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 Studio solves that casting problem with Omni Voice Design. Describe the speaker in your own words, audition takes 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 Studio gives you a repeatable way to test and retain it.
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.
You describe the voice in plain words, for example "a middle-aged man with a deep, resonant voice. Speaks slowly with a calm, steady tone". Gender, age, pitch, tone, emotion, pace and vocal texture all shape the result. You can start from one of the examples and edit it, or start from an existing voice and describe how it should sound instead. Each take sounds a little different, so generate a few and keep the one that 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 carries a designed voice across its 30 languages, and you steer the delivery by describing it in plain words.
For creators, that means a designed voice can be:
- saved and reused across projects;
- kept consistent as a project expands to supported languages; and
- assigned directly to the speaker who needs it.
Omni does not use bracketed reaction tags. To shape a line, describe the delivery you want, such as "warm and amused, with a light laugh in the voice", then preview the complete sentence before approving it.
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 Studio 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 Studio 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:
- Open Voice Design and choose a source voice or start from scratch.
- Set the profile attributes that matter for the role.
- Enter preview text that reflects the line the voice will actually perform.
- Generate several previews and compare them in the project context.
- Save the voice only when the performance fits.
- 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.
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 Studio 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