> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vivix.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Choose a voice

Choose a built-in voice, add it to your session request, and test it with your own lines. The voices below do not require a separate provider account.

## 1. Start with Qwen Audio

Qwen Audio is the default speech service. Start with Long An Huan and supply only its voice ID.

```json theme={null}
{
  "pipeline_config": {
    "tts_config": {
      "tts_voice_id": "longanhuan_v3.6"
    }
  }
}
```

| Voice | Gender | Voice ID / language and sound |
| - | - | - |
| **Long An Feng Yue** | Female | `longanfengyue`  <br />Mandarin and English · Natural and friendly |
| **Long An Yuan Fei** | Female | `longanyuanfei`  <br />Mandarin and English · Poised and commanding |
| **Long An Ling Xi** | Female | `longanlingxi`  <br />Mandarin and English · Sweet and bright |
| **Long An Xiao Xin** | Female | `longanxiaoxin`  <br />Mandarin and English · Friendly and lively |
| **Long An Huan** | Female | `longanhuan_v3.6`  <br />Mandarin and English · Lively and playful |
| **Long Jie Li Dou** | Male | `longjielidou_v3.6`  <br />Mandarin and English · Youthful boy voice |
| **Long Pao Pao** | Female | `longpaopao_v3.6`  <br />Mandarin and English · Soft young girl voice |
| **Long Huo Huo** | Male | `longhuohuo_v3.6`  <br />Mandarin and English · Playful youthful voice |
| **Long Chuan Shu** | Male | `longchuanshu_v3.6`  <br />Mandarin and English · Mature male voice with a Sichuan accent |
| **Mary** | Female | `loongmary`  <br />English · Warm British voice |
| **Eva** | Female | `loongeva_v3.6`  <br />English · Composed American voice |
| **John** | Male | `loongjohn`  <br />English · Calm, friendly American voice |

## 2. Choose an ElevenLabs voice

When changing speech services, provide the `provider`, model, and voice ID together. This example selects George.

```json theme={null}
{
  "pipeline_config": {
    "tts_config": {
      "tts_provider": "elevenlabs",
      "tts_model_id": "eleven_v3",
      "tts_voice_id": "JBFqnCBsd6RMkjVDRZzb"
    }
  }
}
```

| Voice | Gender | Voice ID / language and sound |
| - | - | - |
| **George** | Male | `JBFqnCBsd6RMkjVDRZzb`  <br />Warm, mature British male voice |
| **Charlie** | Male | `IKne3meq5aSn9XLyUdCD`  <br />Deep, energetic Australian male voice |
| **Jessica** | Female | `cgSgspJ2msm6clMCkdW9`  <br />Bright, warm American female voice |

## 3. Create a session and listen

Merge `pipeline_config` into your Create Session request, keeping the images and character instructions. Test product names, numbers, and mixed-language lines in the new session. You can also save the configuration in a Character for reuse.

## 4. Clone a voice

Upload a recording through the [Voices API](/streaming-avatar/api-references/voices), then put the returned `voice_id` in `tts_voice_id`. When switching from another provider, replace the previous `tts_config` with this configuration.

```json theme={null}
{
  "pipeline_config": {
    "tts_config": {
      "tts_voice_id": "VOICE_ID_FROM_CLONE_RESPONSE"
    }
  }
}
```