My brand voice framework and the exact scripts, tags and templates I used to audition a voice for my new company, Avori.
Every brand picks its colors, fonts and logo on purpose. Almost nobody picks what their brand sounds like. So for Avori's walkthroughs, onboarding videos and demos, I auditioned a voice with Eleven v4 from ElevenLabs.
Eleven v4 performs a script instead of just reading it. It picks up tone from how you write, you direct it with plain-English tags, and the voice stays the same across every take. Use Eleven v4 for anything you produce ahead of time, and Eleven v4 Turbo for real time, like a support agent.
Write down what your brand should sound like before you hear a single voice. Fill this in and keep it with your brand files.
The easiest path is Voice Design in ElevenLabs: describe a voice in words and it creates a brand-new one that's yours. You can also browse the Voice Library's recommended voices, or clone a voice actor with their written permission (an Instant Voice Clone needs about 10 seconds of clean audio).
Write three descriptions that push in different directions, then have each candidate read the same script so you're comparing voices, not words.
Tags are directions in square brackets, right in your script. Some set the performance, like [calm and reassuring]. Others trigger a moment, like [laughs] or [whispers].
Pick a handful of tags as your brand's vocabulary and reuse them in every video. Here's a starter set plus two templates.
For a support agent or in-app assistant, Eleven v4 Turbo in ElevenAgents uses the same voice in real time, so your product sounds like your videos.
Reality check: commercial use needs a paid ElevenLabs plan, and only clone a voice you have permission to use.