Ultra

Telnyx Ultra text-to-speech delivers sub-100ms latency across 44 languages, available exclusively through the TTS REST API for ultra-fast voice synthesis.

Voice format: Telnyx.Ultra.<voice>

Sub-100ms latency. 36 languages.

REST only — Ultra is not available over public WebSocket.

Ultra voices tagged as deprecated remain usable, but will not receive upgrades.

Voice Samples

Voice	Language	Gender	Sample
Telnyx.Ultra.Asher	en	Male
Telnyx.Ultra.Callie	en	Female
Telnyx.Ultra.Clara	en-US	Female

---

Expressive Mode for AI Assistants

Enable Expressive Mode when using an Ultra voice with AI Assistants. The assistant can select emotion tags and nonverbal cues from the conversation context without hard-coding each expression in the system prompt.

For example, an assistant can use an excited tone for good news, respond empathetically to a problem, insert [laughter] in a friendly exchange, and keep routine information neutral.

Enable in the portal

1. Open the assistant in the Telnyx Portal.
2. Under Voice Settings, select a Telnyx.Ultra.* voice.
3. Enable Expressive Mode.
4. Save the assistant.

Enable via API

Set expressive_mode: true in the assistant’s voice_settings:

curl -X PATCH "https://api.telnyx.com/v2/ai/assistants/YOUR_ASSISTANT_ID" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "voice_settings": {
      "voice": "Telnyx.Ultra.YOUR_VOICE_ID",
      "expressive_mode": true
    }
  }'

expressive_mode requires a supported expressive voice. Use a Telnyx.Ultra.* voice for Ultra Expressive Mode.

SSML Emotions

Ultra supports inline SSML emotion tags. Place the tag before the text:

<emotion value="excited" />Great news — your order shipped early!

Primary emotions: angry, excited, content, sad, scared.

Additional: happy, enthusiastic, curious, calm, grateful, affectionate, sarcastic, surprised, confident, hesitant, apologetic, determined, frustrated, disappointed.

Omitting the tag = neutral delivery. Use sparingly — Ultra interprets emotional subtext from the text itself.

Pauses and Pacing

Insert <break time="Xs" /> inline to add a pause. Use short pauses for pacing and emphasis; longer pauses can make a caller think the call dropped.

Repeat an ID slowly by splitting it into groups:

Let me repeat that slowly: 4-F-2-A <break time="0.4s" /> 9-B-1-E <break time="0.4s" /> 7-C-3-D.

Pause while looking up information:

Let me look that up for you. <break time="1s" /> Okay, I found your account.

Emphasize important information:

Your total comes to <break time="0.3s" /> four hundred and twelve dollars.

Combine a pause with an emotion:

<emotion value="apologetic" />I'm sorry, but <break time="0.5s" /> that item is no longer available.

As a starting point, use 0.2–0.4 seconds for slight emphasis or dictation, 0.5–0.8 seconds for topic transitions, and 1–1.5 seconds for a lookup or significant shift. Avoid pauses longer than 2 seconds in a conversation.

Nonverbal Cues

Insert [laughter] inline for natural laughing:

That's hilarious! [laughter] Anyway, let me check your account.

---

Language Support

Set language_boost to improve pronunciation for the target language:

Arabic, Bengali, Bulgarian, Chinese, Czech, Danish, Dutch, English, Finnish, French, German, Gujarati, Hebrew, Hindi, Indonesian, Italian, Japanese, Korean, Malay, Marathi, Māori, Norwegian, Polish, Portuguese, Punjabi, Romanian, Russian, Slovak, Spanish, Swedish, Tamil, Telugu, Thai, Turkish, Ukrainian, Vietnamese.

---

REST API

Fields

Field	Type	Default	Description
voice_speed	float	1.0	Speech rate multiplier.
language_boost	string	—	Target language hint.
volume	float	—	Output volume.
emotion	string	—	neutral, happy, sad, angry, fearful, disgusted, surprised.
sampling_rate	integer	—	Output sample rate in Hz.
output_type	string	binary_output	binary_output, base64_output, or audio_id.

Response

Default (binary_output): chunked audio bytes with Content-Type: audio/mpeg.

With output_type: "base64_output": JSON with base64-encoded audio.

With output_type: "audio_id": JSON with an audio_url for deferred retrieval.

See also

xAI Grok is the second TTS provider supporting Expressive Mode. For Grok voice options, see Grok Voices. Note: Grok voices have higher latency than Ultra.