Miso One API

Overview

The Miso One API turns a written multi-speaker script into an audio file. Send a dialogue request, receive a task ID, then poll the task until the audio result is ready.

Miso One currently authenticates with your logged-in web session. Include your session cookie when calling the API from a browser session or a trusted tool that can access that session.

Authentication

Cookie: <session cookie>

If the session is missing or expired, the API returns an auth error.

Create a dialogue task

POST /api/ai/generate
Content-Type: application/json
Cookie: <session cookie>

Request body

{
  "mediaType": "speech",
  "presetId": "miso-one",
  "options": {
    "dialogue": [
      { "voice": "Adam", "text": "[excited] Did you hear the news?" },
      { "voice": "Brian", "text": "[laughs] Tell me everything." }
    ],
    "language_code": "en"
  }
}

Field	Type	Required	Description
`mediaType`	string	yes	Always `"speech"`
`presetId`	string	yes	Always `"miso-one"`
`options.dialogue`	array	yes	Ordered speaker lines. Each item needs `voice` and `text`.
`options.language_code`	string	no	Language hint such as `"en"`, `"zh"`, `"ja"`, or `"ko"`.
`options.stability`	number	no	Voice stability setting. Omit this field to use the default value.
`prompt`	string	no	Optional note saved with the task.

Dialogue line format

{
  "voice": "Adam",
  "text": "[excited] The launch went perfectly."
}

Use one dialogue line per speaker turn. The total spoken text in a single request can include up to 5,000 characters.

Poll task status

The create endpoint returns immediately with a task object. Use the task id to check progress.

POST /api/ai/query
Content-Type: application/json
Cookie: <session cookie>

Query body

{
  "taskId": "task_xxx"
}

Pending response

{
  "code": 0,
  "message": "ok",
  "data": {
    "id": "task_xxx",
    "status": "processing",
    "mediaType": "speech",
    "presetId": "miso-one",
    "displayModel": "Miso One",
    "prompt": null,
    "taskInfo": null
  }
}

Completed response

When status is "success", parse data.taskInfo. It contains an audios array with the generated audio URL.

{
  "code": 0,
  "message": "ok",
  "data": {
    "id": "task_xxx",
    "status": "success",
    "mediaType": "speech",
    "presetId": "miso-one",
    "displayModel": "Miso One",
    "taskInfo": "{\"audios\":[{\"id\":\"task_xxx-0\",\"audioUrl\":\"https://cdn.example.com/audio.mp3\"}]}"
  }
}

Task statuses:

Status	Meaning
`pending`	The task has been created.
`processing`	The audio is still being generated.
`success`	The audio is ready in `taskInfo.audios`.
`failed`	The task failed. Check `taskInfo.errorCode`.
`canceled`	The task was canceled.

Supported voices

Voice	Character
Adam	Warm, expressive male
Brian	Natural, conversational male
Roger	Clear, authoritative male
Rachel	Calm, professional female
Antoni	Smooth, focused male
Bella	Bright, energetic female
Elli	Friendly, approachable female
Josh	Deep, confident male
Arnold	Firm, grounded male
Clyde	Relaxed, casual male
Domi	Clear, neutral female
Freya	Warm, empathetic female
Grace	Light, pleasant female
Nicole	Soft, thoughtful female
Sam	Versatile, neutral
Sarah	Crisp, professional female

Emotion tags

Add emotion tags at the start of a line to guide delivery:

[excited] Great news!
[whispers] Can you keep a secret?
[laughs] That is hilarious.
[sighs] I was not expecting that.

Credit cost

Dialogue generation uses dynamic credit billing based on total spoken text:

Rule	Value
Min	30 credits per generation
Rate	500 credits per 1,000 characters
Limit	5,000 characters per request

curl example

APP_URL="https://misooneai.com"

TASK_RESPONSE=$(curl "$APP_URL/api/ai/generate" \
  -H "Content-Type: application/json" \
  -H "Cookie: <your-session-cookie>" \
  -d '{
    "mediaType": "speech",
    "presetId": "miso-one",
    "options": {
      "dialogue": [
        { "voice": "Adam", "text": "[excited] The launch went perfectly." },
        { "voice": "Rachel", "text": "Incredible. What happens next?" },
        { "voice": "Adam", "text": "We scale." }
      ],
      "language_code": "en"
    }
  }')

echo "$TASK_RESPONSE"

Then poll the returned task ID:

curl "$APP_URL/api/ai/query" \
  -H "Content-Type: application/json" \
  -H "Cookie: <your-session-cookie>" \
  -d '{"taskId":"task_xxx"}'

JavaScript example

const createResponse = await fetch('/api/ai/generate', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
    mediaType: 'speech',
    presetId: 'miso-one',
    options: {
      dialogue: [
        { voice: 'Adam', text: '[excited] The launch went perfectly.' },
        { voice: 'Rachel', text: 'Incredible. What happens next?' },
      ],
      language_code: 'en',
    },
  }),
});

const createPayload = await createResponse.json();
if (createPayload.code !== 0) {
  throw new Error(createPayload.message);
}

const taskId = createPayload.data.id;

const queryResponse = await fetch('/api/ai/query', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({ taskId }),
});

const queryPayload = await queryResponse.json();
const taskInfo = queryPayload.data.taskInfo
  ? JSON.parse(queryPayload.data.taskInfo)
  : {};
const audioUrl = taskInfo.audios?.[0]?.audioUrl;

Errors

All API responses use the same envelope:

{
  "code": -1,
  "message": "generate ai task failed",
  "data": {
    "errorCode": "auth_required"
  }
}

Common error codes:

Error code	Meaning
`auth_required`	The user is not logged in or the session expired.
`insufficient_credits`	The account does not have enough credits.
`invalid_input`	The request body is missing required fields.
`content_policy`	The dialogue content was rejected.

Miso One API

On this page