Opus 5.5 builds an explainer video from one prompt: self-judging voices, Whisper QA, word-level sync
spdustin · reddit · 2026-09-27
A Reddit user shares how Opus 5.5 generated a 60-120s science explainer video (atoms, elements, molecules, compounds, mixtures for 8th graders) from a single one-shot prompt, with one follow-up editorial fix. The model's autonomous workflow
- Tool discovery first: Claude queried the OpenRouter API for available TTS, music, and audio-intelligence models before planning anything.
- AI judge panel for voices: two Gemini models critiqued 13 candidate voices; after noticing judges favored whichever clip played first, it rotated A/B comparison order to remove the bias. "Kore" won, and it rewrote the style prompt after feedback that every voice sounded "sing-song, too elementary".
- Word-for-word QA: four takes per script with the winning voice; any take whose Whisper transcript didn't match the script exactly was discarded, then a listening panel picked the best.
- Audio-first timeline: the whole animation timeline was built around the audio, aligning animation cues to Whisper word timings — the bond animation snaps on the word "bond". It generated its own JSON cue sheet plus .srt/.vtt subtitle files.
The author shares the full original prompt; budget was capped at $10 of OpenRouter credits with all creative choices left to the model. Result was "pretty damn good for a one-shot," though one pedagogically shaky script choice needed a human fix.
More from coding & agent
- Building a Semantic and Knowledge Layer on BigQuery and Exposing It to Agents — batmanparam · 2026-09-27
- OpenAI selects 35 builders for first Codex Physical Builds hardware cohort — OpenAIDevs · 2026-09-27
- Multi-Agent Bot Pipelines Keep Stalling: Direct-to-Harness Still Beats Bot Teams — Old_Grapefruit8774 · 2026-09-27
- Claude Opus 5.5 Chains Blender, Image-Gen and three.js Into Browser-Running 3D — burny_tech · 2026-09-27
- Claude kills plan mode the same day Google announces launching it — dotey · 2026-09-27
- Gary Marcus amplifies warning: large teams using AI agents likely have unknown security incidents — GaryMarcus · 2026-09-27