Make a Faceless YouTube Tutorial with AI Voice and Screen Recording (Step by Step)

By The Keep or Skip AI team · 2026-11-16

This post may contain affiliate links. We may earn a commission at no extra cost to you. Learn more.

Short answer: Write the script first, generate the voiceover in ElevenLabs (use Starter, $6/month or $5/month billed yearly, once you publish, since that's the plan that lists a commercial license), then record your screen in OBS Studio (free and open source) while you play the narration back, section by section. Cut the two together in any video editor, add captions, and upload. Under YouTube's current help page, an AI voiceover on a real screen recording generally doesn't trigger the "AI use" disclosure unless the video shows something realistic that didn't actually happen, but read the rules yourself before you publish.

How we wrote this: Tool details come from ElevenLabs pricing, OBS Studio's site, and YouTube's help article on disclosing use of GenAI content, all checked November 16, 2026. The workflow is a production method; we haven't published our own channel's results here and make no claims about views or earnings. For a price comparison of other voice tools, see our AI voiceover tools roundup.

Why narration-first works better for tutorials

Most beginners record their screen, then try to talk over it. The result is "um, so now I'm going to… wait…" or a voiceover that runs ahead of the clicks. Writing the narration first and generating it with an AI voice flips that:

What you'll need

Tool Job Cost (November 16, 2026)
OBS Studio Screen recording Free and open source (Windows 10/11, macOS 13+, Linux)
ElevenLabs AI voiceover Free: 10,000 credits/month; Starter: $6/mo or $5/mo billed yearly, 30,000 credits/month, commercial license
Any AI assistant Script drafting and tightening Free plans work
A video editor Cutting voice and screen together Whatever you already use; Descript and Canva are covered in our other posts

Sources: ElevenLabs pricing and OBS Studio, checked November 16, 2026. ElevenLabs text to speech uses about 1 credit per character, and Starter's card lists a commercial license and instant voice cloning. Our voiceover roundup recorded the free plan as non-commercial with attribution, so plan to be on a paid plan before you publish on a monetized channel.

Step 1: Pick a tutorial that fits the format (10 minutes)

Faceless screen tutorials work best for "how do I do X in Y software" topics: setting up a tool, fixing a common error, building something specific. They work poorly for opinion pieces, which usually benefit from a face or a strong personal voice.

Good: "How to set up a Kit signup form in 5 minutes." Weak: "Is email marketing dead?"

Step 2: Write the script in sections (30 minutes)

Write in short sections, one per screen action. Each section should be 2–4 sentences. Use this template:

[HOOK – 10 sec] What the viewer will have by the end. [SECTION 1 – screen: open the app] ... [SECTION 2 – screen: click Settings > Forms] ... [SECTION 3 – screen: ...] ... [RECAP – 15 sec] What they did, and one next step.

Copy-paste prompt to tighten it: Here's my YouTube tutorial script in sections. Tighten it for spoken narration: short sentences, no filler, plain words, each section under 4 sentences. Keep every step and every on-screen label exactly as I wrote it. Don't add steps or claims. Flag anything unclear. [PASTE SCRIPT]

Read it out loud once. If you stumble, a listener will too.

Step 3: Generate the voiceover (15 minutes)

  1. In ElevenLabs, open text to speech, pick a voice, and paste one section at a time.
  2. Generate, listen, and regenerate only the sections that sound off. Every regeneration uses more credits.
  3. Download each section as a separate audio file named in order: 01-hook.mp3, 02-open-app.mp3, and so on.

Credit math: a 1,000-word script is often around 6,000 characters including spaces. At roughly 1 credit per character, Starter's 30,000 monthly credits cover several scripts that length if you don't regenerate heavily. That's arithmetic from ElevenLabs' published rate, not a measured test.

Pronunciation tip: spell product names phonetically in the voice script if the voice mispronounces them, but keep the correct spelling in your captions.

Step 4: Record the screen to match the narration (30–45 minutes)

  1. In OBS Studio, add a Display Capture or Window Capture source for the app you're demonstrating.
  2. Set your screen to a clean state: close notifications, hide bookmarks, use a test account with no private data.
  3. Play section 1's audio through headphones and perform the actions at the narration's pace. Stop recording.
  4. Repeat for each section. Short clips are much easier to fix than one long take.
  5. Move the mouse slowly and pause briefly after each click so viewers can follow.

Before you hit record, double-check that nothing private is visible: email inboxes, API keys, customer names, browser autofill.

Step 5: Edit voice and screen together (30–60 minutes)

  1. Place audio files on the timeline in order.
  2. Drop each screen clip under its matching narration. Trim or speed up dead time (loading screens) so visuals match the voice.
  3. Add zoom-ins on small UI elements; viewers on phones can't read tiny menus.
  4. Add on-screen text for key settings or values.
  5. Add captions. Many editors can auto-generate them; correct product names by hand.
  6. Export at the resolution you recorded in.

Step 6: Upload and handle YouTube's AI disclosure

YouTube's help page says creators must disclose AI-generated or meaningfully altered content that seems realistic, using the "AI use" setting during upload. Its examples of what you don't need to disclose include production assistance such as using AI to create or improve a script, thumbnail, or title, voice or audio repair, and cloning your own voice for voiceovers. Its examples that do need disclosure include realistic depictions of events that didn't happen.

For a screen tutorial with an AI voice, that usually means the content itself (a real recording of software) isn't a realistic fabricated scene. But YouTube's page doesn't list "stock AI voice narrating a real screen recording" as an explicit example either way, so read the current help article and decide for your video. When in doubt, disclosing is the cautious choice.

Also review YouTube's channel monetization policies before applying to monetize; low-effort or repetitive content is a common concern with faceless channels.

Pre-publish checklist

Who this workflow is for (and who should skip it)

Keep or skip? Our verdict

These verdicts are based on published plans and YouTube's help pages, not on our own channel's results.

Common mistakes

FAQ

Is OBS Studio really free? OBS describes itself as free and open source software for video recording and live streaming, available for Windows, macOS, and Linux.

Can I clone my own voice instead? ElevenLabs lists instant voice cloning on Starter. YouTube's help page lists cloning your own voice for voiceovers among the examples that don't require disclosure.

Do I need an expensive editor? No. Any editor that handles multiple audio and video tracks works. If you want to edit by transcript, Descript is covered in our voiceover tools roundup.

How do I make thumbnails and channel art? See Canva vs Adobe Express. For a broader list of free tools, see Best free AI tools for side hustlers.

Disclosure: some links on this page are affiliate links. If you buy, we may earn a commission at no extra cost to you. See our disclosure.

Get the free Keep or Skip AI Starter Stack PDF

The AI tools we'd keep (and skip) to run a small business in 2026, with prices.