Dadlexa: The No-AI Smart Speaker
Forget AI — Dadlexa is a smart speaker that lets a dad secretly send voice responses to his child's questions.
It’s hard to fully appreciate how revolutionary radio and television were when these technologies first arrived in people’s homes. Hearing the voices and seeing the faces of people who were not physically present was so shocking that some people imagined little people actually lived in the new boxes that sat in their living rooms. After experiencing decades of technological advancement, we are not so easily amazed by new gadgets today. But Sean Clark’s son might be starting to wonder if a tiny version of his dad lives in a little box called Dadlexa.
Dadlexa is a DIY smart speaker created by Clark. Unlike its commercial counterparts, Dadlexa is not powered by artificial intelligence. Rather, it is powered entirely by good old biological intelligence supplied by Clark himself. When his son presses the button and asks a question, the audio is forwarded to Clark’s phone. He then answers, and the audio is played by the smart speaker.
The project is built around a $24 Waveshare ESP32-S3 development board that is well-suited for making a smart speaker. The board includes everything needed for the job: a microphone, speaker, ESP32-S3 microcontroller, and a small LED ring for status indication. Clark wrote the firmware using ESP-IDF and integrated several components from Espressif’s speech recognition stack to create an experience that feels just like using a commercial device.
Dadlexa is always listening for the wake phrase “Hi ESP” using ESP-SR WakeNet. Once activated, the speaker emits a tone, illuminates its LED ring, and begins recording audio. Recording continues until several seconds of silence are detected. The captured speech is then processed locally before being sent to Telegram, where Clark receives the message on his phone. He can respond with a voice message, which is downloaded by the speaker and played back through its onboard audio hardware.
While the remote parent-to-child messaging feature is the main attraction, Dadlexa includes a few extra tricks. Clark added support for local voice commands that trigger music playback directly from a microSD card. A manifest file maps spoken phrases to MP3 files, allowing commands such as “play milkshake” or “play bedtime” to start playback without involving Telegram (or Clark) at all.
Once the device joins a Wi-Fi network, it hosts a small web interface that displays available songs and voice commands. New MP3 files can be uploaded through a browser, assigned to custom trigger phrases, and stored directly on the SD card. After a reboot, the new commands become available.
In a world increasingly focused on replacing human interaction with automated systems, Dadlexa uses modern hardware and software to accomplish the exact opposite. Instead of putting an AI assistant between a child and a parent, it creates a direct line of communication. In this case, the smartest thing inside the smart speaker is still a person.