I am fine.
A voice-based mental health check for people who are definitely fine.
Your voice never leaves your session. We keep the model; we don't keep the audio.
English only — though the model won't judge your accent. It has one too.
Built on Kintsugi's open-source voice biomarker model. Peer-reviewed. Not a diagnosis.
Talk naturally for 35 seconds. Silence doesn't count — the model needs to hear your voice (pitch, tone, timbre, and pauses), not your thoughts.
Hold to record · 35s
Analyzing
Powered by Kintsugi DAM v3.1 — an open-source voice biomarker model trained on 35,000 voices and published in the Annals of Family Medicine.
Here's why "I am fine" exists: to use AI to deconstruct the anxiety of the AI age — with humor, so you can look at the paradox with a good laugh.
Tested a friend? Send them the link, not just the screenshot. They deserve the full experience.
This model was trained on English-language speech data. Results for other languages are unvalidated and provided for entertainment only.