Short answer: Voice is implemented in the current pre-beta build. You can assign voices to characters and narration, render spoken lines, and play completed audio alongside the matching dialogue.
Dialogue and narration
Choose a character's voice in their Sound settings or the project's Words manifest. The narrator voice and the voicing mode are in project Settings. The Words manifest shows voice coverage, the estimated render cost, and controls for starting a render job and reviewing its results. Rendering runs in the background and reports progress.
The player uses completed audio that matches the current line and selected voice. Changing the words or voice can require a new render. Voice does not need a separate publish or bake step, and incomplete voice coverage does not block publication. Readers can choose text, voice or both; a line without audio still shows its text.
Current limits
Inline emotion or performance tags do not direct the generated delivery. Custom voice recordings are audio assets; an uploaded clip is not automatically attached to a dialogue line. Studio has no tool for generating a one-off voice clip at the moment; you can upload your own audio in Assets. Availability remains restricted during pre-beta.