How to Make YouTube Shorts with AI in Under 5 Minutes

Pricing update: Kling Standard is $8.8/month. Suno Pro is $10/month. ElevenLabs Starter is $6/month (Source). The budget table below reflects current prices.

Let me show you exactly how it works.

The Full AI Workflow

The workflow has six steps. Script, voiceover, captions and assembly all have a free option. Visuals and music are where the free tiers run out: Kling’s free Basic plan includes no monthly credits (Source), Pika’s free plan is 0 credits a month (Source), and Suno’s free plan comes with no song downloads (Source). Free stock footage covers the visuals gap. AI music means a paid plan. I’ll note the paid upgrades as we go.

Step 1: Script Step 2: Voiceover (ElevenLabs, free tier) Step 3: Visuals (Pika or Kling for B-roll, or stock) Step 4: Captions (auto-caption in CapCut) Step 5: Music Step 6: Assembly (CapCut or Descript)

The AI generation steps are fast. ElevenLabs generates audio in seconds (Source). CapCut produces captions instantly (Source). You are writing a 45-second script, not a television treatment. The manual inputs stay brief because the output is brief. The only step with real wait time is AI clip generation in Kling or Pika; use Pexels stock footage for the visuals and the whole workflow lands under 5 minutes.

I’m going to walk through each step.

Step 1: Script

Shorts can run up to three minutes (Source).

The mistake most people make is writing a full script and then chopping it down. Better approach: write specifically for the format from the start.

Here’s the ChatGPT prompt template I use:

“Write a YouTube Shorts script (45-55 seconds when spoken aloud, conversational pace) about [TOPIC]. Format: hook in the first 3 seconds that creates curiosity, 3-4 punchy information points, strong CTA at the end. No intro, no ‘Hey guys,’ no padding. Make it sound like a knowledgeable person talking, not a script being read.”

For a Short about AI tools, I’d fill in: “the 3 most useful free AI tools most people don’t know about.”

I edit for my voice – add a specific example, change something that sounds too formal, cut a sentence that’s redundant.

Every sentence should make the viewer think “wait, I need to hear the next part.” Write the script with that in mind – front-load the value, don’t save the good stuff for the end.

Step 2: Voiceover with ElevenLabs

ElevenLabs generates synthetic voiceovers in various voices (or your cloned voice) from text. The free tier gives you 10,000 characters per month – more than enough for Shorts.

Workflow:

1. Go to ElevenLabs, create a free account
2. Navigate to “Text to Speech” (Source)
3. Paste your script
4. Select a voice
5. Adjust stability and similarity sliders (start at default, adjust if the output sounds too robotic or too erratic)
6. Generate and download

The quality on ElevenLabs free tier is genuinely good. For most audiences, you can’t distinguish it from a human voiceover at normal playback speed. In the Shorts context – phone speakers, often with captions on – it’s more than sufficient.

One important note: ElevenLabs’ free tier does not include commercial use rights. If you’re monetizing your YouTube channel or using the voiceover in paid content, you need the Starter plan at $6/month minimum (Source).

Optional upgrade: ElevenLabs Starter at $6/month (Source) unlocks commercial rights and voice cloning (train it on your voice, generate voiceovers that sound like you). For building a consistent presence without having to record every video yourself, the voice clone is compelling. See our full How to Clone Your Voice with AI guide for the complete step-by-step process, and the ElevenLabs Review 2026 for detailed pricing and feature breakdown. Want to compare all major AI voice tools? The Best AI Voice Generators 2026 roundup covers everything.

Step 3: Visuals – AI-Generated B-roll

This is the step most guides skip over. You have voiceover audio. What do you show on screen?

Option A: Show yourself on camera (traditional talking-head). Valid, no AI needed for visuals.

Option B: Generate AI visuals using Pika or Kling. This is what makes the workflow fully AI-native.

For each sentence or concept in your script, generate a matching 3-5 second video clip:

Using Kling AI (recommended for realism):

• Go to Kling AI – the free Basic plan includes No monthly Credits, so you need a subscription or a credit purchase (Source)
• Prompt: describe the visual you want – “aerial drone shot over a modern city at night with glowing lights” or “close-up of hands typing on a keyboard, soft blue light” – short, descriptive, specific
• Generate at 5 seconds
• Download

Using Pika (recommended for stylized/animated look):

• Go to Pika
• Same prompting approach, but add a style modifier: “anime style” or “illustrated, painterly” if you want a non-realistic aesthetic

For a 45-second Short, you need roughly 8-10 clips.

Alternative: Free stock footage. Pexels Video (pexels.com/videos) has free HD and 4K footage. Not AI-generated, but completely free and often faster than generating the exact clip you need. For abstract topics, stock footage is underrated.

See our Pika vs Kling comparison for more detail on which platform fits which use cases.

Step 4: Captions

Non-negotiable in 2026.

CapCut (free, recommended): Import your video, click “Auto Captions,” select language, done. The accuracy is high on clean voiceover audio. CapCut also offers style presets for captions – the animated word-by-word highlight style that’s common on viral Shorts.

Descript (free tier, 1 hour/month): Import your voiceover audio, generate transcript, export with burned-in captions. Descript’s caption export is cleaner for professional formatting.

For Shorts specifically, I like CapCut’s caption styles better – they’re tuned for the format. For longer YouTube videos, Descript is more flexible.

Caption placement: center screen works well for Shorts viewed on phone. Don’t put captions too low (they get cut off on some devices) or too high (conflicts with the title area).

Step 5: Music with Suno

Background music changes the feel of a Short more than most creators realize. Silence under a voiceover sounds clinical. The right background music makes the same script feel energetic, emotional, or professional.

Suno generates original music from text prompts. The free plan gives you 50 credits per day, but it comes with no monthly song downloads and no commercial rights (Source). That makes it a place to audition ideas, not a source for your Short. To download a track and use it on a monetized channel, you need the Pro plan ($10/month) for commercial use.

My prompting approach:

• Keep it simple: “upbeat background music, no lyrics, positive energy, corporate”
• Or: “ambient electronic, minimal, tech-focused, no lyrics”
• Or: “light acoustic guitar, warm, approachable”

Generate 3-4 options and use the one that fits your specific content.

The important thing: keep the music volume low. The voiceover should be clearly dominant. Music is texture, not centerpiece.

Step 6: Assembly in CapCut

CapCut is free, available on both desktop and mobile, and has become the de facto standard for Shorts and TikTok editing. It’s genuinely good for this format.

The assembly process:

1. Create a new project, select 9:16 ratio (vertical, standard for Shorts)
2. Import your video clips (AI-generated B-roll or stock)
3. Import your voiceover audio – lay it on the audio track
4. Trim video clips to match voiceover pacing
5. Add captions (use CapCut’s auto-caption feature)
6. Import background music track, set volume low
7. Add transitions (subtle: fade or cut. Skip the spinning star transitions)
8. Export at 1080x1920 (1080p vertical)

Budget Options at Every Step

For completeness, free options at each step:

Script, voiceover, captions and assembly run at $0. Visuals run at $0 only if you use free stock footage. AI B-roll and AI music need a paid plan.

Common Mistakes (and How to Avoid Them)

Mistake 1: Making the hook too slow. Start with the most interesting thing you’re going to say – not a greeting, not an intro, not “today I’m going to tell you about.” Jump straight in.

Mistake 2: Generating too many visual styles. If your B-roll mixes realistic Kling clips, anime-style Pika clips, and stock footage, the visual inconsistency is jarring. Pick one aesthetic and stick with it for the whole Short.

Mistake 3: Overloading captions. Captions should feel like they’re helping, not competing with the video. Keep fonts clean, size readable, avoid distracting animations for every single word.

Mistake 4: Setting music too loud. If I can’t understand the voiceover because the music is competing, I’m out. For catching this before you upload, monitor your final mix on proper headphones rather than laptop speakers — laptop audio masks audio balance problems that headphones reveal immediately. Our best noise-cancelling headphones guide covers options across every budget for content creators.

Mistake 5: Skipping the iteration step. Generate your first Short and post it. Then generate another one next week. The improvement from iteration one to iteration ten is dramatic. Don’t optimize endlessly before starting.

Getting Better Results from AI Video

The biggest variable in AI-generated Shorts quality is prompt specificity. “Show a futuristic city” produces generic output. “Aerial drone shot above a glowing neon-lit city at night, fog settling in the streets, cyberpunk aesthetic” gives the AI something to work with.

Specific details that help:

• Camera angle (aerial, close-up, wide shot, side profile)
• Lighting (golden hour, neon, studio lighting, backlit)
• Motion (slow pan, static shot, zoom in)
• Aesthetic (realistic, anime, illustration, cinematic)
• Time of day / environment

Good prompt, good output. Vague prompt, vague output.

What You’ll Need (Complete Checklist)

• ChatGPT free account (script)
• ElevenLabs free account (voiceover)
• Kling AI paid plan (video B-roll; the free Basic plan includes no monthly credits)
• Pika paid plan or credit pack (stylized B-roll, optional; the free plan is 0 credits/month)
• Suno Pro (music; the free plan has no song downloads)
• CapCut free (assembly and captions)
• A YouTube channel

The Honest Part

This workflow produces good Shorts, not great Shorts. The ceiling is lower than what you’d create with your own on-camera presence, professional audio, and original footage.

What AI workflow gives you: speed, accessibility, and the ability to produce content consistently without recording yourself every time.

What it doesn’t give you: your personality on screen, the kind of parasocial connection that builds loyal audiences, the spontaneous moments that make video feel real.

The best creators I’ve watched build Short-form audiences combine both: AI-assisted workflow for speed and volume, with real on-camera moments mixed in for connection. The tools do the production work; you bring the perspective and voice.

Start with the AI workflow to build the habit and test what topics your audience responds to. Add more of yourself as you find your footing.

For the detailed comparison of Pika vs Kling specifically, see Pika vs Kling AI: Budget AI Video Generators Compared. For the full landscape of AI video tools at every price point, Best AI Video Generators 2026 has the complete picture. And if you want to go deeper on AI video for marketing content, How to Create Marketing Videos with AI covers the longer-form workflow.

For pricing breakdowns on the video generation tools mentioned in this guide: Kling AI Pricing 2026 | Pika Pricing 2026

FAQ