> 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/talking-to-audenai.md).

# Talking to audenAI

**In short:** tap the microphone in audenAI chat and talk instead of typing. What you say is sent on its own after a short pause, audenAI tells you what it understood and waits for you to say go before it does anything, and it can read its replies aloud. It uses your browser's own speech features, and VoiceMoat never receives, records, or stores audio.

#### Start talking

1. Open **audenAI chat**.
2. Tap the **microphone** next to Send. Your browser may ask for microphone permission the first time.
3. Talk. Your words appear in the box as you speak.
4. Stop talking. About **two seconds** later, the message sends on its own.

audenAI replies, reads the reply aloud, and the microphone opens again when it finishes, so you can keep the conversation going hands-free.

#### Ending it

* Tap **End** to stop talking mode. If you tap it mid-sentence, your words stay in the box to edit instead of sending.
* Talking mode also ends if nothing is said for about 30 seconds, or when you close the chat panel.

#### audenAI checks before it acts

Speech recognition sometimes mishears a word: "cloud" for "Claude", or a name written as ordinary words. So when you ask for something out loud, audenAI works differently from typed messages:

* **Questions and lookups** are answered straight away.
* **Anything that does something** (writing or rewriting a post, rating, visuals, scheduling, saving, changing a rule or a memory, posting) waits. audenAI tells you in a sentence or two what it heard and what it will do, then asks you to say go.
* If a word could mean something else, it asks: "Did you mean Claude, the AI, not cloud?"
* Say **go** (or yes, do it, go ahead) and it carries out exactly that plan. Say what to change and it restates the plan and waits again.

{% hint style="warning" %}
**Saying go never presses a confirm card for you.** If an action opens a card, such as posting now or scheduling, audenAI tells you it is on screen and you tap the button yourself. Nothing is posted by voice alone.
{% endhint %}

Common names are fixed as you talk, so "linked in" becomes LinkedIn and "voice moat" becomes VoiceMoat. Ordinary words are never changed in the box; audenAI works those out from context instead.

#### Replies read aloud

When you talk, audenAI's replies are read aloud in your browser's own voice, drafts included. It keeps its replies short and conversational while you are talking.

* Tap **Skip** to stop the current reply.
* Tap the **speaker** icon to switch reading aloud on or off.

#### Which browsers work

Talking works in browsers that include speech recognition, such as **Chrome** and **Edge**. In browsers without it, such as Firefox, the microphone button does not appear. Some browsers, such as Brave, show it but refuse the speech service; VoiceMoat tells you when that happens.

If the microphone is blocked, allow it from the icon in your browser's address bar and try again.

#### What it costs

Talking is part of audenAI chat, on every paid plan. Each message you send costs the same as a typed one: 1 credit for a plain answer, 2 when audenAI uses a tool.

{% hint style="info" %}
Because audenAI checks with you before acting on a spoken request, a spoken action usually takes one extra message (your "go"), so it costs 1 credit more than typing the same request.
{% endhint %}

#### Your privacy

{% hint style="success" %}
VoiceMoat never receives, records, or stores audio. Your browser turns what you say into text, and VoiceMoat receives only that text, handled like anything you type. Reading aloud is done by your browser too. In Chrome, for example, the browser's speech service is provided by Google, under Google's privacy policy. See Security and your data.
{% endhint %}

#### Common questions

**Does it send before I have finished?** It waits for about two seconds of silence after you stop talking. If you pause longer mid-thought, just keep talking and add to the next message, or tap End and edit the words in the box.

**Can I type and talk in the same conversation?** Yes. Typed messages work exactly as they always do, without the go step.

**Can I post by voice?** You can ask for a post by voice. The publish card still needs your tap on **Post**.

**Does VoiceMoat become a voice or audio product?** No. VoiceMoat is the personal brand operating system for Twitter/X and LinkedIn, and it writes text. Talking is just another way to tell audenAI what you want.