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

# MiniMax Speech (T2A)

> MiniMax text-to-speech models, OpenAI-compatible audio API.

The MiniMax speech models are available through Upmore via the standard audio API (`/v1/audio/speech`). Send text, get audio bytes back — billing is per input character.

## Available models

| Model                                | Notes                            |
| ------------------------------------ | -------------------------------- |
| `speech-2.8-hd`                      | Latest generation, high fidelity |
| `speech-2.8-turbo`                   | Latest generation, low latency   |
| `speech-2.6-hd` / `speech-2.6-turbo` | Previous generation              |
| `speech-02-hd` / `speech-02-turbo`   | Legacy                           |
| `speech-01-hd` / `speech-01-turbo`   | Legacy                           |

<Note>
  `hd` variants prioritize audio fidelity; `turbo` variants prioritize latency and cost less. All models share the same request format.
</Note>

## Quick example

<CodeGroup>
  ```bash cURL theme={null}
  curl https://api.upmore.net/v1/audio/speech \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "speech-2.8-hd",
      "input": "你好，欢迎使用语音合成服务。",
      "voice": "male-qn-qingse",
      "response_format": "mp3"
    }' --output speech.mp3
  ```

  ```python Python theme={null}
  from openai import OpenAI

  client = OpenAI(
      api_key="YOUR_API_KEY",
      base_url="https://api.upmore.net/v1"
  )

  response = client.audio.speech.create(
      model="speech-2.8-hd",
      voice="male-qn-qingse",
      input="你好，欢迎使用语音合成服务。",
      response_format="mp3",
  )

  response.write_to_file("speech.mp3")
  ```

  ```typescript TypeScript theme={null}
  import fs from 'node:fs'
  import OpenAI from 'openai'

  const client = new OpenAI({
    apiKey: process.env.UPMORE_API_KEY,
    baseURL: 'https://api.upmore.net/v1',
  })

  const speech = await client.audio.speech.create({
    model: 'speech-2.8-hd',
    voice: 'male-qn-qingse',
    input: '你好，欢迎使用语音合成服务。',
    response_format: 'mp3',
  })

  fs.writeFileSync('speech.mp3', Buffer.from(await speech.arrayBuffer()))
  ```
</CodeGroup>

## Parameters

| Parameter         | Type   | Required | Description                                                                        |
| ----------------- | ------ | -------- | ---------------------------------------------------------------------------------- |
| `model`           | string | Yes      | One of the model IDs from the table above                                          |
| `input`           | string | Yes      | Text to synthesize, billed per character                                           |
| `voice`           | string | Yes      | MiniMax voice ID (e.g. `male-qn-qingse`, `female-shaonv`) — not OpenAI voice names |
| `response_format` | string | No       | `mp3` (default), `wav`, `flac`, `aac`, or `pcm`                                    |
| `speed`           | number | No       | Playback speed, `0.5` – `2`. Default: `1`                                          |
| `metadata`        | object | No       | Provider-specific fields merged into the upstream request                          |

## Advanced: provider passthrough

Anything MiniMax's native T2A API supports can be passed through `metadata` — the object is merged into the upstream request body:

```json theme={null}
"metadata": {
  "voice_setting": { "emotion": "happy", "pitch": 0 },
  "audio_setting": { "sample_rate": 32000, "bitrate": 128000 },
  "language_boost": "Chinese"
}
```

By default the platform returns the audio bytes directly. Set `"metadata": {"output_format": "url"}` to receive a `302` redirect to a temporary audio URL instead (use `curl -L`).

<Card title="API Reference" icon="code" href="/api-reference/model-api/minimax/speech">
  View the interactive API playground.
</Card>