Your voice is the most powerful comfort tool you have. Voice cloning makes it available all the time.

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, a familiar voice may feel recognizable or comforting to some people, although responses vary. Voice cloning technology now makes it possible for that comfort to be available when the service is active and supported calls can be handled, 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 may support a sense of recognition that exists even when declarative memory has declined. A person with moderate dementia may not remember what day it is, but some people may recognize or respond to a familiar voice.

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.

Is voice cloning safe and ethical for dementia care?

This is the question we think about most carefully. Using a 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 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 is set up and authorized by the family caregiver, the person who holds legal decision-making authority for their loved one's personal care. This is the same framework that governs every other care decision made on behalf of a person living with dementia: which facility, which medications, which care approaches. KindredMind sits within that framework, not outside it. 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. We have also written in full about the ethical framework behind voice cloning in dementia care, including the consent and legal authority question. Read our ethics position →

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 examined familiar-voice interventions and reported varied outcomes; findings are specific to the interventions and populations studied.

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. KindredMind is designed to provide responsive conversation; this product capability has not been clinically established by the cited simulated-presence study.

PMID 38646703 included 85 patients described as having senile dementia: 40 received routine nursing and 45 received routine nursing plus simulated presence therapy. Reported measures included agitation, protective-restraint duration, apathy, functional independence, depression, anxiety, clock-drawing performance, and caregiver burden; between-group findings were reported at p < 0.05. The abstract does not establish repetitive calling, telephone delivery, AI, interactive conversation, or outcomes for KindredMind. The trial did not test KindredMind, voice cloning, or AI-delivered conversation. KindredMind provides interactive familiar-voice conversation, but no clinical outcome should be inferred from the separate study.

How the voice clone is built

KindredMind uses professional-grade voice cloning technology. The process:

1. You have 20-second express onboarding of natural conversation.
2. The model trains on your recordings, capturing the specific characteristics of your voice.
3. 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.
4. The system goes live. Your loved one calls. They hear you.

As calls accumulate, the voice companion deepens. 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.