Advanced use cases for tags

Updated November 3, 2026, based on the xAI Text to Speech page: every example uses the official tag list. Source: Text to Speech.

Get the most out of expressive tags

Beyond the basic combinations, expressive tags open up creative possibilities for specific use cases. This lesson explores real scenarios where tags turn standard speech synthesis into an engaging audio experience.

Audiobooks and fiction

Fiction narration needs shifts in emotion and pace to hold the listener. Expressive tags let you set dialogue, description, and moments of tension apart.

Dialogue

Use different styles to distinguish characters and their emotions:

The detective walked into the room. [long-pause]
<lower-pitch><slow>Where were you last night between 8 PM and midnight?</slow></lower-pitch> [pause]
<higher-pitch><soft>I... I was at home. Alone.</soft></higher-pitch> [pause]
<lower-pitch><emphasis>Alone. Really?</emphasis></lower-pitch> [sigh]

By giving each character a distinct style (low for the detective, high and soft for the suspect), you create an audible difference without changing voices.

Atmospheric descriptions

Descriptive passages become more immersive with the right pace:

<soft><slow>Night was falling over the city. [breath] The lights came on one by one, like stars that had drifted down to earth.</slow></soft> [long-pause]
<loud>Suddenly, a scream tore through the silence.</loud>

E-learning and training

For educational content, tags help direct the learner’s attention.

Highlight definitions

Now let's look at a fundamental concept. [pause]
<slow><emphasis>An API, or application programming interface, is a set of rules that lets two pieces of software talk to each other.</emphasis></slow>
[long-pause] Make sure you remember this definition.

Signal hands-on steps

[breath] Let's put this into practice. [pause]
<slow>Step one: open your code editor.</slow> [long-pause]
<slow>Step two: create a new file named app.py.</slow> [long-pause]
<slow>Step three: paste in the following code.</slow>

Slowing down makes learners follow the pace you set instead of letting the audio run past them.

Voice assistants and chatbots

Voice assistants need to sound natural and empathetic, and tags go a long way toward that.

Empathetic responses

<soft>I understand how frustrating this is.</soft> [pause]
Let me help you sort out this problem. [pause]
<emphasis>Your satisfaction is our priority.</emphasis>

Handling errors

[breath] <soft>Hmm</soft>, there seems to be a problem with your order. [pause]
<slow>Could you confirm your reference number for me?</slow> [pause]
I'll look into it right away.

Audio marketing content

Ad spots

<loud><fast>Special offer!</fast></loud> [pause]
Until March 31, <emphasis>50% off</emphasis> all our courses. [pause]
<slow><emphasis>Don't miss out.</emphasis></slow> [long-pause]
<fast>Visit corsen.ai. Terms apply.</fast>

Jingles and audio signatures

<singing>Corsen Academy, your future starts here.</singing>

The <singing> tag turns the text into a light melody, perfect for short brand signatures.

Accessibility and inclusion

Tags make audio content more accessible to people who have difficulty following speech.

Adapted reading

<slow>Welcome to our platform. [long-pause]
You can navigate with the arrow keys on your keyboard. [long-pause]
For help, press F1.</slow>

Dictation

<slow>The word is [pause] <emphasis>antidisestablishmentarianism</emphasis>. [long-pause]
Once more. [pause] <emphasis><slow>an-ti-dis-es-tab-lish-men-ta-ri-an-ism</slow></emphasis>.</slow>

Optimization tips

Test with different voices

The same tagged text can sound very different depending on the voice. Eve makes [laugh] sound more spontaneous, while Leo makes it more restrained. Always test.

Match tag density to the pace

For text read quickly, use fewer tags. For text read slowly (meditation, relaxation), add more pauses and breaths.

Keep the text readable

Even with tags, your source text should stay readable by a human. If the tags make the text hard to read, you probably have too many.

Key takeaways

• Tags let you tell characters apart in fiction
• Slowing down (<slow>) is essential for educational content
• Voice assistants sound more empathetic with <soft> and [pause]
• Audio marketing benefits from <loud>/<fast> contrasts to create urgency
• Test every tag combination with the voice you’re targeting
• The readability of the source text comes first