Voice prompts for Arduino and ESP32 projects
A talking project feels finished: a door lock that says “welcome home”, a plant sensor that asks for water, a quiz box for kids, an escape room prop. Microcontrollers cannot run a neural voice, but they can play audio files, and a natural voice makes a hobby project sound like a product.
The free SuperClonica engine voices every phrase on your computer in advance. This guide shows how to prepare the files, keep them small and name them so your player finds them.
How a talking project works
An Arduino or an ESP32 cannot run the speech model itself; pre-voiced files are the reliable way.
- The board decides when to speak and which phrase to play.
- The phrases are audio files on an SD card or in flash memory.
- A player module or the board itself plays them through a small speaker.
- The voice is made in advance on your computer, so the board needs no internet.
Pick the voice free
On the page of any voice, type one of your prompts, 30 to 100 characters, choose the language and press Voice it. In a few seconds you hear how your project will sound, with no account.
Voice all the phrases at once
The --normalize option evens the loudness, so no prompt is suddenly louder than the others.
- Install the free engine on your computer: python -m pip install superclonica-cpu.
- Keep the phrases in a list with a number for each.
- A short script voices each one: superclonica --text "Welcome home!" --voice F1 --lang en --normalize --output 0001.mp3.
- Name the files with numbers, the way your player module or code expects them.
Small files for small boards
The engine writes 44.1 kHz audio in WAV, MP3, FLAC or OGG. MP3 keeps files small for SD-card player modules; for playback from the flash memory of an ESP32, convert the WAV to a lower sample rate, such as 16 kHz mono, with an audio tool like ffmpeg or sox. Short prompts of one sentence take a few kilobytes to a few dozen kilobytes as MP3.
Voices you can build in
For a project you sell or show publicly, use a voice you have the rights to: the ten stock voices of the model, M1 to M5 and F1 to F5, are free under the model license, or train your own voice in the SuperClonica studio from a recording of a speaker who agreed. Character voices from the catalog are fan-made and fit personal and fan builds.