Can I keep using a former employee's voice in our intro message?

This page may contain affiliate links from which Blur Studio earns commissions.

A great intro message going away when its voice leaves the company is a real loss, and the instinct to "just clone the voice and keep going" is understandable. Before you do anything, it's worth knowing that the obvious-seeming path isn't actually open to you on the platform most people reach for.

It's also worth pausing on why this feels urgent in the first place. Usually it's one of two things: the message itself is genuinely well-written and you don't want to lose the copy, or the voice specifically carries warmth and familiarity that customers associate with your brand. Those have different fixes. If it's the writing, you can reuse the exact script with a new voice and lose nothing that actually matters to a customer hearing it for the first time. If it's specifically the voice, that's the harder problem this article is actually about, and it's worth being honest with yourself about which one you're solving before you spend time chasing consent for a clone.

The thing most people get wrong about this

ElevenLabs has two voice cloning tools with very different rules, and the mix-up matters. Professional Voice Cloning, the higher-fidelity option, can only ever be used to clone your own voice. ElevenLabs states this plainly: even with someone's consent, you cannot use Professional Voice Cloning on anyone's voice but your own, and the platform enforces this with a live identity-verification step the person being cloned has to complete themselves. There is no "get their written permission and clone it anyway" path through this specific tool. If your former employee isn't going to personally log in and verify their own voice, Professional Voice Cloning is off the table entirely, regardless of what's in their exit paperwork.

The tool that can clone a voice other than your own is the separate, lower-fidelity Instant Voice Cloning. It uses a self-attestation step, you confirm you have the right and consent to clone the voice, rather than a live verification from the voice's owner. That makes it technically possible to clone a former employee's voice from an old recording with their documented consent, but ElevenLabs' own guidance points toward the safer route being the opposite of what you'd assume: have the person create and verify the clone on their own account, then share it with you, rather than you uploading a recording of them yourself. If they've already left and aren't willing to do that, you're relying on a recording-based clone and a paper consent trail instead, which is weaker footing. The rules for both tools are covered in the ElevenLabs voice cloning guide.

Your real options, in order of how solid they are

1. Keep the original recording, with consent. If your former employee is willing to let you keep using the exact audio they recorded, get that in writing now, even if it was implied while they worked for you. This is the cleanest option and needs no cloning at all.
2. Have them create and share a voice clone themselves. If they're willing to help, ask them to set up the clone on their own account and share it with you. This keeps consent and verification with the actual voice owner, which is the setup every platform's own guidance points toward.
3. Clone from an old recording, with documented written consent. Technically possible through ElevenLabs' Instant Voice Cloning, but weaker ground: you're relying on a recording and a signed agreement rather than the person's live participation. Get written, specific consent for this exact use, not a vague agreement from years ago.
4. Record a new intro with someone else, or hire a voice actor. Often the fastest, cleanest option if the former employee isn't reachable or willing, and it sidesteps the consent question entirely.

What "written consent" should actually say

A vague email saying "sure, that's fine" isn't the kind of consent that holds up if the relationship later turns sour or someone asks questions. A document worth relying on should specify, in plain language: exactly what's being used (the original recording, or a clone made from it), exactly what it will be used for (your current intro message, specifically, not "any future company messaging"), how long the permission lasts (indefinitely, or until a stated date or event), and that it can be withdrawn with reasonable notice. If you're paying them anything for this, even a small one-time fee, put that in writing too, since a paid consent is generally sturdier ground than an unpaid favor if it's ever questioned later. None of this needs to be a formal legal contract drafted by a lawyer for a single intro message, but it should be specific enough that a reasonable person reading it later would agree on exactly what was permitted.

Worth knowing: several places have specific laws governing a person's voice and likeness, separate from general consent principles, and these vary by where the person is based, not necessarily where your company is. If this intro message is going out broadly, on a website, in ads, to a large customer base, rather than staying internal, it's worth a few minutes with a lawyer who can confirm your specific situation, rather than assuming a generic consent email covers every jurisdiction your former employee might have standing in. Other tools' rules are compared in best AI voice cloning tools and their consent rules.

A worked example, start to finish

Say your former sales lead, Maria, left on good terms six months ago, and her recorded intro ("Hi, I'm Maria, and I'd love to help you find the right plan") still plays on your website's demo-booking page. Here's how the decision actually plays out: you message her, explain you'd like to keep the recording or make a voice clone for new lines, and ask directly. If she says yes to keeping the exact recording, you're done, get her agreement in a short email specifying this one use, and you're on solid ground with no cloning involved at all. If she's open to a clone but wants to help rather than hand over an old file, she creates the account, records a fresh sample herself, verifies it, and shares access with you, which is the version that holds up best if anyone later asks how it was made. If she agrees to a clone from the old recording but doesn't want to do the setup herself, that's option 3 above, technically available but weaker ground, and worth a specific written agreement rather than a casual yes. If she doesn't respond, or you'd rather not ask, you record a new version with a current team member, which sidesteps every question above and usually takes less time than chasing a consent conversation with someone who's moved on.

What I'd tell you if you asked me directly

Don't clone someone who left on bad terms, even if you technically have an old signed form that might cover it. The legal and reputational risk of a disgruntled former employee's voice circulating in your company's messaging isn't worth whatever continuity you'd gain. If the relationship ended well and they're willing to help, option 2, them creating and sharing the clone themselves, is worth the extra ten minutes of their time, since it's the version that holds up if anyone ever questions how the clone was made. If they're not reachable or not willing, re-recording with someone new is almost always the better call than fighting for a weaker, recording-based clone of someone who's moved on. The same live-verification rule applies if you want to give an AI receptionist your real receptionist's voice.

When to skip all of this

If this is a one-off intro message, the cleanest path above will take you a day or two, not a project. Skip the whole cloning question entirely if the message isn't actually that important to keep in that exact voice, a surprising number of "we have to keep this voice" situations turn out, on reflection, to be more about not wanting to deal with the hassle of re-recording than any real attachment customers have to that specific voice. If customers are responding to the warmth and clarity of the message rather than literally recognizing who's speaking, a new recording with someone else, written in the same style, usually gets you 90% of the value with none of the consent complexity. For a founder's voice, see getting an AI voice that sounds like your founder.

If you're trying to build something bigger, a system that preserves and manages voice assets properly as people join and leave your company, with real consent tracking, expiry dates and an audit trail rather than a one-time favor chased down over email, that's a genuine custom build, not a single voice clone. This comes up more than people expect once a company has several recorded voices across different tools (an IVR greeting, a demo video, a receptionist line) and no central record of who consented to what, for how long. I build that kind of thing, and if that's closer to what you actually need, tell me what you're trying to solve and I'll tell you plainly whether it's worth building or whether a simple shared document is all you actually need right now.