A flat AI voice kills retention faster than bad visuals. These are my starting presets; change one slider at a time from here. They're my own settings, not official recommendations.
| Preset | Stability | Similarity | Style |
|---|---|---|---|
| Calm documentary | 60% | 80% | 10% |
| Energetic hook | 35% | 75% | 40% |
| Storyteller | 45% | 80% | 30% |
| Deadpan narrator | 70% | 85% | 0% |
Lower stability = more emotion but more random takes. Higher style = more drama but slower and occasionally odd.
Eleven v3: steer with one tag per sentence instead of sliders. Put the tag right before the words it should affect.
[urgent] So you've fallen through the ice. [deadpan] Here's the thing. The cold won't get you first. [whispers] Panic will.
Voice cloning rule: only clone your own voice, or one you have written permission to use.
Save this for later:
9 pages: the 1-hour workflow, 15 visual prompts, voice presets, caption rules, a real credit-cost sheet and a 30-day calendar.
Free. One useful email a week. Unsubscribe anytime.
Some links below may be affiliate links. If you subscribe through them I may earn a commission, at no extra cost to you. I've listed a free option next to each one.
| Job | Tool I use | Why | Free option |
|---|---|---|---|
| Visuals | Higgsfield | Cinematic motion and consistent characters | Pexels stock + Canva |
| Voice | ElevenLabs | The most natural narration I've tested | ElevenLabs free tier |
| Captions & edit | Submagic | Animated captions in minutes | CapCut auto-captions |
| Email list | Kit | Owning your audience | Kit free plan |








