Illustrative · E-learning
A Canadian e-learning platform's multi-language narration
The brief landed with twelve modules, four languages, and a hard launch date tied to a provincial curriculum update. The client's earlier narration felt rushed — correct pronunciation but flat prosody that lost teenage listeners by minute three.
We cast licensed neural voices per locale, then ran directed text-to-speech sessions emphasizing warmth and pause structure. Speech-to-speech passes tightened emphasis on assessment instructions without sounding robotic. Each language received its own revision round with bilingual reviewers. Consent records covered all custom pronunciation guides the client supplied.
Deliverables included mastered WAV files, broadcast-safe loudness, and a usage-rights schedule covering LMS distribution for three years. Human-in-the-loop review caught synthesis artefacts in two modules before release. Total CAD project scope ran mid-five figures across languages — quoted upfront, not metered like a SaaS tool.