voice-calibrator

To calibrate a voice profile, you need to collect 50+ writing samples and run the deterministic calibration workflow, which extracts quantitative metrics like sentence length and punctuation to generate a machine-readable profile.

Deterministic voice analysis works by running scripts on writing samples to extract quantitative metrics, interpreting the results with AI, and validating the outputs against measurable targets to ensure accurate authorship matching.

Yes, you can validate generated content against a voice profile by comparing the content's metrics to the established tolerances, allowing you to verify authorship matching and perform A/B style comparisons between outputs.

Voice calibration requires a minimum of 50 writing samples from diverse sources to accurately extract quantitative voice metrics and generate a robust profile with measurable targets.

You should not use voice calibration for general content generation, general writing, or non-voice editing, as the workflow is specifically designed for creating, refining, and validating voice profiles.

The voice calibration workflow outputs a machine-readable profile.json with measurable targets and tolerances, alongside a SKILL.md voice skill utilizing a sample-first architecture for authorship matching.