How I Created an AI Clone of Myself Using HeyGen and ElevenLabs (Step-by-Step for Real Estate Pros)
- Jerad Larkin
- 17 hours ago
- 5 min read
Can you actually create a realistic AI clone of yourself using your own voice and video?
Short answer: yes. And if you do it the right way, it can save you a massive amount of time creating content. I want to walk you through exactly how I built my own AI clone using HeyGen for video and ElevenLabs for voice. I’ll also share the mistakes I made, what worked, what didn’t, and how I’m actually using this in my real estate and title marketing today.
This is not theory. I went through this process five different times before I got a result I was happy with.
Why I Decided to Create an AI Clone of Myself
As someone who teaches classes, hosts events, records videos, and constantly creates content for real estate agents, time is always the bottleneck.
I still believe authentic video performs best. But I also know I do not need to physically record every single short educational video myself anymore.
My goal with AI cloning was simple:
Create short-form educational videos faster
Still sound like me
Still look like me
Use my real voice, not a robotic AI voice
Layer in B-roll, captions, and screen recordings afterward
When done right, the final product is shockingly good.
The Two Tools That Actually Matter
There are a lot of AI avatar tools out there. After testing multiple platforms, here is what I landed on.
HeyGen for Video Cloning
HeyGen handles the visual side. This is where your AI avatar lives.
You upload a training video of yourself, and HeyGen builds a digital avatar that matches your facial expressions, lip movement, and on-screen presence.
You can start with the free plan to test things out. Eventually, if you want to use this seriously, you will need to upgrade to the $29 per month plan.
My advice: start free, learn the system, then upgrade when you hit limits.
ElevenLabs for Voice Cloning
HeyGen’s built-in voices were not good for me. Some people get lucky. I didn’t.
That’s where ElevenLabs comes in. ElevenLabs lets you create a professional-grade clone of your real voice. When you combine this with HeyGen, you get the best of both worlds. Expect another $22 per month once you move into professional voice cloning. Yes, the costs add up. But if this replaces hours of filming and editing every month, it pays for itself fast.
How to Record Your Training Video (This Matters More Than You Think)
This is where most people mess things up.
I tested multiple setups, including a Sony A7S III with a full camera rig. Ironically, the avatar I actually use today was created from an iPhone video.
Here is what mattered most.
Camera Setup
iPhone on a tripod shooting in 4K worked extremely well
DSLR is great but not required
Do not shoot handheld
Keep your framing consistent
Audio Setup
Clean audio matters more than perfect video
I used DJI lav mics, but any quality lav mic works
Avoid background noise
Body Movement Tips
Do not move your head too much
Try to keep your upper body fairly still
Hand gestures are fine, but less is better
Look directly into the lens the entire time
How Long to Record
HeyGen says two minutes minimum.
That is not enough.
I recommend recording at least five minutes of continuous footage. More footage gives the AI more data, which improves realism.
I literally set a timer on my watch and talked for five minutes straight.
The Biggest Mistake I Made Early On
HeyGen instructs you to pause between sentences when recording.
In my experience, that made the avatar look robotic and unnatural.
Once I stopped pausing and just talked like myself, the quality improved dramatically. The goal is not perfection. The goal is realism.
Creating Your Avatar Inside HeyGen
Once you upload your video, HeyGen will:
Review your footage
Score it based on quality
Confirm identity and consent
Process your avatar
This can take up to an hour.
Pro tip: Sometimes avatars fail to process. I had this happen multiple times. If it does, contact HeyGen support. They were helpful once I reached out. When it works, your avatar will appear in your dashboard and be ready to use.
Why I Don’t Use HeyGen’s Default Voice
This is important. Even if your avatar looks great, a bad voice will ruin everything.
HeyGen’s default voice did not sound like me. It was flat and quiet.
That is why I moved everything over to ElevenLabs.
How I Built My Voice Clone in ElevenLabs
Inside ElevenLabs, you will create a professional voice clone.
Here is what I learned.
How Much Audio You Need
Minimum required: 30 minutes
What I recommend: 50 minutes to 2 hours
More audio equals better results.
I extracted audio from existing videos and tutorials I had already recorded. You can do this easily using tools like CapCut.
The key rule: clean audio, no background noise, only your voice.
Fine-Tuning the Voice
Once your voice is created, you can adjust:
Stability
Similarity
Style exaggeration
Volume and speed
I kept things subtle. I wanted it to sound like me on a normal day, not a dramatic version of me.
Connecting ElevenLabs to HeyGen
This step is technical but simple.
Generate an API key inside ElevenLabs
Paste that key into HeyGen
Import your custom voice
Select it as your default voice
Once connected, your avatar will speak in your real voice.
This is the moment everything clicks.
How I Create Scripts That Sound Like Me
I do not manually write scripts anymore.
I use ChatGPT with a custom project trained on how I speak, teach, and explain things. I generate short scripts, copy and paste them into HeyGen, and preview the audio before generating video. Always listen to the audio first before generating the full video. Fix pacing and wording early to save credits.
Generating the Final Video
When you are ready:
Choose 1080p resolution
25 frames per second
MP4 format
Turn off the watermark
Submit
Processing takes a few minutes. Once complete, download the file.
At this stage, the video is usable. But I take it one step further.
How I Edit AI Videos to Make Them Look Real
I use CapCut for editing.
Here is my workflow:
Import the AI video
Remove the background
Resize the avatar smaller on screen
Add screen recordings or B-roll underneath
Add captions
Trim awkward pauses
This is where the magic happens.
Once you layer in B-roll, captions, and motion, it becomes very difficult to tell it was AI-generated.
When This Makes Sense for Real Estate Professionals
I want to be clear.
This does not replace authentic, in-person video. It supplements it.
This works extremely well for:
Educational tips
Market explanations
Tool walkthroughs
Short social videos
Consistent content without constant filming
If you are already recording long-form videos, this is a force multiplier.
Final Thoughts on AI Cloning in 2025
We are at a point where AI video cloning is good enough when used correctly.
Is it perfect? No.
Is it powerful? Absolutely.
If you follow the steps above, avoid the mistakes I made, and layer in smart editing, you can create content faster without sacrificing your voice or your brand.
Questions? Contact:
If you want more real estate tools, marketing ideas, and AI walkthroughs like this, I share everything I’m testing in real time.
Subscribe at MileHighTitleGuy.com/subscribe for access to tools, resources, and invites to my Colorado classes and events.
Questions or want help setting this up?
Jerad Larkin Chicago Title Colorado
📞 303.630.9430

