# What are Voice Tags? SSML vs Prompt Tags

> Voice tags tell a text to speech voice how to say a line. Learn the difference between SSML markup and the natural-language prompt tags modern AI voices use, with examples.

Canonical: https://zenmic.com/glossary/what-are-voice-tags/
Last updated: 2026-10-12

🏷️ Voice & TTS Glossary

Voice tags are small instructions inside a script that tell an AI voice how to say a line: pause here, whisper this, laugh there.

📖 6 min readUpdated October 2026

## The Simple Definition

Voice tags are markup or short instructions placed in a text to speech script to control delivery: pauses, emphasis, speed, emotion and non-verbal sounds. Older TTS systems use SSML (Speech Synthesis Markup Language), an XML format. Modern AI voices also understand plain-language tags like [excited] or <laugh>, which are easier to write and can describe emotion SSML never could.

## SSML: The Older Standard

SSML wraps text in XML elements such as <break time="500ms"/>, <emphasis> and <prosody rate="slow">. It is precise about timing and speed, but verbose, and it has no real way to say "sound sarcastic" or "barely holding back tears".

## Prompt Tags: How Modern AI Voices Are Directed

→

**Line styles** — [excited], [whispering], [deadpan], [softly]: how the next line is performed.

→

**Sounds** — <laugh>, <sigh>, <gasp>: performed exactly where they appear.

→

**Pauses** — <short pause>, <long pause>: timing without milliseconds.

## SSML vs Prompt Tags at a Glance

→

**Ease of writing** — SSML: XML markup. Prompt tags: plain words in brackets.

→

**Emotion** — SSML: limited. Prompt tags: any emotion you can describe.

→

**Precision** — SSML: exact milliseconds and rates. Prompt tags: natural, approximate.

→

**Languages** — Both work across languages; prompt tags stay in English even in a German or Japanese script.

## The same line, two ways

```
SSML:
<speak>I can't believe it. <break time="700ms"/> <prosody rate="slow">We actually won.</prosody></speak>

Prompt tags (ZenMic):
Maya: [stunned] I can't believe it. <long pause> [slowly, joyful] We actually won.
```

Prompt tags describe the performance; the voice works out the timing.

[See All Voice Tags →](https://zenmic.com/features/expressive-ai-voices/) [More Glossary Terms](https://zenmic.com/glossary/)

## Frequently Asked Questions

### What is SSML?

SSML (Speech Synthesis Markup Language) is an XML standard for controlling text to speech: pauses, emphasis, pitch and speed.

### Does ZenMic support SSML?

ZenMic uses natural-language prompt tags instead: [excited], <laugh>, <short pause> and similar. They are easier to write and control emotion better.

### Should tags be in English for other languages?

Yes. Keep tags like [whispering] in English even when the script is in German, Spanish or Japanese.

### Will the voice read the tags aloud?

No. Tags are performed, not spoken. Only the words outside the tags are read.

### Related Resources

[Expressive AI voices: all tags](https://zenmic.com/features/expressive-ai-voices/) [What is text to speech?](https://zenmic.com/glossary/what-is-text-to-speech/) [What is a podcast script?](https://zenmic.com/glossary/what-is-a-podcast-script/)
