Nice work. Although I'd probably had used one of the newer esp32 from seeed Studio with integrated battery pads (unfortunately the BatteryPal from PNLabs is still in the making). A DFPlayer Mini might also habe saved some time...
There also is a german project called Tonuino, which is quite similar and has lots of examples and documentation - they even designed a custom PCB which unfortunately is no longer available:
I agree with the commenters who were jarred by the heavy use of LLMisms -- the writing was genuinely painful to read. What a really fun project though!
Props for seeing it through a debugging phase, it can really be worrying when you get your boards and they just don't work.
It’s odd because the first part sounds like it was written or at least copy-edited by a human who started getting tired and gave up, and as it goes on it sounds more and more sloppish. Nice project though.
I read it through. There have always been articles, even before LLMs, that exceed your personal "drivel" threshold, that you abandon. That's ultimately what matters, do you want to keep reading or not.
That was much rarer though. Now the sad norm is someone's idea, AI code with AI docs and AI "article".
I'm not spending my time on reading something someone claims it's not important enough to write or at least edit. At most I'll send my LLM to check if there's anything useful in the code (but i don't reuse it because I know my harness and tests are sound, not sure about someone so lazy).
One thing I will say about the Yoto player is that there is an API: https://yoto.dev/api/ . You can do all sorts of things: get button presses over a MQTT queue, redirect the player to download audio from a server that _you_ host, and even dynamically update the playlist behind the scenes. Oh, and add static or animated GIFs for display on the device. It's really fun to mess around with - I created a travelogue for my daughter when I was away for a work trip a couple of months ago…
This is an excellent project! Congrats and thanks for sharing! (LLM writing notwithstanding)
I wonder if it would be a good base to create a kids phone / yogurt-cup telephone: no screen, no keyboard, just a box and figurines, and to call someone up you put the figurine on the box. It could be boxes-only (only people with a box can call or be called) or it could be more general: kids only have boxes and figurines, but the machine is able to call actual phones.
This is incredible - I definitely wanted to do the same for my kid once he gets a bit older so I'm bookmarking this.
I read through the whole thing and found it informative, LLM or not. To all AI low effort naysayers I think for a working parent this maybe would not have been possible to write in reasonable time without the aid of Claude.
Wow thanks for the info about the toniebox reverse engineering!
My kids got offer a toniebox this year, and when discovering the product I was so upset about this locked and privacy nightmare I brought it back to the seller for a refund and started to look for alternatives (got time, still very young). Will check it out!
What are you talking about? "I hate him"? He literally learnt how to build an interactive electronic toy for his children and then put in the effort to share it online. You didn't even take the time to read it because he talks about the Toniebox at the very start of article.
So he used AI to do the write up? He also references paying in Euros so English is probably not his first language. I think you need to calm down and critically examine your priorities.
I loved the article. This is a project I've been wanting to do myself.
Fast dopamin? Did you read the article? I did. I think he mentions spending 3 months on this. Going from zero to hero on a project like this failed on 3 attempts, because figuring stuff out from forums alone is just way too hard.
Sorry man, but I really don't see your problem. A guy made something and wrote an article about making it. Did LLMs help? Yes, he's quite open about that. So what?!
I absolutely hate AI writing as well, but I suspect since the author is French, and the underlying code in their GitHub project [1] uses a lot of French naming, that it was more of a translation aid.
But the problem is invented - if you invested a fraction of the time to research, you would see that custom tonies sell for like $8 a piece for up to 90min of your audio, or you can diy all of it by getting the ICODE SLIX-L chips for like $1.5 a piece and adding any content by accessing the internal micro sd card.
First off - well done, this is amazing. Secondly, it's a fascinating read how you basically figured it out, one mistake at a time, like not connecting or fastening the battery correctly. This is the way! We make mistakes and figure things out in our journey to make something interesting, and learn along the way.
I am mostly into software my whole life. I want to dabble with hardware and electronics in the future and this post taught me many things that’ll be very useful when I do.
Tnx for writing every bit of the building process in an easy to understand way and also open sourcing it.
This is really sweet. Reminds me of an old client I worked for called Museum in a Box (https://museuminabox.org/). It was started by George Oates (ex. Flickr).
3D printed figurines on top of a box which would activate museum style narration and multimedia. It was based on a Raspberry PI though which is different from this project but in many ways, it was the same.
Toniebox has a handful of patents on parts of the functionality. Not that it matters here, just a heads-up in case anyone thinks of productizing something like it :)
Things like detecting a figurine placed on top of the box with RFID/magnet and then doing something.
love it! incredible! and you got it to work at first attempt!
i really need to start to find time to my own unfinished..or not started projects that are little bit similiar.
annd you made your own pcb?! what the actual .. i was thinking that's something you'll need to have years of experience and 5 revisions
Love that it's open source, nice alternative to the closed boxes like Toniebox. How do the kids pick a story, NFC cards or buttons? And how long does the battery last on the ESP32?
I wanted to give my three kids a screenless story box (inspired by the Toniebox gesture) but I was deeply annoyed by the closed ecosystem and subscription-like models of commercial alternatives.
So, despite knowing absolutely nothing about electronics three months ago, I decided to build one myself using an ESP32, an NFC reader, and a Bambu Lab printer. I learned everything the hard way, from accidentally creating short circuits to routing 491 tracks by hand on my first custom PCB with KiCad (and using Claude to double-check my math).
The box serves its own local web page to upload MP3s/M4As, streams web radios, and handles self-updates over Wi-Fi with a built-in safety net in case the network stack crashes.
Everything is fully open-source (firmware, KiCad files, 3D models under MIT license) on GitHub: link in the article
I’d love to hear your thoughts, feedback on the PCB routing, or answer any questions about the build!
I'm planning something like this. I already have Home Assistant with Music Assistant working, so playing radio and music and audiobooks in rooms are already possible with my phone. I plan to add ESPHome device with NFC reader to trigger plays so it's easy for the kids to use it. Just need to find where I left my roll of 50 NFC stickers.
One suggestion: In my opinion, kids nowadays are very familiar with charging devices with USB-C. Powering the device with USB-C will be much easier. Worse case scenario, I'll use some small portable power bank and place it inside.
It‘s awesome, thanks and your progress from breadboard to finished project is amazing! Don‘t let all the complaints here keep you from building and sharing.
You could consider replacing the potentiometer with an encoder that has no fixed range and can’t be broken by turning it too much.
There also is a german project called Tonuino, which is quite similar and has lots of examples and documentation - they even designed a custom PCB which unfortunately is no longer available:
https://github.com/tonuino/TonUINO-TNG
Another rasperry focussed project that has a web interface is Phoniebox:
https://phoniebox.de/index-en.html
There also is a toniebox hack, which lets you setup a self-hosted content server:
https://gt-blog.de/toniebox-hacking-how-to-get-started/
https://github.com/biologist79/ESPuino
Props for seeing it through a debugging phase, it can really be worrying when you get your boards and they just don't work.
I'm not spending my time on reading something someone claims it's not important enough to write or at least edit. At most I'll send my LLM to check if there's anything useful in the code (but i don't reuse it because I know my harness and tests are sound, not sure about someone so lazy).
They seem like a solid company
I wonder if it would be a good base to create a kids phone / yogurt-cup telephone: no screen, no keyboard, just a box and figurines, and to call someone up you put the figurine on the box. It could be boxes-only (only people with a box can call or be called) or it could be more general: kids only have boxes and figurines, but the machine is able to call actual phones.
I think even as an adult I would enjoy that.
I read through the whole thing and found it informative, LLM or not. To all AI low effort naysayers I think for a working parent this maybe would not have been possible to write in reasonable time without the aid of Claude.
Kudos to the author.
My kids got offer a toniebox this year, and when discovering the product I was so upset about this locked and privacy nightmare I brought it back to the seller for a refund and started to look for alternatives (got time, still very young). Will check it out!
So he used AI to do the write up? He also references paying in Euros so English is probably not his first language. I think you need to calm down and critically examine your priorities.
As long as they had fun and enjoyed the process, that’s what counts. Enterprise software deployment is a different story, of course.
Fast dopamin? Did you read the article? I did. I think he mentions spending 3 months on this. Going from zero to hero on a project like this failed on 3 attempts, because figuring stuff out from forums alone is just way too hard.
Sorry man, but I really don't see your problem. A guy made something and wrote an article about making it. Did LLMs help? Yes, he's quite open about that. So what?!
[1] https://github.com/Zebra-Rodeo/marelleo
had to stop reading from that point onwards.
But the problem is invented - if you invested a fraction of the time to research, you would see that custom tonies sell for like $8 a piece for up to 90min of your audio, or you can diy all of it by getting the ICODE SLIX-L chips for like $1.5 a piece and adding any content by accessing the internal micro sd card.
Tnx for writing every bit of the building process in an easy to understand way and also open sourcing it.
I liked the look of your box too. Keep up!
3D printed figurines on top of a box which would activate museum style narration and multimedia. It was based on a Raspberry PI though which is different from this project but in many ways, it was the same.
Things like detecting a figurine placed on top of the box with RFID/magnet and then doing something.
You’re not buying a player, you’re buying a subscription [...]
I particularly hate the first one, don't AI sloppers notice how horrible it is to read "No x, no y, no z"?
I wanted to give my three kids a screenless story box (inspired by the Toniebox gesture) but I was deeply annoyed by the closed ecosystem and subscription-like models of commercial alternatives.
So, despite knowing absolutely nothing about electronics three months ago, I decided to build one myself using an ESP32, an NFC reader, and a Bambu Lab printer. I learned everything the hard way, from accidentally creating short circuits to routing 491 tracks by hand on my first custom PCB with KiCad (and using Claude to double-check my math).
The box serves its own local web page to upload MP3s/M4As, streams web radios, and handles self-updates over Wi-Fi with a built-in safety net in case the network stack crashes.
Everything is fully open-source (firmware, KiCad files, 3D models under MIT license) on GitHub: link in the article
I’d love to hear your thoughts, feedback on the PCB routing, or answer any questions about the build!
One suggestion: In my opinion, kids nowadays are very familiar with charging devices with USB-C. Powering the device with USB-C will be much easier. Worse case scenario, I'll use some small portable power bank and place it inside.
You could consider replacing the potentiometer with an encoder that has no fixed range and can’t be broken by turning it too much.