How To Use ElevenLabs
Generate, clone, and dub voices that actually sound human.
Overview
ElevenLabs is the gold standard for AI voiceover. This walks you through generating natural speech, cloning your own voice safely, and dubbing video into 30+ languages.
Step by step
- 1
Pick a voice
Browse the Voice Library — filter by gender, accent, age, and use case. 'Conversational' voices are best for narration.
- 2
Generate your first clip
Paste a script (~150 words). Adjust Stability and Style sliders — lower stability = more expressive, higher = more consistent.
- 3
Clone your own voice
Upload 1–5 minutes of clean audio of yourself. Use it only on accounts and content you own — read the consent rules.
- 4
Dub a video
Upload an MP4, pick a target language, and ElevenLabs translates AND keeps your voice. Great for short-form distribution abroad.
- 5
Export and edit
Download as MP3 or WAV. Bring into Descript or your editor for the final mix.
Tips
- Write punctuation like a director would — commas, ellipses and line breaks shape the delivery.
- Use the v3 alpha model for emotional reads; v2 for long-form narration consistency.
- Generate three versions and pick the best — it costs almost nothing.
Common mistakes
- • Cloning a voice that isn't yours or that you don't have written permission to use.
- • Cranking Style to 100% — output gets unstable and slurs words.
- • Using long run-on sentences without punctuation — pacing collapses.