Audit any text against a four-axis "voice print" — and optionally against your own brand reference.
The four dimensions. Each is scored from 0–100, where 50 is neutral:
- Formal ↔ Casual — contractions, slang, emoji, first-person vs. third-person, jargon density.
- Confident ↔ Cautious — hedges (maybe, might, perhaps), intensifiers (definitely, certainly), passive voice, conditional framing.
- Optimistic ↔ Concerned — positive vs. negative sentiment vocabulary, future framing, risk language.
- Playful ↔ Serious — exclamation marks, humor markers, conversational openers, puns vs. measured academic register.
How it works. A curated lexicon of ~1500 hand-picked markers (hedges, intensifiers, contractions, fillers, sentiment-bearing words, formality cues) is matched against your text, normalized by word count, and projected onto each axis. Sentence length, punctuation density, and emoji are factored in. It is a lexicon + heuristic engine, fast and offline — no LLM calls, no data leaves your browser.
Brand reference. Paste a sample of your brand's existing copy (≥50 words) and the auditor computes the same four-axis voice print for it, then reports your text's deviation 0–100 per axis plus an overall drift score. A drift <15 is on-brand; >30 suggests a voice mismatch worth revisiting.
Use cases.
- Catching a junior writer's draft that drifts from your house style.
- Checking that a chatbot reply matches the brand's confident-but-warm tone.
- Comparing two tone options for a launch email.
Limitations. Lexicon-based scoring is approximate and English-focused (it understands the other supported languages with reduced coverage). It does not detect irony, sarcasm, or register shifts mid-text — for those, a human editor still wins.