ElevenLabs Voice Clone for Videos: How I Voice My Shorts With My Own Voice

Quick answer

An ElevenLabs voice clone is an AI copy of your voice that reads any script you type. Instant cloning needs a few minutes of audio. Professional cloning needs 30 minutes or more and gives the best quality. I use a Professional clone to narrate my short videos, with stability 0.5 and similarity 0.75.

Key takeaways

  • I made a Professional Voice Clone of my own voice and use it to narrate my short-form videos.
  • Instant cloning is fast and needs little audio. Professional cloning needs much more and sounds closer.
  • Clean training audio matters more than anything else.
  • My settings are stability 0.5 and similarity 0.75. Test and adjust for your own voice.
  • Only clone a voice you have the rights to, and label AI audio where the app asks.

Tools in this guide

ElevenLabs

Clone your own voice and let it narrate your videos.

Affiliate link: I earn a commission if you sign up or upgrade, at no extra cost to you.

Try ElevenLabs →

I do not like recording voiceovers. I film the video, then I have to sit down, set up a mic, and read the script five times until it sounds right.

So I made an ElevenLabs voice clone of my own voice. Now I film silent B-roll, type the script, and the AI voice narrates the video. This is how it works, what I'd tell you to watch out for, and where it falls short.

What is an ElevenLabs voice clone?

It is an AI copy of a voice. You give ElevenLabs recordings of someone speaking. It builds a voice from them. After that you type text and it reads the text in that voice.

I use mine for short-form affiliate videos. I film silent video, then the AI voice does the narration. I do not record each script myself.

What is the difference between Instant and Professional voice cloning?

ElevenLabs offers two kinds. Here is how they compare, based on their site.

Instant clone Professional clone
Audio needed About 1 to 5 minutes 30+ minutes, up to about 3 hours for best quality
Speed Ready in seconds Takes more setup
Quality Good for quick use Their highest-quality option
Verification Not stated Voice verification
Best for Trying it out A voice you will use a lot

I went with Professional because I plan to use the voice on a lot of videos. If you only want to try it, start with Instant. Check which plans include each type, because that can change.

How do I record good training audio?

The clone can only be as good as what you feed it. Bad audio makes a bad clone. Follow these steps.

  1. Pick a quiet room. Soft things like a closet full of clothes or a room with a rug cut echo.
  2. Use the same mic every time. Do not mix a phone, a headset and a laptop.
  3. Keep the same distance from the mic. A hand width or two works.
  4. Speak the way you will want it to sound. If your videos are upbeat, record upbeat. The clone copies your style.
  5. Read real text, not a stiff script. Read out loud from something you wrote, a blog post or notes.
  6. Leave out background noise. No fans, music, or typing.
  7. Listen back before you upload. If you hear echo or hum, re-record.

More clean audio is better than a lot of messy audio.

What settings do I use?

My settings are stability 0.5 and similarity 0.75. That is what I like for my own voice.

The ElevenLabs page I checked did not explain these two sliders, so treat this as my preference, not a rule. In my experience, a lower stability sounds more varied and a higher one sounds flatter. A higher similarity stays closer to the training voice. Your voice and your scripts may want different numbers.

Here is how to find yours.

  1. Pick one short script you know well.
  2. Render it at my settings.
  3. Change one slider at a time and render again.
  4. Keep the version that sounds most like you talking to a friend.

What are the honest limits?

This is useful, but it has real downsides.

It can mispronounce names

Business names, people's names and odd words can come out wrong. I listen to every render. When a word is off, I respell it in the script until it sounds right.

Some platforms want an AI label

Apps like Meta and TikTok ask creators to label AI-generated content, and that can include AI audio. Rules change often. Read the current policy on each app, and use the label when it applies. Do not hide it.

You must have the rights

Only clone a voice you have the right to clone. For me, that is my own. If you want a different voice, get clear permission in writing, or use a stock voice.

It does not replace your face

A clone gives you a voice. It does not give you a personality on camera. For videos where people want to see you, record yourself.

The first version is rarely the best

Expect a few tries on every script. A quick listen and a rewrite saves a bad post.

When is a voice clone worth it?

It works well when you:

  • Make a lot of short videos with the same voice
  • Hate recording audio
  • Film B-roll and want the narration to be separate
  • Need to fix a line without re-recording

It is less useful if you only make one or two videos a month. In that case, recording yourself is faster than setting up a clone.

Bottom line

A voice clone lets me film silent video and have my own voice narrate it. ElevenLabs gives you two ways to do it: Instant for a quick test, Professional for the best quality. Record clean audio, start from stability 0.5 and similarity 0.75, and tune from there. Listen to every render for mispronounced names, follow each app's AI-label rules, and only clone a voice you have the rights to.

Frequently asked questions

How much audio do I need for an ElevenLabs voice clone?

ElevenLabs lists about 1 to 5 minutes for an Instant clone and 30 or more minutes for a Professional clone. Their page says about 3 hours gives the best Professional quality. Check their current guidance before you record.

What is the difference between Instant and Professional voice cloning?

Instant is quick and takes a short sample. Professional uses much more audio and is described by ElevenLabs as their highest-quality cloning option, with voice verification. Professional takes more effort up front.

Can I clone someone else's voice?

Only if you have the rights to do it. That means your own voice, or a person who gave you clear permission. Do not clone a customer, a celebrity or anyone who has not agreed.

Do I have to tell people the voice is AI?

Platforms like Meta and TikTok ask creators to label AI-generated content, including audio. Rules change, so read the current policy for each app and use the label when it applies.

Will the AI pronounce every word right?

No. Names, brands and unusual words can come out wrong. Listen to every render, and respell the word in your script until it sounds right.

ElevenLabs

Clone your own voice and let it narrate your videos.

Affiliate link: I earn a commission if you sign up or upgrade, at no extra cost to you.

Try ElevenLabs →
Rob P.

Builds AI systems for small businesses and tests AI tools on camera. About Rob.