Inline expressive tags

Updated November 3, 2026, based on the xAI Text to Speech page and Voice API reference: the fourteen official inline tags, in xAI’s four families (pauses, laughing and crying, mouth sounds, breathing). Tags are billed like the rest of the text. Sources: Text to Speech · Voice API.

Add emotion and realism to speech

Inline tags are markers you insert directly into your text to trigger a one-off vocal expression: a pause, a laugh, a mouth sound, a breath. xAI publishes fourteen of them, in four families. They make speech synthesis far more natural and expressive.

Inline tag syntax

You write inline tags in square brackets, [tag], and insert them directly into the text, right where you want the effect to happen:

Hello! [pause] How are you today?

During synthesis, the tag is replaced by the matching sound effect. The surrounding text is spoken normally.

Pauses

Pauses give speech structure and a natural rhythm:

• [pause]: a short pause
• [long-pause]: a longer pause, for a change of topic or to let an idea sink in
• [hum-tune]: humming a tune, which xAI files in the same family

First idea. [pause] Second idea. [long-pause] Now let's move on to the next topic.
In the meantime [hum-tune] let's see what we've learned.

Pauses are especially useful between the sections of a long text, or to give listeners time to take in an important piece of information.

Laughing and crying

These tags add emotional reactions in the middle of speech:

• [laugh]: a hearty laugh
• [chuckle]: a quiet chuckle
• [giggle]: a giggle
• [cry]: crying

That's really great news! [laugh] I'm so happy for you.

They work especially well in dialogue, podcasts, and narrative content, where they break the monotony of classic speech synthesis.

Mouth sounds

• [tsk]: a “tsk” of annoyance or disapproval
• [tongue-click]: a tongue click
• [lip-smack]: a lip smack

[tsk] Another meeting rescheduled.

Breathing

Breaths make speech sound more human:

• [breath]: a breath
• [inhale]: an inhale
• [exhale]: an exhale
• [sigh]: a sigh

[breath] Welcome to this new chapter.
And yes, another meeting [sigh] but this one will be short, I promise.

The [breath] tag is subtle but helps long texts: without breaths, an extended reading can sound mechanical.

Place tags strategically

How well inline tags work depends on where you put them. Here are the best practices.

At the start of a sentence

Put [breath] at the start of a new paragraph or section to mimic someone naturally starting to speak:

[breath] Now let's move on to the hands-on part of this lesson.

Between two sentences

Use [pause] or [long-pause] to create a natural break:

The first point is important. [long-pause] The second one matters even more.

After a key piece of information

A [pause] after an important figure gives listeners time to absorb it:

Revenue grew by 47%. [pause] It's our best quarter ever.

Full example

Here’s a text that combines several tags for a natural result:

{
  "text": "[breath] Hello, everyone, and welcome to this presentation. [pause] Today, we're going to talk about a fascinating topic. [long-pause] Artificial intelligence that speaks. [chuckle] Yes, it's a little meta, I know. [pause] But you'll see, the results are impressive. [breath] Let's start with the basics.",
  "voice_id": "eve",
  "language": "en"
}

Key takeaways

• You write inline tags in square brackets, [tag], inside the text
• [pause] and [long-pause] shape the rhythm of speech
• [pause], [long-pause], and [hum-tune] make up the pause family
• [laugh], [chuckle], [giggle], and [cry] add emotion; [tsk], [tongue-click], and [lip-smack] add mouth sounds
• [breath], [inhale], [exhale], and [sigh] make speech sound more human
• Put tags where a person would naturally pause or react
• Tags count toward the 60,000-character limit per request and are billed like the rest of the text