Voice variants

A voice variant is a single TTS voice from one of JARAI’s speech providers. The catalogue spans many providers, languages, and accents, and you pick a variant per avatar. This page covers how to find the right voice.

🎬 Find the right voice

Prefer to read? Open the step-by-step transcript

1. Open the voice picker on an avatar’s form.
2. Use the trait chips to filter — accent, age, gender, style, use-case.
3. Preview a variant with the inline player.
4. Pick it for your avatar; if the accent you want has no voice yet, ask JARAI support for it.

Where you browse voices

For account users, the voice picker on the avatar form is the catalogue. The full Settings → Voice Variants page — the whole catalogue across providers, plus its request button for missing accents — is a platform-admin page; see Languages & voices.

Filtering by traits

Every voice carries provider metadata parsed into trait facets you can filter by:

• Accent — the regional accent (see below).
• Age, Gender, Style — broad voice character.
• Use cases — what the provider suggests the voice suits (narration, ads, audiobook…).
• Tags / colour — extra descriptors where the provider supplies them.

Select chips across categories to narrow down: filtering is AND across categories (accent and gender) and OR within a category (British or Irish). Platform admins get the same filter on the Voice Variants page, so a selection drills to the same voices in both places.

Accents

Accent is a first-class trait. Voices are tagged with a precise regional label — from country level (e.g. Spain — Castilian, UK — Received Pronunciation) down to sub-regional accents (e.g. Andalusia — Sevillano, Brazil — Carioca). Use the accent filter to match a voice to your audience.

Tip

The accent catalogue is intentionally deep. If you produce for a specific region, filtering by its accent makes the output feel local rather than generic.

Previewing

Use the inline preview player before committing a voice to an avatar — a voice that reads well for ads may not suit long-form narration, and accent nuance is easier to hear than to read.

When there’s no voice for an accent yet

The accent taxonomy is broader than any single provider’s catalogue, so some accents have no voice yet. When the avatar’s voice picker has nothing for the accent or language you chose, it says No voices available — there’s no request button there. If the accent you need has no voice:

• Pick the closest available accent (often the parent country/region), or
• Contact support to request the specific accent — demand signal helps prioritise which voices get added.

Note

New voices are added over time as providers expand and as the voice harvester discovers and vets them, so the catalogue you see grows.