Voices

The Voices API returns available Text-to-Speech (TTS) voices for your organization. Voice availability depends on your TTS provider configuration.

List Available Voices

GET /voices

See List Available Voices for the complete API reference.

Voice Providers

Local TTS (Free for All Users)

Our local TTS service provides cost-efficient, low-latency voice synthesis with 96 voices across 35+ languages. Unlike cloud TTS providers, local voices are:

Free - No per-minute charges

Fast - Low-latency synthesis

Private - Audio processed on our infrastructure

Reliable - No third-party API dependencies

Why use local voices?

Local voices have a slightly robotic quality which research shows can actually improve user engagement in phone conversations. Users feel more comfortable being direct and concise with a clearly-AI voice, reducing awkward social dynamics. Learn more about voice psychology

Available Languages:

The /voices endpoint returns all 96 available voices grouped by language:

Language	Voices	Sample IDs
Dutch (NL)	2	`nl_NL-pim-medium`, `nl_NL-ronnie-medium`
Dutch (BE)	2	`nl_BE-nathalie-medium`, `nl_BE-rdh-medium`
English (UK)	9	`en_GB-alan-medium`, `en_GB-alba-medium`, `en_GB-cori-medium`
English (US)	18	`en_US-amy-medium`, `en_US-joe-medium`, `en_US-ryan-medium`
German	3	`de_DE-thorsten-medium`, `de_DE-thorsten_emotional-medium`
French	3	`fr_FR-siwis-medium`, `fr_FR-tom-medium`
Spanish (ES/MX/AR)	5	`es_ES-davefx-medium`, `es_MX-ald-medium`
Italian	1	`it_IT-paola-medium`
Portuguese (BR/PT)	4	`pt_BR-cadu-medium`, `pt_PT-tugão-medium`
Russian	4	`ru_RU-denis-medium`, `ru_RU-dmitri-medium`
Polish	3	`pl_PL-darkman-medium`, `pl_PL-gosia-medium`
+ 24 more languages	42	Arabic, Catalan, Czech, Danish, Farsi, Finnish, Hindi, Hungarian, Icelandic, Georgian, Kazakh, Luxembourgish, Latvian, Malayalam, Nepali, Norwegian, Romanian, Slovak, Slovenian, Serbian, Swedish, Swahili, Turkish, Ukrainian, Vietnamese, Chinese

Use the /voices endpoint to get the complete list with voice IDs, names, and bilingual descriptions.

ElevenLabs (BYOK)

When using ElevenLabs with your own API key, this endpoint does NOT return ElevenLabs voices.

To use ElevenLabs voices:

Check your ElevenLabs Voice Library

You only have access to voices you've added to your Voice Library

You only have access to custom voices you've created

Copy the Voice ID from ElevenLabs

Use that Voice ID in your agent's tts_config.voice_id field

Example agent with ElevenLabs voice:

{
  "name": "Customer Support",
  "tts_config": {
    "provider": "elevenlabs",
    "voice_id": "21m00Tcm4TlvDq8ikWAM",
    "model": "eleven_flash_v2_5",
    "stability": 0.5,
    "similarity_boost": 0.75
  },
  "llm_config": {
    "provider": "openai",
    "model": "gpt-4o-realtime"
  }
}

See the BYOK documentation for setup instructions.

xAI Grok Realtime (BYOK)

xAI Grok Realtime provides built-in ultra-low latency voices optimized for real-time conversations. These voices are not returned by the /voices endpoint because they're natively integrated into the Grok Realtime model.

Available xAI Realtime Voices:

Voice ID	Name	Style	Best For
`ara`	Ara	Female, professional	Business, support
`rex`	Rex	Male, friendly	Sales, casual
`sal`	Sal	Neutral, calm	Support, instructions
`eve`	Eve	Female, warm	Customer service
`leo`	Leo	Male, energetic	Sales, marketing

To use xAI Realtime voices:

Set the voice field in llm_config when creating or updating an agent:

{
  "name": "Sales Agent",
  "llm_config": {
    "provider": "xai_realtime",
    "model": "grok-realtime-v1",
    "voice": "ara"
  }
}

INFO

xAI Realtime combines LLM and TTS in a single model for ultra-low latency. Voice selection is part of the model configuration, not a separate TTS step.

See the BYOK documentation for xAI configuration.

Request

Response

Local TTS Response

{
  "voices": [
    {
      "id": "nl_NL-pim-medium",
      "name": "Pim",
      "language": "nl",
      "description": {
        "en": "Pim (Dutch, Male)",
        "nl": "Pim (Nederlands, Man)"
      }
    },
    {
      "id": "en_US-amy-medium",
      "name": "Amy",
      "language": "en",
      "description": {
        "en": "Amy (English US, Female)",
        "nl": "Amy (Engels US, Vrouw)"
      }
    },
    {
      "id": "de_DE-thorsten-medium",
      "name": "Thorsten",
      "language": "de",
      "description": {
        "en": "Thorsten (German, Male)",
        "nl": "Thorsten (Duits, Man)"
      }
    }
  ]
}

ElevenLabs BYOK Response

{
  "voices": []
}

The array is empty because ElevenLabs voices must be retrieved from your ElevenLabs account directly.

xAI Realtime Response

{
  "voices": []
}

The array is empty because xAI Realtime voices are built into the model and configured via llm_config.voice.

Voice Selection Best Practices

For Local TTS

Match voice language to your agent's primary language

Test different voices to find the best fit for your use case

Consider the robotic advantage - Users often respond better to clearly-AI voices

Use for cost-sensitive applications - No per-minute charges

Best for:

High-volume campaigns

Transactional calls (appointments, reminders)

Support/FAQ bots

Internal tools

For ElevenLabs

Use consistent Voice IDs across your agents

Keep a mapping of Voice IDs to names in your system

Test voices in your actual use case before deploying

Consider voice cloning for brand consistency

Best for:

Sales and marketing

Brand-critical interactions

Emotional/empathetic conversations

High-value customer touchpoints

For xAI Realtime

Choose voices based on agent personality and use case

Professional contexts → ara or sal

Sales/marketing → rex or leo

Customer service → eve or sal

Test latency and voice quality in your region

Best for:

Ultra-low latency requirements

Real-time conversations

Interactive voice experiences

Complex dialogue flows

Agents API - Configure agent voices

Voice Selection Psychology - Why robotic voices can be better

BYOK - Set up ElevenLabs or xAI

List Available Voices#

Voice Providers#

Local TTS (Free for All Users)#

ElevenLabs (BYOK)#

xAI Grok Realtime (BYOK)#

Request#

Response#

Local TTS Response#

ElevenLabs BYOK Response#

xAI Realtime Response#

Voice Selection Best Practices#

For Local TTS#

For ElevenLabs#

For xAI Realtime#

Related Documentation#

List Available Voices

Voice Providers

Local TTS (Free for All Users)

ElevenLabs (BYOK)

xAI Grok Realtime (BYOK)

Request

Response

Local TTS Response

ElevenLabs BYOK Response

xAI Realtime Response

Voice Selection Best Practices

For Local TTS

For ElevenLabs

For xAI Realtime

Related Documentation