voice-casting

Voice casting selects the right TTS voice variant for each script section by matching text register—like openers or direct address—to the correct sibling voice in a family, ensuring narration sounds natural instead of flat.

Voice casting in a family-based TTS workflow assigns the correct sibling voice variant to individual narration clips based on their text, allowing mixed-register scripts to use different delivery styles like warm, dynamic, or phonetic.

You split a narration script into mixed-register clips—such as hooks, explanatory exposition, and pronunciation-heavy lines—then voice casting assigns the appropriate family-based voice slug to each segmented clip.

Yes, voice casting handles pronunciation-heavy lines by selecting a phonetic delivery variant, ensuring that technical model names and complex segments in your narration script land correctly.

You need brand guidance, section-level segmentation, family-specific voice slugs, and metadata output that preserves stable section IDs and clip-level casting decisions for the workflow to function correctly.

Use per-clip voice variant selection when a script shifts between narrative beats, such as moving from a hook to explanatory exposition, requiring different delivery variants so each beat lands correctly instead of sounding mismatched.