sunholo/explainer

The narration half of an explainer-film pipeline (the picture stays in Manim): voice a whole script in ONE std/ai.callSpeech call so the voice does not drift between lines, align speech-to-text word timings to the script (difflib matching blocks, re-implemented), cut the take at the longest pause between lines, then trim and RMS-level each line into mono PCM16 WAV. Word timings come from the caller (e.g. faster-whisper); per-sample work is meant for the bytecode VM.

Field	Value
Latest	0.1.1
Stability	experimental
Effects	AI, IO
Exports	6 modules
Last Updated	2026-10-26T13:39:06Z

Source Code ↗Documentation ↗License ↗

Agent Guide (AGENT.md)

Loading agent guide…

Exported Modules

• sunholo/explainer/take
• sunholo/explainer/words
• sunholo/explainer/align
• sunholo/explainer/cut
• sunholo/explainer/level
• sunholo/explainer/narrate

Dependencies

No dependencies (pure standalone package).

Loading version history from registry...