Voice and Languages
How the agent sounds: one voice with a live preview, then a language list where every row carries its own voice and dialect strength. Rules decide who hears which, and a pronunciation editor keeps the hard words right.
At a glance
- A voice picker with preview, overrides per language
- Dialect strength per language, the default up front
- Rules by caller group, country code or channel
- A pronunciation editor with a hear-it button

The languages card
One row per language: a Default radio — checking one moves it to the front — a voice select with inherit and preview, and a dialect-strength select. A locale picker adds languages, and a mirror-caller switch lets the agent follow the caller's lead. There are no pace or pitch dials — the voice engine takes a voice and a language, nothing more.
Who hears which, and what may change
Rules map a caller group, a country code or a channel to the profile the agent speaks with them, so one agent meets each audience its own way. The caller-may-change card sets what may shift mid-call: allow a language switch, allow an accent switch, preserve technical terms.
Pronunciation and the announcement
The pronunciation editor holds the words the agent must say exactly right, each with a hear-it button to check the result. Last on the pane sits the workspace recording announcement — the recording notice callers hear.
- Pick the voiceChoose the agent's voice and hear it with the preview; the spoken name sits beside it.
- Add the languagesThe locale picker adds a row; each row takes its own voice — or inherits the main one — and a dialect strength. Check Default and the row moves to the front.
- Decide who hears whatWho-gets-which rules say which profile the agent speaks for a caller group, a country code or a channel.
