Gives Assist a voice and an ear: text to speech and speech to text through Uttera, processed on our own hardware in Spain.
HACS → Integrations → ⋮ → Custom repositories → add
https://github.com/uttera/uttera-home-assistant as an Integration, then
install and restart Home Assistant.
Then Settings → Devices & Services → Add Integration → Uttera and paste your API key. Get one free at app.uttera.ai — no card needed.
Uttera then appears as a speech-to-text and text-to-speech engine when you configure an Assist pipeline.
Your audio does not leave Spain. It is processed on our own machines, discarded once answered, and never used to train anything. For a microphone that lives in your house, that tends to matter more than it does elsewhere.
It speaks 24 languages and the voice is selectable per call.
Credits from your monthly allowance, the same as the API. The free plan needs no card and is enough to try it properly.
No dependencies. The integration uses the HTTP session Home Assistant already has. An integration that pulls in packages is the first thing to break when the core updates.
Line breaks are stripped before synthesis. Each one inserts a pause of about 1.3 s and is billed, and Assist often sends text with them.
Apache-2.0