I've been posting AI-narrated YouTube Shorts consistently for the past three weeks. My workflow has gotten stupidly simple. No Premiere. No DaVinci. No CapCut timeline wrestling. Just a browser, a script, and 60 seconds.
Here's exactly how I do it — step by step.
The Problem With Most "AI Video" Tutorials
Every tutorial I found online follows the same bloated workflow:
- Write a script
- Generate AI voice in one tool
- Export audio
- Open a video editor
- Find royalty-free background footage
- Import audio, align it manually
- Add subtitles by hand or use a separate subtitle tool
- Render and export
That's 7 steps too many. I quit twice before finding a faster way.
My Actual 60-Second Workflow
Step 1: Write (or paste) the script — 10 seconds
I keep a running list of hooks in my phone's Notes app. Reddit threads, shower thoughts, hot takes about my niche. When it's time to post, I grab one.
Example script I used last week:
"Most people think AI voices sound robotic. Five years ago, they were right. But in 2026, the best neural TTS engines can cry, whisper, and pause for dramatic effect. Here's what changed."
That's about 200 characters. Perfect length for a 30–45 second Short.
Step 2: Generate voice + subtitles in one click — 20 seconds
This is where the magic happens. I open VoiceIndex AI, paste the script, pick a voice (I rotate between Yunze for deep narration and Xiaoxiao DragonHD for emotional storytelling), and hit generate.
The key feature: it outputs both the audio file AND a perfectly time-synced SRT subtitle file. Automatically. No extra tool, no manual timestamps.
I download both files. Two clicks.
Step 3: Combine into a video — 30 seconds
Here's where people assume you need a full editing suite. You don't.
Option A — CapCut (mobile, free):
- Open CapCut on your phone
- Pick any vertical background from their stock library (satisfying clips, nature, gameplay — whatever fits your niche)
- Drop in the audio
- Import the SRT file — CapCut auto-positions the subtitles
- Export. Done.
Option B — Even faster with CapCut desktop:
- Drag background footage to timeline
- Drag audio to timeline
- File → Import Subtitles → select the SRT
- Subtitles snap perfectly because the timestamps are already accurate
- Export.
No manual alignment. No retyping. The SRT file does all the heavy lifting.
Why the SRT File Changes Everything
Most people underestimate this. Let me explain why auto-generated SRT is the single biggest time saver:
Without SRT: You paste your script into CapCut's auto-caption feature, wait for it to process, then spend 5–10 minutes fixing wrong words, adjusting timing, and reformatting. For every. Single. Video.
With pre-synced SRT: Import → done. The timestamps match the audio perfectly because they were generated together. Zero corrections needed.
Over 20 Shorts, that's roughly 3 hours saved on subtitles alone.
The Voices I Actually Use (And Why)
After testing dozens of options, I settled on a small rotation:
| Voice | Best for | Why I like it |
|---|---|---|
| Yunze (云泽) | Deep narration, documentary style | Calm authority, great for "did you know" hooks |
| Xiaoxiao DragonHD | Emotional storytelling | Natural pauses, genuine tonal shifts |
| Yunxi (云希) | Energetic explainers | Slightly faster pace, keeps attention |
| Andrew Multilingual | English content | Clean American accent, no uncanny valley |
All available for free on VoiceIndex AI — no sign-up, no credit card, no character limits that cut you off mid-sentence.
My Posting Schedule
I post 3–4 Shorts per week. Each one takes about 60–90 seconds to produce once the script is ready. I batch-generate 4 audio files on Sunday night, then assemble and post one per day Monday through Thursday.
Total weekly time investment: under 15 minutes for 4 videos.
Quick Tips From 3 Weeks of Doing This
- Hook in the first 2 seconds. Your AI voice needs to say something surprising immediately. No intros.
- Keep scripts under 300 characters. Shorts that run 25–40 seconds get the best completion rates.
- Use satisfying or mildly chaotic backgrounds. Soap cutting, pressure washing, subway surfing gameplay — sounds dumb, works incredibly well.
- Bold white subtitles with black outline. Don't get creative with subtitle styling. Readability wins.
- Post between 6–9 PM in your target timezone. For US audiences, that's 6–9 AM Beijing time.
Try It Yourself
The whole workflow costs $0 and requires zero software installation:
- Go to VoiceIndex AI
- Paste a script, pick a voice, generate
- Download audio + SRT
- Open CapCut, add background + audio + SRT
- Export and upload to YouTube Shorts
If you've been overthinking your content pipeline, stop. The best Shorts are raw, fast, and consistent. The AI voice is just the engine — your ideas are the fuel.
I'm documenting my AI content creation journey as I go. Follow for more real workflows, no fluff.
Top comments (0)