When someone you love has dementia, your voice is irreplaceable. Not just what you say, but the sound of you. The familiar cadence. The specific warmth. For your loved one, hearing your voice activates something deeper than memory: a felt sense of safety that dementia cannot fully erase. Voice cloning technology now makes it possible for that comfort to be available at any hour, from any phone, without you needing to be the one answering.
What voice cloning actually is
Voice cloning creates a digital model of a specific person's voice, capturing its tone, rhythm, warmth, and individual characteristics, that can then generate new speech that sounds like that person. Modern voice cloning uses short recordings (typically 10 to 30 minutes of natural speech) to build a model accurate enough that people who know the voice well recognize it as familiar.
In dementia care, this matters because the familiar voice is doing something specific and documented: it triggers the emotional recognition that exists even when declarative memory has declined. A person with moderate dementia may not remember what day it is, but they know when they're hearing someone they love.
How KindredMind uses voice cloning
KindredMind's approach is built on a principle from simulated presence therapy: a familiar family voice, responding to the specific needs and history of a person with dementia, provides real emotional comfort. The voice clone is only one component. The other is knowledge: the stories, routines, reassurances, and personal details that make a response feel genuine rather than generic.
When you set up KindredMind, you record your voice and build a knowledge base about your loved one. The AI uses both to answer calls: in your voice, with your words, about the things that matter to your loved one specifically.
You can read more about the principles behind this on our approach page, and about how we protect the person receiving these calls on our protection page.
The voice clone is only one part. The knowledge base is what makes the response feel like you.
Is voice cloning safe and ethical for dementia care?
This is the question we think about most carefully. Using a cloned voice to answer calls to someone with dementia requires being honest about what we're doing and why. Our position:
The person calling is seeking emotional comfort, not factual information. They are calling because they need to feel connected to someone safe. The cloned voice provides that connection. It does not deceive them about anything that matters to their wellbeing. It gives them exactly what the call was reaching for.
KindredMind requires full informed consent from the family caregiver, who authorizes the system on behalf of their loved one. We recommend transparency with care facility staff. We build safety protocols that escalate to real human contact when the call indicates an emergency. The goal is always to serve the person with dementia, not to replace human connection, but to make it available in the moments when it can't be physically present.
Our protection framework covers this in detail, including exactly how emergency escalation works and what information is stored.
What the research says about familiar voice comfort
The use of familiar voice recordings to comfort people with dementia has been studied under the framework of simulated presence therapy since the 1990s. Research has consistently found that familiar voices reduce agitation, anxiety, and distress in people with dementia, particularly during periods of separation from primary caregivers.
The key limitation of earlier simulated presence approaches was that they used static recordings: the same message played repeatedly, unable to respond to what the person was actually saying. KindredMind addresses this directly. The voice is responsive, not recorded. It adapts to the specific call.
A Cochrane systematic review identified simulated presence therapy as a recognized non-pharmacological approach to dementia-related distress and studied it in controlled trials. KindredMind builds on that research foundation and goes further by making the voice interactive rather than passive.
How the voice clone is built
KindredMind uses professional-grade voice cloning technology. The process:
- You record your voice. We recommend 20 to 30 minutes of natural conversation, which our onboarding call guides you through.
- The model trains on your recordings, capturing the specific characteristics of your voice.
- We build a knowledge base from an intake form covering your loved one's history, preferences, routines, and the things they most commonly ask about.
- The system goes live. Your loved one calls. They hear you.
As calls accumulate, the voice model improves. More of your voice means a more accurate, warmer representation. You can learn more about the full setup process on our how it works page.
Record your voice once. Be there every time they call.
KindredMind takes 20 minutes to set up. After that, every call your loved one makes is answered with your warmth, your words, and the things only you would know.
Your voice. Always available.
Start Your First Month