Grok Voices & Speech Tags

Available Grok voices and expressive speech tags for natural, emotional AI speech

Grok voices are powered by xAI and support expressive speech synthesis through a tag system that lets you embed emotion, pacing, and vocal effects directly in the input text.

Available Voices

Use GET /api/v1/voices to retrieve the current list of voices and filter the results by model: "xai":

curl -X GET 'https://sexyvoice.ai/api/v1/voices' \
  -H 'Authorization: Bearer sk_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx'

The response includes each voice's name, language, and model. Pass the name as the voice parameter in speech requests.

Speech Tags

Grok supports two kinds of speech tags that you embed directly in the input text:

Instant tags — single sound effects inserted at a specific point, written as [tag-name]
Wrapping tags — style effects applied to a span of text, written as <tag-name>...</tag-name>

Instant Tags

Tag	Description
`[pause]`	Short pause
`[long-pause]`	Longer pause
`[laugh]`	Laughter
`[chuckle]`	Brief chuckle
`[giggle]`	Giggle
`[cry]`	Crying sound
`[sigh]`	Sigh
`[breath]`	Audible breath
`[inhale]`	Inhale sound
`[exhale]`	Exhale sound
`[tsk]`	Tsk sound
`[tongue-click]`	Tongue click
`[lip-smack]`	Lip smack
`[hum-tune]`	Humming

Wrapping Tags

Tag	Description
`<soft>...</soft>`	Soft, gentle delivery
`<whisper>...</whisper>`	Whispered speech
`<loud>...</loud>`	Louder, projected voice
`<emphasis>...</emphasis>`	Emphatic stress
`<slow>...</slow>`	Slower speech rate
`<fast>...</fast>`	Faster speech rate
`<higher-pitch>...</higher-pitch>`	Raised pitch
`<lower-pitch>...</lower-pitch>`	Lowered pitch
`<build-intensity>...</build-intensity>`	Gradually increasing intensity
`<decrease-intensity>...</decrease-intensity>`	Gradually decreasing intensity
`<laugh-speak>...</laugh-speak>`	Laughing while speaking
`<sing-song>...</sing-song>`	Sing-song delivery
`<singing>...</singing>`	Sung vocal style

Examples

Instant tags

{
  "model": "xai",
  "voice": "eve",
  "input": "Welcome everyone! [laugh] I'm so happy to be here. [pause] Let's get started."
}

Wrapping tags

{
  "model": "xai",
  "voice": "eve",
  "input": "<emphasis>Do not miss this deadline.</emphasis> <soft>I'm counting on you.</soft>"
}

Combining instant and wrapping tags

{
  "model": "xai",
  "voice": "eve",
  "input": "<build-intensity>The crowd grew louder and louder</build-intensity> [laugh] until the stadium erupted."
}

Whispered secret

{
  "model": "xai",
  "voice": "eve",
  "input": "I have something to tell you. <whisper>Don't tell anyone.</whisper> [pause] Promise?"
}

Output Format

Grok voices support both mp3 (default) and wav:

{
  "model": "xai",
  "voice": "eve",
  "input": "Hello there!",
  "response_format": "mp3"
}

Notes

Tags count toward the character limit and are billed as input characters
Unsupported or misspelled tags pass through as plain text in the audio
Grok voices do not support the style parameter — use speech tags for expression instead
The seed parameter is not supported for Grok voices

Grok Voices & Speech Tags

On this page