AI Voiceovers in Pencil, with ElevenLabs

Creative production found its rhythm with AI images and video. Audio never quite joined it.

Voiceover has lived outside the workflow: generated in a separate tool, downloaded, renamed, uploaded, and - all too often - lost the moment the project wrapped. The next campaign started the hunt all over again. Which file, which voice, which version. Sound was an asset everywhere except in your asset library.

That changes today. You can now generate AI voiceovers in Pencil using ElevenLabs models. And for the first time, voiceover can be saved to your asset library and reused across projects, like any other asset.

What's New

• ‍AI voiceover generation, powered by ElevenLabs, available in the Editor, Canvas, and Workflows.
• ‍Voiceover as a reusable asset. Every voiceover appears as a Generation and can be saved to your asset library. The read that worked for one campaign is one click away for the next.‍
• A dedicated Voiceover panel in the Editor, sitting alongside the Image and Video panels. Generate new VO, view past generations, and browse saved voiceover assets in one place.‍
• A Generate Voiceover node in Workflows, so spoken audio becomes a step in your production runs, not an errand outside them.

Three Models, Three Jobs

Pencil launches with three ElevenLabs models, and they split neatly by use case:

• ElevenLabs v3 is the most expressive of the family. Reach for it when the read is the creative: character, emotion, performance.
• Flash v2.5 is the fastest. Built for quick turnaround and high-volume runs where every generation counts.
• Multilingual v2 is the most stable and lifelike for narration, and the natural pick for campaigns that speak more than one language.

Default voices are provided by ElevenLabs, and per-voice controls are available in Advanced Settings.

How It Works

1. An admin enables Voice Generation for the workspace: Settings → AI Governance → Voice Generation. Models must be enabled before they appear in the platform.
2. Open the Voiceover panel in the Editor, or add the Generate Voiceover node to a workflow.
3. Select a model and a voice, and adjust voice behaviour in Advanced Settings if needed.
4. Style the delivery line by line, by writing audio tags in square brackets in your script. For example: "[whispers]", "[excited]" or "[angry]". You can add as many of these intonation tags as you need, to guide the voiceover styling line-by-line. 
5. Generate. The voiceover appears as a Generation, ready to save to your asset library and reuse later.

Good to know: for Agents, voiceover is currently available via Scribble.

What's Coming Next

• A full ElevenLabs voice library with categorisation. Browse and filter voices by age, gender, language and more.
• Voiceover generation in Feeds.
• Additional Advanced Features.

Why This Matters

Every asset type that lives outside the platform is a handoff: a version risk, a governance gap, and a tax on every iteration. Bringing voiceover into Pencil closes the loop on another one. Script, visuals, motion, and now voice, generated, reviewed, and reused in the same place, under the same controls.

‍

ElevenLabs voiceover generation is live in Pencil now.

Try it today: https://pro.trypencil.com/login

‍