> For the complete documentation index, see [llms.txt](https://docs.voicemoat.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.voicemoat.com/voice-and-audenai/the-10-signals.md).

# The 10 signals

**In short:** audenAI reads your content across ten signals of voice, not just "topic" or "tone". Together they are what make a draft read like you instead of like a generic assistant.

#### Why signals, not a style setting

Most tools treat voice as a slider (more casual, more formal) or a brand-voice document you fill in. That produces content that is roughly in the right direction but still reads as generic. audenAI instead measures your actual content across many dimensions at once, so a draft matches the specific fingerprint of your posts.

#### The ten signals

1. **Sentence rhythm**: how long or short you run, where you break, when you fragment.
2. **Vocabulary**: the words and register you reach for, and the ones you never use.
3. **Hooks**: how you open, whether a question, a bold claim, a stat, a story, or a direct address.
4. **Structure**: how you build a post or a thread from start to finish.
5. **Tone**: your mix of formality, confidence, humour, warmth, and edge.
6. **Punctuation**: your habits with commas, line breaks, ellipses, and casing.
7. **Recurring references**: the ideas, people, and themes you come back to.
8. **Taboos**: the phrasings and clichés you avoid.
9. **Mode-specific register**: how your voice shifts between a tweet, a reply, and a thread.
10. **Persona markers**: the small tells that make your content recognisably yours.

Training also measures **how your posts feel**: the feelings they carry, your humour, and your usual strength. audenAI uses that range to pick a feeling for each post it writes. See Feelings in your drafts.

#### What audenAI counts

audenAI does not just read your posts: it counts how you actually write in them. The counts sharpen the signals above, especially rhythm, vocabulary, hooks, structure, and punctuation:

* how long your sentences run,
* how many lines a post runs, and whether you put one sentence on each line,
* the words you open with,
* the filler words you do and don't use,
* your punctuation,
* how often you say I, you, and we,
* your reading level,
* on LinkedIn, how long your first line is before "see more",
* and your hashtag habit.

audenAI's writing rules follow these counts too. If you put one sentence on each line, drafts keep that shape; if you rarely end on a question, drafts don't either. See Feelings in your drafts.

#### How you think

When you train or re-train, audenAI also reads how you think on the page: how you make a point (from experience, with a story, with numbers, step by step, and so on), how you build an idea, the metaphors and references you come back to, your humour, and how you present yourself. Only examples and names that really appear in your posts are kept. You see it on the **How you think** card in Voice Lab, and audenAI writes the same way.

#### How they show up in the product

* When audenAI drafts, it writes to match these signals. Studio's **Style** card then counts how many of your measurable habits the draft matches ("4 of 6 checks match"): sentence length, emoji use, casing, post length, opener style, filler words, line shape, plus two checks for AI tells and drift away from how you normally write. Each check that is off shows a real line of yours, or your usual pattern, next to it.
* What a draft **says** is a separate check. **Stance** reads each line against your claim bank, now called Your stories and views (the views, stories, and framings in your own posts). Style and Stance are never blended into one number. See Style and Stance.
* The **Your voice** tab in Voice Lab shows several of these signals: How you write, six habit tiles (How you open, Sentence length, Emoji, Format, Numbers and full stops, Words you use), and How you think.
* The **Teach audenAI** tab lets you nudge some of them directly: **How you like it** for emoji, capitals, punctuation, reply length, and lists, and **Your rules** for words to never use.

{% hint style="info" %}
Signal 9, mode-specific register, is why a reply audenAI writes sounds different from a thread it writes. It learns that you are not the same on a quick reply as you are in a long post, and keeps each in its own register.
{% endhint %}

#### An honest note

The ten signals are a way of describing what audenAI measures, not ten dials you set by hand. You do not tune each one; you train on your posts and steer with How you like it and Your rules. The point is that voice is many things at once, and audenAI treats it that way.

Next: audenAI Standard vs Deep.