Short answer: yes, you can have an AI answer your phone in a voice that sounds like yours. OsmO's Custom Voice feature records a short sample of your speech, clones it, and uses that voice whenever your assistant picks up. That makes the interaction feel more personal than the fixed system voices people usually hear in caller-ID apps and built-in screening tools.
Why the voice matters more than people expect
The single most common complaint about automated call screening is that callers hang up. A robotic voice announcing that it's screening the call reads, to a lot of people, as "this number is dead" — so the recruiter, the clinic confirming your appointment, and the contractor you were waiting on all give up before saying anything useful. You end up with a log full of abandoned screenings and no idea which ones mattered.
That's a voice problem, not an intelligence problem. A natural, familiar voice keeps people on the line long enough to say why they called, which is the entire point of screening in the first place. Using your own voice takes that a step further: your regular contacts hear something they recognise, and the interaction feels like reaching your assistant rather than hitting a wall.
How Custom Voice works
You record about fifteen seconds of your speech inside the OsmO app. That sample is used to build a voice model, which then speaks your assistant's side of every call it handles — both the calls it answers for you and the calls it places on your behalf.
A few practical details worth knowing before you start:
- Fifteen seconds is enough. You're not reading a script for ten minutes — it's one short take.
- It's used on outbound calls too. When OsmO rings a restaurant to book your table, that's your voice on the line as well, not just on the calls it answers.
- One voice per account. You have a single custom voice at a time, not a library of them.
- You can re-record any time. If the first take is rushed, noisy, or just doesn't sound like you, record again and it replaces the old one.
- Cloning runs in the background. You don't have to sit on the screen watching a progress bar — OsmO tells you when it's ready.
- You preview before you commit — and you can preview it in each language your assistant speaks, so you hear how it sounds before any caller does.
Setting it up
The whole process is a few minutes, most of which is the recording itself.
- Open Custom Voice in the app's voice settings. You'll be asked for microphone access the first time — if you declined it earlier, you'll need to enable it in your phone's Settings before the recorder will start.
- Record your sample — about fifteen seconds, in a quiet room. Speak at your normal pace, the way you'd actually answer the phone. The clone reproduces your delivery, not just your timbre, so reading stiffly produces a stiff-sounding assistant.
- Let it clone. Generation happens in the background; you can close the screen.
- Preview, then switch. Listen to the result. Use it if it sounds right, re-record if it doesn't.
One tip from people who've done this: don't over-perform. The instinct is to use a bright, clear "phone voice," but the clone that people find most natural is the one recorded in an ordinary speaking register.
Does the caller know it's an AI?
Yes — and this is deliberate. OsmO's assistant identifies itself as an assistant answering on your behalf. It doesn't claim to be you.
The cloned voice is there so that disclosure lands naturally rather than jarringly. There's a meaningful difference between "you've reached a screening robot" and a warm, familiar voice saying it's picking up on your behalf and asking what the call is about. The first ends conversations; the second continues them.
This distinction matters for more than etiquette. Impersonating a real person on a call is the mechanic behind voice-cloning fraud, and it's not what this feature does or is designed to do.
The voice-cloning safety question
Voice cloning has a reputation problem, and it's earned — AI-cloned voices are now a standard tool in scam calls, most visibly in the "grandparent scam," where a cloned voice claims to be a relative in trouble. So it's fair to ask what happens to a voice you hand to an app.
Two things are worth separating here.
The clone in your OsmO account is scoped to your account. It speaks your assistant's side of your own calls. It isn't offered to other users and isn't handed out as a downloadable voice model.
The realistic cloning risk in the wild isn't apps — it's public audio. Criminals building a cloned voice sample it from material that's already public: social videos, podcast appearances, a company "meet the team" clip, or an outgoing voicemail greeting. A few seconds is enough with current tools. If you're worried about voice cloning generally, the useful defences are reducing how much of your voice is publicly posted, and agreeing a verification word with your family that a caller claiming to be in trouble has to produce.
There's a small, real upside here too: once an AI assistant is answering unknown calls, your recorded voicemail greeting stops being what strangers hear when they call you — which removes one of the easiest public samples of your voice from circulation.
If you'd rather not use your own voice
Custom Voice is optional, and plenty of people skip it. OsmO includes a set of preset voices, and there are good reasons to prefer one:
- You want it unmistakably clear to callers that an assistant is answering.
- You'd rather keep your voice out of any system, on principle.
- You just prefer how one of the presets sounds.
You can move between a preset and your custom voice whenever you like — it isn't a one-time decision.
Frequently asked questions
Can I make an AI answer my phone in my own voice?
Yes. OsmO's Custom Voice feature records a sample of your speech in the app, clones it, and uses that voice when the AI answers your calls. You get one custom voice per account and can re-record it any time. It is built for people who want a more familiar voice than the fixed system voices common in caller-ID and built-in screening tools.
Does the caller know they are talking to an AI?
OsmO does not pretend to be you. The assistant identifies itself as an assistant answering on your behalf — the cloned voice makes that interaction sound natural rather than robotic. The purpose is to stop callers hanging up, not to trick anyone into thinking they reached you personally.
Is voice cloning safe? Could a scammer use my cloned voice?
The clone OsmO creates is tied to your account and used only to speak your assistant's side of your own calls. It isn't downloadable as an audio model and isn't offered to other users. The broader voice-cloning risk comes from criminals sampling audio you've already published publicly — social videos, voicemail greetings — not from a clone held inside an app account.
How long does it take to clone your voice for OsmO?
The recording is about fifteen seconds. After that the clone generates in the background, so you don't wait on the screen. When it's ready you can preview it — in each language your assistant speaks — before switching over. If a take doesn't sound right, re-record and it replaces the old one.
Do I have to use my own voice?
No. Custom Voice is optional and OsmO ships preset voices you can use instead. You can switch between a preset and your custom voice at any time.
The bottom line
Call screening only works if callers actually talk to it. The voice is the part of the product that decides whether they do. Custom Voice exists because a familiar, natural voice gets people to explain why they called — and that explanation is the thing you actually wanted from screening in the first place.
It's included with OsmO Pro, which is US$4.99 a month with a 7-day free trial, on both iPhone and Android.
Related reading: All OsmO features · How AI call screening works · How to stop scam calls in 2026 · What is an AI phone assistant?