Videos in your own voice, made automatically

How I turn a list of steps into a narrated screen video and a written guide, without recording a word.

Ask the questionCan AI make my how-to videos, in my own voice, while I do something else? Yes. Every video in this guide was made that way.

Every video in this guide was made the same way. I didn't sit at a microphone and I didn't screen-record myself clicking around. I wrote down the steps, and the system did the rest: it filmed the screen, read the script out in my own cloned voice, added music and a cover, and wrote the transcript you see under each video. One afternoon of setting up, then each new video costs me about ten minutes and a few cents.

This is the biggest time saver I've found for any business that explains things: how to use your product, how to book, what happens after a job, how to care for what you sold. Write it once, and you get a video and a written guide from the same words.

The stack

• ElevenLabs for the voice. You record a few minutes of yourself talking and it makes a voice that sounds like you. The Starter plan (about US$6 a month) does a quick clone and allows commercial use; the Creator plan (about US$22 a month) does the higher-quality professional clone I use. Roughly a thousand credits is a minute of speech.
• A screen recorder. If you'd rather click through it yourself, a desktop recorder like Recordly or OBS is fine. I use a small script instead that drives a hidden Chrome window, so the clicks are perfect every time and nothing pops up on my screen.
• Claude Code in the terminal. This is what writes and runs the scripts. You describe the video in plain English and it builds the recorder, generates the voice lines and joins everything up.
• ffmpeg, a free tool that joins the picture, the voice and the music into one video. Claude Code installs and drives it for you.

How one video gets made

1. Write the steps. Plain numbered steps, the same ones you'd give a customer. This is the only part that needs you.
2. Turn them into spoken lines. Claude rewrites each step as one short sentence you'd actually say out loud.
3. Make the voice. Each line goes to ElevenLabs and comes back as an audio clip in your voice. Clips are saved, so fixing one line never pays for the rest again.
4. Film the screen. The recorder clicks through the steps and notes the exact moment each step starts, so every voice line lands on its step and never talks over the last one.
5. Finish it. Music goes underneath and dips whenever you speak, a branded cover goes on the front, and it's sized for your site.
6. Document it automatically. The same lines become the written transcript and the how-to page. Change the steps, re-run, and the video and the page both update.

I want to make a short how-to video for my customers in my own voice, using ElevenLabs for the voice.

The video shows: [what the viewer learns, e.g. how to book a service on our website]
The page or app it happens on: [web address]
Who's watching: [e.g. a first-time customer on their phone]

Please:
1. Write the steps as a numbered list, 6 to 10 steps.
2. Write the narration: one short spoken sentence per step, plus a one-line hello and a one-line goodbye. Plain, friendly, Australian English. Spell out web addresses the way they're said out loud.
3. Estimate the length in seconds and the ElevenLabs credits it will use (about one credit per character).
4. Write the same steps as a short written guide I can put on my website.

Don't ask me questions. Make sensible assumptions and list them at the end.