Vois
Back to BlogVoice Technology

Omni Voice Design: Build Custom AI Voices Without Recording a Sample

Vois TeamVois Team
May 15, 2026
5 min read

TLDR:Omni Voice Design lets you create reusable voices from attributes such as gender, age, pitch, accent, and style. You can preview locally, save the result, then use it beside built-in and cloned voices across projects.

When a library voice is close but wrong, production slows down. You can settle for "good enough" or ask a real person for a recording to clone. Neither helps when the project needs a new voice for a role that does not exist yet.

Vois solves that casting problem with Omni Voice Design. Choose attributes such as gender, age, pitch, accent, and style, audition the result against the real script, then save the approved voice locally beside the rest of the project.

That keeps voice creation inside the same desktop workflow as script assignment, generation, timeline review, mastering, and export. You still make the taste call. Vois gives you a repeatable way to test and retain it.

Magic wand illustration for voice design

Design from the role, not from a file

Good casting starts with intent. You rarely think, "I need file number 47." You think:

  • a calm audiobook narrator with a low, steady delivery
  • a bright podcast co-host who sounds alert but not shouty
  • a careful product explainer with clean pronunciation
  • a weathered game commander with authority and restraint
  • a friendly app assistant that feels helpful without sounding synthetic

That is the workflow Voice Design is built around.

Instead of writing a fragile prompt from scratch every time, the Vois interface gives you structured controls for the core attributes: gender, age, pitch, accent, and style. You can start from a source voice or design from scratch. You can use a preset as a starting point, then adjust the attributes until the voice sits in the right place.

The important part is that the output becomes a real saved voice. Once it is in your library, you can assign it to speakers, use it in projects, and search it like any other voice.

Why Omni matters to the workflow

Omni gives Voice Design the range to create and reuse a role without forcing you to begin with an audio recording. It can carry a designed voice across broad language coverage and supports selected non-verbal tags where they fit the performance.

For creators, that means a designed voice can be:

  1. saved and reused across projects;
  2. kept consistent as a project expands to supported languages; and
  3. assigned directly to the speaker who needs it.

Tag support is specific to the chosen voice workflow. Omni supports tags such as [laughter], [sigh], [confirmation-en], and [surprise-oh]. Use them only when they help the line, then preview the complete sentence before approving it.

Globe illustration for multilingual voice design

Designed voices, cloned voices, and built-in voices

The three voice types serve different jobs.

Built-in voices are the fastest way to cast a project. Vois now includes 100+ voices across 21 categories, covering audiobooks, podcasts, explainers, games, app assistants, interactive dialogue, broadcasters, educators, characters, and more.

Cloned voices are best when you need to recreate an existing voice with permission. You provide a short sample, and Vois saves a local voice that can be assigned later.

Designed voices are for new voices. You are not copying a person. You are defining a role and creating a reusable voice identity for that role.

That distinction matters for workflow. A podcast might use a built-in host, a cloned creator voice, and a designed recurring guest. A game might use built-in voices for background NPCs, cloned voices for key performances, and designed voices for factions or character classes. An audiobook might use a built-in narrator and several designed character voices to keep the cast consistent.

The production flow

Use Voice Design as a casting review, not as a one-click personality generator:

  1. Open Voice Design and choose a source voice or start from scratch.
  2. Set the profile attributes that matter for the role.
  3. Enter preview text that reflects the line the voice will actually perform.
  4. Generate several previews and compare them in the project context.
  5. Save the voice only when the performance fits.
  6. Assign it to the appropriate speakers in the script editor.

The preview text matters. An audiobook narrator should be tested with audiobook text. A game commander should be tested with game dialogue. A product explainer should be tested with a clear instructional line. The closer the preview is to the real job, the more useful the casting decision becomes.

Production tools illustration

What this changes for creators

Voice Design is a way to build a repeatable voice system for a project rather than a one-off demo clip.

For audiobooks, you can make a stable cast without relying on one narrator to carry every role. For podcasts, you can create a recurring co-host or guest identity that fits the show's tone. For games, you can build voices for factions, enemies, guides, and companions. For explainers, you can establish a clear, calm brand voice.

The designed voice lives in the same Vois workflow as the script and production decisions. It is saved locally, appears in the voice selector, and can be reused rather than recreated from memory.

Compare Voice Design with Voice Cloning, or browse the voice library when an existing voice may fit. When a role needs its own sound, Get started by designing and reviewing it against one real scene.

The Vois Team

Frequently Asked Questions

What is Omni Voice Design?

Omni Voice Design is the Vois workflow for creating reusable custom voices from selected attributes instead of a recorded voice sample.

Do I need to upload audio to design a voice?

No. Voice Design works from attributes and preview text, and the generated voice is saved locally in your Vois library.

How is this different from voice cloning?

Voice cloning recreates an existing voice from audio. Voice Design creates a new voice profile from attributes such as gender, age, pitch, accent, and style.

Can designed voices work in multiple languages?

Yes. Designed voices are built for Omni and can be used across the same broad language coverage as the Omni engine.

TtsAi VoicesMultilingualProductionWorkflow
Share:
Vois Team

Written by

Vois Team

Product Team

The team behind Vois, building the future of AI voice production.