GUIDE

Home Assistant text to speech with a natural local voice

Smart home announcements are only pleasant when the voice is: the washing machine is done, someone is at the door, it is time to leave for school. A natural voice that runs on your own network keeps them private and keeps working when the internet does not.

SuperClonica is not a built-in Home Assistant integration, but its local server connects in a few lines through a shell command and a media player. This guide shows the setup and how to write announcements that sound natural.

Browse voices Install the SuperClonica engine

What you can announce

Everything runs on your own network; only the first model download needs the internet.

  • Appliance and doorbell messages on your smart speakers.
  • A morning briefing with the weather and the first appointment.
  • Reminders for the family at fixed times.
  • Alarm and safety messages in a calm, clear voice.

Pick the voice free

On the page of any voice, type a typical announcement, 30 to 100 characters, choose the language and press Voice it. In a few seconds you hear how your home will sound, with no account.

Connect Home Assistant

Voice fixed messages once and reuse the file; generate a new file only when the text changes.

  • Run the server on a machine in your network: docker run -d -p 8000:8000 -v superclonica-models:/models superclonica/superclonica:cpu.
  • Add a shell_command in configuration.yaml that calls curl with a JSON body, text, voice and lang, and saves the WAV to the www folder of your configuration, for example /config/www/tts/announce.wav.
  • Files in www are served under /local/, so the announcement is at http://YOUR-HOME-ASSISTANT:8123/local/tts/announce.wav.
  • In an automation, run the shell command, then call media_player.play_media with that address and media_content_type music.

Announcements that sound natural

Keep announcements short, one or two sentences. Write times and numbers in words, “half past seven” rather than 7:30, because digits are read as plain numbers. Set lang to the language of the text, and give different rooms or message types their own voice so the family knows at once what is going on. A short phrase is ready in well under a second on an ordinary processor.

Privacy

The server and the audio stay on your network. Anything in the www folder can be opened without logging in, so do not voice private details into it if your Home Assistant is reachable from the internet; set SUPERCLONICA_API_KEY if the speech server is reachable from other machines.