Zarema Khalilova · zmoki.xyz

Voice fingerprint analysis

Your voice is unique. AI rewrites it in its own handwriting. This analysis preserves yours.

I measure what makes your writing yours, so your voice survives in every post you publish with AI.

Analysis plan
1

Baseline

your idiosyncrasies, measured from texts in your original voice

2

Score

every Substack and LinkedIn post against your baseline, showing how far AI pulled each toward its own voice

3

Improvements

how each published post should sound, and what to change to get there

4

Guidelines

how to keep publishing with AI and keep sounding like you

Methodology

Voice lives in structure, deeper than word choice.

So I measure 20 discourse-level features built on two published studies.
StoryScope · COLM 2026. Found AI's idiosyncrasies survive style editing because they live at the discourse level.
NarraBench · EACL 2026. The taxonomy of narrative understanding the features are drawn from.
10 core features · where AI flattens any writer
how a post opens concrete specifics how it ends emotion handling hedging transitions sentence rhythm
10 fingerprints · your idiosyncrasies, patterns only you produce
recurring themes signature phrasings real people & numbers how you address readers your quirks
What you get

Visual report

Your baseline overview. How far AI pulled toward its own style. All posts ranked, all gaps mapped.

Improvement plan

For the posts already published. Flagged passages quoted, and exactly how to rewrite each one to bring your voice back.

Brief for future posts

Your editor's guide for writing with AI, so every new post keeps sounding like you.