space seemed to even out the treble. Additionally I hope to lower the lower mids a bit to get more distinct bass and leaner vocals which I simply prefer.

6 Likes

That’s really awesome! You are on your way to be another ZMF if you are not careful. Congrats!

1 Like

Thank you for your kind words! I am currently not looking to start any kind of business tho so I may be “safe” :smiley:

Well its time for an update. And boy oh boy did it improve in my opinion.
I experimented with the distance and angle of the driver in the beginning but the drivers are strongly reacting to the sealing of the chamber. With added seal not only does the bass rise in level, but also the frequency where the rise begins rises accordingly. Seal it totally and you receive a 10dB+ boost from 400hz down and with no seal it reminds me visually of the free field target.
The added distance on the other hand seems to even the treble and additionally does boost it in general. Another part that I added is an added angle to the driver. With an 8 degree tilt and a slight shift forward I am now very happy with the sound. But lets get into more detail with added measurements.

The Distance

My first try also used the angled distant design. But I didnt manage a correct seal amount so that was mostly held back by the bass. The second design is the flat, much more robust design you also saw in the last post, with cotton fabric and extra foam smushed into the holes. This current design is a stronger and more sophisticated variant of the first angled version. The driver is not held in with friction anymore but with an ring that also features hexagonal holes for the nuts. This allowed me to just clamp backing material instead of glueing it.


In this graph you can see the impact simply of the distance of a raw driver without housing or pad. This was measured again with a 3d Printed ear that does not seem to be perfect but it works as a representation.
It can be seen that with the added distance the treble above 4k rises in comparison to the 3.5k peak. With an added angle we see again the upper treble rise a bit but also the 7k peak travels further upwards.
Note that in the chassis this will change but its easier to show with the bare driver.

The Seal

To get the seal right I began with a brown felt fabric. With that alone I was still missing the bass I was looking for but it also slightly evened out the 4k to 10k treble. To get the bass back I the added a paper towel behind the felt. This was suprisingly effective giving a 10dB+ boost to the lower mids into the subbass. After cutting away bit by bit I ended up with about 1/3 of the area covered. This gave me a rise at about 160hz for a few dB. It also added more treble while taking a few dB in the mids giving a more engaging and clear sound in my opinion.


The last change I made was to use the hybrid pads instead of the pleather ones. This was mostly a comfort change since they seem to dissaper easier with those.

As you can see the change ist negligeble.

Comparisons

Lastly lets compare this revision against my other headphones, beginning with graphs and thus the tuning.


As you can see the ADX is a bit more uneven in the upper mids and then adds more air while the Tymphany adds a bit more Subbass. Shown here is also an average since the position on the head or rather in respect to the ear changes the frequency response a bit. Generally further back adds treble while pulling it further back reduces that range.
This tuning now is really nice to my ears with one “oddity”. Some vocals seem undecided whether they are slightly more rounded or hushy in the overtones. This doesnt happen a lot but is still rather strange.

Lastly the “subjective” stuff. Staging and imaging is in no comparison to the old version. It is wider, more defined and seprated in all cases. The bass seems more distinct and impactful. What I assume to be detail also seems clearer to me which likely just stems from the added treble and better separation just giving tones a easier time to come through.

In conclusion these have not left my head the last day and I really enjoy them like this. I will have to see if this is just the initial romance with a positive change or if these really are one of my favorites in general like this.
As like in the last posts feel free to ask questions or give suggestions. I know this post is a bit all over the place in structure but so is my process lol.

3 Likes

The next headphones were inspired by a friend constantly whining about his shitty headset yet not willing to spend some money to get actual good sound. With his birthday on the horizon I thought “oh well, still have a pair of spare drivers and pads lets see what we can achieve with the 3D printer”.
Since that would make a good “community” headphone i thought i’d do my best to build them with of the shelve parts.
And this is what ive come up with:

Meet the “Jalter”

As you can see they are very much open backed and kept simple but fairly elegant in design.
Its a neutral with a slight V tonality with good spaciality. It weighs 308g at this point and i can wear it for hours with no issues.
The off the shelve parts are:

A pair of Tymphany HPD-50N25PR00-32 which you know from the previous headphone aswell as the Borealis. They often are sold out but as of writing this are in stock at DigiKey

Next a Beyerdynamic spring steel. Just go to their website, on the bottom to Spare Parts and you can choose between the high pressure or low pressure variant originating from the “Pro” or “Edition” variants from the 880 for example. For me the low pressure is ideal but your mileage may vary.

Then a pair of 3.5mm Sockets. I used the literally first result from AliExpress for 2 bucks for this design. They rattle a bit with no cable inserted but work just fine otherwise.

For the comfort Strap I use about 22cm of leather but you can use whatever width and material that you want for it

You will also need some screws. To attach the headband, the baffle and fasten the yolks to the rods and into the chassis I use M3x10 and M3x6 Wood screws. Use the short ones on the headband blocks as to not peirce the material. The driver gets clamped with M3x12mm countersunk metal metric screws and the corresponding m3 nuts. For the headband you also need m3x8mm countersunk screws. You also need a little bit of wire to solder to the sockets and the driver.

For Tuning some felt and paper towels or whatever floats your boat.

Lastly the pads I landed on are Brainwavs XL Hybrids. They seem to give the best frequency response. But you can of course choose other pads. 105mm is minimum and can be too tight to fit but 110mm or more should fit just fine.

The Printed Parts:


The housing consinst of the structural Outer Housing which holds the Arc and the Socket. The Baffle which holds the driver screws directly into it aswell. Through the grid which keeps your fingers out of the driver and the baffle yuu screw into the nuts set into the clamp. This holds the driver really tightly. Then you screw the keyed end of the rod through the Arc with another wood screw. Not shown here are the “blocks” that hold the spring steel and the comfort strap aswell as the grill on the outside but that is just jammed in the hole and held with some Pattafix. On the blocks the Spring steel gets screwed down with some m3x8 metric screws and the leather strap uses the shorter wood screws. The middle hole is to optionally use a m2,5 screw to fix the Rods which is not necessary in my experience.

https://www.thingiverse.com/thing:5482763
As you saw in the Picture I only modeled the right side. For the left side mirror the parts in the slicer. The rods and the Blocks are symetrical so just print them twice
On thingiverse I also wrote short instructions how to print what.

The Assembly

Fairly Straight forward. Clean the 3D Parts from the support and check for fitment. The driver should just fit into the baffle and the clamp and the rod should just fit into the blocks. If it doesnt first drill out the block with a 6mm drill. If it still doesnt fit print the rod again a bit smaller, maybe 98% scale. It should slide with friction. It will loosen over time and usage a bit.

Then insert the Headphone socket and solder the wires to it. I like to route it around the housing to end at the top. Take the baffle and lay it on some felt or fabric with the convex side. Cut out the driver hole and cut around the perimeter. The thickness of this fabric is a necessary tuning decision I will talk more about later. With the tuning fabric cut out, slot the driver in the baffle, put the fabric around it and push the clamp on the driver taking care not to poke the diaphragm. Now take the Grid and the m3 screws. Use the 4 holes with no bevel on them and screw the Grid - Baffle - Clamp sandwich together with the nuts to secure the driver and the tuning fabric in place. Now add the other 4 screws for good measure. You can now screw the baffle with the wood screws onto the outer housing again clamping the tuning fabric in between. Now solder the wires onto the driver. You can also use an interconnect beween the driver and the socket to make it easier to change things up.
No try to stick the rod into the arcs hole. The rod should be angled away from the portrusion on the arc. This can be too tight to fit. In that case just carefully cut the keying on the rod down until it somewhat fits. With a wood screw screw it down through the arc until it feels really solid. Now you can push the Arc into the housing making sure its the right side. It should take a bit of force but once in turn fairly well. Now you can screw another wood screw into the sides to keep the hooks from depressing, keeping the arc fixed solidly inplace.
Lastly the Headband part itself. Shove the spring steel carefully into the slots on the front until the hole line up and the use a 6-8mm m3 screw with pressure to fasten it. Cut a roughly 22cm strip of whatever you want as a comfort strap and just thread a wood screw directly through it into the blocks holes.
Add the pads and after inserting the rods into the blocks the headphones are basically done. The extra holes on the end of the rods can be used to keep them from ever sliding out on their own but arent necessary.

The Tuning


As this is an DIY headphone I highly want to encourage you to try stuff for yourself. And since the stuff and way I used them are very hard to exactly reproduce its easier for me to just give you intructions on how to tune it and what things do.
The goal is a mixture of reflectivity control aswell as finding the “correct” amount of seal in the chamber. The Tymphany drivers react strongly to the seal of the chamber as shown in the previous project. So you dont want to just seal it completely (so either open pads or more closed pads and open housing).
What i used is some medium thick felt and some paper towel. The felt by itself was already pretty ok, giving some seal so that the bass was essentially linear until a slight rolloff in the subbass, but i like a little more sauce in there. So the paper towel again is used to seal it up a bit more. You can of course achieve this with another fabric altogether thats just more solid than the felt i used. Sealing the complete lower portion was too much in my case but adding the slits there gave the bass a very tastefull boost of maybe 2 dB. In my experience I also was able to reduce a bumb in the upper mids by putting the paper inside instead of outside, probably due to its reflectivity so maybe try that aswell and look what works best for you. Also you can try to add some front or rear dampening but I personally like to keep the driver as “nude” as it gets and just change the environment. But you aint me so do whatever you feel is correct.

Measurements
As per usual i like to follow up with some measurements to show what it does. The first graph will again be made with my janky DIY setup with a surroud receiver mic and a plastic printed Pinnae. This is NOT a standard and should NOT be compared outside of what I show. I had the opportunity to measure my stuff with another rig that has IEC couplers for a bit more accuracy which I will show after my own.

The Jalters are shown in green, the Sundara in blue and the previous Tymphany build in red. You can see A slight boost in the bass followed by very smooth mids. Compared to the Sundara we have a slight bump at 2-3kHz with a drop at 4.5kHz. The big spike at 7kHz is a lot less prominent in a sine sweep than it seems in the graph. Lastly one could argue for similar upper treble levels but this is so far beyond the point where it is accurate that it could just be measurement error.
If we compare it to the previous Tymphany headphone we can see that we have more smoother mids from 500Hz - 3kHz, slightly less bass and more treble. This checks out with listening with the Jalter being very neutral in general and very clear to my ears. Due to the similar angled driver design spaciousness is again excellent in my opinion.
Lastly something of interest in my opinion is distortion. Due to the very open and undampaned nature of the driver we see rising distiortion in the subbass. I tried to replicate the levels with my Sundara which is a lot cleaner in those regards. My finding was that the second harmonics that I see with the jalter up to 50ish Hz just give bass a thicker tone which i thoroughly enjoy. Note that a more closed chamber with the resulting more bass level will also dampen the driver and reduce distortion a bit. So keep that relation in mind.

A few final Points
This is my personal design which you may print, modify and use for personal non commercial use. I am no professional CAD or Acoustic designer so all parts can be a little rough and/or non optimal but as a DIY project I encourage people to try things out for themselves and use this works as a starting point that works pretty well.
If somethings was unclear or if you have any suggestions please feel free to ask/post. Since it is rather late at night at this point things can be a little jumpy and duplicates here and there.

Lastly there is already another 'phone in the pipeline a bit more exotic :upside_down_face:

7 Likes

This is really impressive.

When you use something like a paper towel as part of the build, is there any risk that it’s going to deteriorate over time and need to be replaced? Or are paper towels likely to last as long as any of the other components?

Paper will deteriorate. So do foam, leather, and rubber. Vintage speakers routinely require new surrounds after perhaps 20 years, and paper cone speakers are made of paper. I put my finger through some old paper speakers when I was a child. They were as brittle as a dead leaf.

2 Likes

Yes that’s my assumption as well. I don’t have it in headphones long enough to be certain but I think it should last just fine given that you don’t dunk them in water. Since it’s not a moving part I guess that the natural deterioration should be fairly slow.
Id guess that toilet paper lasts less long since it’s engineered to dissolve easily in water and paper towels are the exact opposite but that just a random thought with no evidence to back it up

1 Like

Impressive work on all these headphones.

Im intersted in trying to create my own custom 3d printed headphones and im looking for inspiration. “Jalters” baffle design looks like a design i would like to tinker with. But the problem is that the link to thingivers goes straight to 404.
Could you help me out with a working link to the design.

Thanks.

Man I wish I could whine to my friends and they’d build me works of art. Need new friends.

1 Like

Hey for me the Thingiverse link works just fine. I wont deny the possibility that I configured something wrong. Would be cool if somebody else could chime in whether the link works for them

Seems that the link is working now. Could have been a problem on my end.

Cant thank you enough for sharing such an amazing design for free! This will be a huge help for me to design my own 3d printed headphones.

How cool is that! Really impressive work, congrats!

By any chance… are you familiar with infra red audio? I was wondering how good it really is, or can be if done right.

Funnily enough I am but maybe not in the exact concept you are referring to.
In an university course we learned about the photoacoustic effect for gas spectroscopy.
There you send a periodic laser pulse through the medium you want to analyse. Any gas absorbs light at some wavelengths and if you want to see how much methane is in a medium, you choose a laser wavelength that will be absorbed ideally only by methane. In the absorption process the gas will heat up and thus expand. Due to the periodic nature the expansion and contraction produce soundwaves that can get measured.

That is basically what’s used in plasma tweeters to an extend. Otherwise I don’t think it makes too much sense given the nature of heating and cooling for expansion and contraction. For a low note a lot of heat would be needed for it to work and I am not too keen on quickly heating my room.
I am curious about the plasma tweeters though it’s ozone byproduct is a bit sucky

How good that is I don’t know. You obviously loose a lot of problems with traditional transducers but that surely doesn’t leave you without issues. But maybe look into plasma tweeters.