Voiceover Studio
Gemini TTS production tool that turns health-program scripts into styled audio, as both a CLI and a multi-user web app.
Health programs ship with a lot of narrated audio, and recording it is the bottleneck. Voiceover Studio converts a script — a .docx or .txt file — into finished audio in the voice and pacing the program calls for.
Because Gemini TTS is LLM-based, style is controlled through natural language rather than SSML: you describe the delivery you want and the model produces it. That turned out to matter more than voice selection, and it is why the tool exposes prompt controls per segment rather than a fixed set of voice presets.
It ships as two entry points from one codebase — a CLI for batch conversion and a multi-user web app with per-session encrypted API keys behind AWS Cognito, so a content team can use it without anyone sharing credentials.
Highlights
- Natural-language style control per segment instead of SSML markup.
- Batch conversion of multi-section scripts to MP3 and WAV with per-segment prompt records.
- Multi-user web app with Cognito auth and per-session encrypted API keys.
- Handles multilingual scripts — the production output includes full French program sets.