Name: IndexTTS-2 API: Emotional TTS and Voice Cloning
Brand: magicapi
SKU: api-magicapi-emotionally-expressive-tts-lucataco-indextts-2
Availability: InStock
Author: API Market

Shashank Agarwal

# 🗣️ Expressive Text-to-Speech and Voice Cloning API (Bilibili IndexTTS-2)

## Overview
IndexTTS-2 is a text-to-speech model from Bilibili's Index team. Its developer describes it as emotionally expressive text-to-speech.

Send the `text` to speak and a `speaker_audio` recording of the target voice; the model reads the text in that voice. The emotion of the speech comes from the speaker recording by default. You can instead give a separate `emotion_audio` clip, describe the emotion in `emotion_text`, or pass 8 emotion weights in `emotion_vector`. Each request returns the URL of one WAV file.

## Features
- **Voice from a reference clip**: `speaker_audio` (WAV, 16 to 48 kHz) sets the voice; no training step is needed.
- **Separate emotion reference**: `emotion_audio` lets one clip set the voice and another set the emotion; `emotion_scale` (0 to 1) blends them.
- **Emotion from text**: `emotion_text` describes the emotion, which is detected automatically.
- **Manual emotion weights**: `emotion_vector` takes 8 weights as a comma-separated or JSON list and skips the classifier.
- **Long text handling**: text is split into segments of up to `max_text_tokens_per_segment` tokens, with `interval_silence_ms` of silence between them.
- **Sampling controls**: `temperature`, `top_p`, `top_k`, `num_beams`, `repetition_penalty` and `length_penalty`.

## Use Cases
1. **Narration in a chosen voice**: voice-overs and audiobook passages read in the voice of a consenting speaker.
2. **Character dialogue**: the same voice with different emotions per line, set by `emotion_audio` or `emotion_text`.
3. **Product and app audio**: spoken prompts, notifications and tutorials.
4. **Localization drafts**: read translated scripts in a reference voice for review.

## Visual Examples
### Example Input
Text: "Every long journey starts with one small step, and today is the day to take it. Start now with IndexTTS 2 on api.market."

Speaker reference: [Listen to the speaker audio](https://blog.api.market/wp-content/uploads/2026/10/indextts-2-input-api-market.mp3)

All other fields were sent with their default values (no emotion audio, text or vector).

### Example Output
[Listen to the example output](https://blog.api.market/wp-content/uploads/2026/10/indextts-2-output-parmb2kw-api-market.wav) (WAV)

## Input Parameters
| Field | Type | Required | Default | Allowed values | Description |
| --- | --- | --- | --- | --- | --- |
| `text` | string | Yes |  |  | Text to synthesize |
| `speaker_audio` | string (URL) | Yes |  |  | Reference audio URL for the target voice (WAV, 16 to 48 kHz) |
| `emotion_audio` | string (URL) | No | none |  | Optional emotion reference audio URL. Defaults to the speaker audio when omitted. |
| `emotion_scale` | number | No | `1` | 0 to 1 | Blend ratio for the emotion reference when both speaker and emotion references are used |
| `emotion_vector` | string | No | none |  | Optional comma-separated or JSON list of 8 emotion weights; skips the emotion classifier |
| `emotion_text` | string | No | none |  | Optional text describing the emotion; the emotion is detected from it automatically |
| `randomize_emotion` | boolean | No | `false` |  | Pick emotion embeddings randomly instead of nearest-neighbour selection when emotion weights are provided |
| `interval_silence_ms` | integer | No | `200` | 0 to 2000 | Silence inserted between long segments, in milliseconds |
| `max_text_tokens_per_segment` | integer | No | `120` | 32 to 300 | Maximum text tokens per generated segment |
| `top_p` | number | No | `0.8` | 0 to 1 | Top-p nucleus sampling |
| `top_k` | integer | No | `30` | 1 to 200 | Top-k sampling |
| `temperature` | number | No | `0.8` | 0 to 2 | Sampling temperature |
| `length_penalty` | number | No | `0` | 0 to 5 | Beam search length penalty |
| `num_beams` | integer | No | `3` | 1 to 8 | Beam width |
| `repetition_penalty` | number | No | `10` | 1 to 30 | Penalty for repeated tokens |
| `max_mel_tokens` | integer | No | `1500` | 256 to 4096 | Maximum audio (mel) tokens to generate per segment |

## Pricing
- **Pay As You Go (PAYG) pricing**: you only pay for what you use.
- **Price**: $0.0049 per API unit.
- **1 API unit = 1 request with up to 4 seconds of processing.** A request that takes longer is billed one more unit for each started 4-second block of processing time.
- **Example cost**: the example request above took 7.67 seconds of processing, which is 2 started 4-second blocks, so it cost 2 units = $0.0098.
- Longer text takes longer to synthesize and can use more units.
- **Failed or cancelled predictions are not billed.**
- **GET requests to check prediction status** are free and don't count toward API unit usage.
- **No monthly subscription fee.**

## API Documentation

### Base URL
All API requests go to: `https://prod.api.market/api/v1/magicapi/emotionally-expressive-tts-lucataco-indextts-2`

### Authentication
Send your api.market API key in the `x-api-market-key` header:
```
x-api-market-key: YOUR_API_MARKET_KEY
```

### Asynchronous Processing
**IMPORTANT**: This API is asynchronous. The flow is:
1. Send a POST request to `/predictions`. A successful request returns **201 Created** with a prediction `id` and a `status` of `starting` or `processing`.
2. Poll GET `/predictions/{id}` with that ID.
3. Keep polling while `status` is `starting` or `processing`.
4. Stop when `status` is `succeeded` (read `output`), `failed` or `canceled` (read `error`).

A polling interval of 1 second and an overall timeout of 600 seconds (10 minutes) work well.

### Endpoints

#### Create a Prediction
```bash
curl -X POST "https://prod.api.market/api/v1/magicapi/emotionally-expressive-tts-lucataco-indextts-2/predictions" \
  -H "Content-Type: application/json" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY" \
  -d '{
    "version": "b219b0f22f95fd97cb2c8e3bbea6827a450a7fff05674c996d83171d70b3f685",
    "input": {
      "text": "Every long journey starts with one small step, and today is the day to take it. Start now with IndexTTS 2 on api.market.",
      "top_k": 30,
      "top_p": 0.8,
      "num_beams": 3,
      "temperature": 0.8,
      "emotion_scale": 1,
      "speaker_audio": "https://blog.api.market/wp-content/uploads/2026/10/indextts-2-input-api-market.mp3",
      "length_penalty": 0,
      "max_mel_tokens": 1500,
      "randomize_emotion": false,
      "repetition_penalty": 10,
      "interval_silence_ms": 200,
      "max_text_tokens_per_segment": 120
    }
  }'
```

Response (201 Created):
```json
{
  "id": "parmb2kwdxrmr0d1d209f9z56w",
  "version": "b219b0f22f95fd97cb2c8e3bbea6827a450a7fff05674c996d83171d70b3f685",
  "status": "starting",
  "created_at": "2026-10-23T11:23:33.520Z"
}
```

#### Check Prediction Status
```bash
curl -X GET "https://prod.api.market/api/v1/magicapi/emotionally-expressive-tts-lucataco-indextts-2/predictions/{id}" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY"
```

Response when the prediction has finished (200 OK):
```json
{
  "id": "parmb2kwdxrmr0d1d209f9z56w",
  "status": "succeeded",
  "output": "https://blog.api.market/wp-content/uploads/2026/10/indextts-2-output-parmb2kw-api-market.wav",
  "error": null,
  "created_at": "2026-10-23T11:23:33.520Z",
  "started_at": "2026-10-23T11:24:25.185Z",
  "completed_at": "2026-10-23T11:24:32.855Z"
}
```

`output` is a single URL string pointing to the generated WAV file.

### Status Codes
- **201 Created**: prediction created (POST `/predictions`). This is the normal success response.
- **200 OK**: prediction details returned (GET `/predictions/{id}`).
- **400 Bad Request**: invalid input, for example a missing `text` or `speaker_audio`, or a `num_beams` above 8.
- **401 Unauthorized**: missing or invalid API key.
- **404 Not Found**: no prediction with that ID.
- **429 Too Many Requests**: rate limit exceeded.
- **500 Internal Server Error** or **503 Service Unavailable**: temporary server problem; retry later.

## Rate Limits
All plans allow 10 requests per second. Requests above that return `429 Too Many Requests`; back off before retrying and honour any `Retry-After` header. Polling a prediction with GET is free but still counts toward the per-second limit, so poll every 1 to 5 seconds rather than in a tight loop.

## Error Handling
- Check the HTTP status code of every request and read the response body for the error message.
- A prediction can also end with `status` set to `failed`; the `error` field then explains why. Failed predictions are not billed.
- Retry 429, 500 and 503 responses with exponential backoff. Fix the input before retrying a 400.

## Implementation Guide
### Example Code

#### JavaScript
```javascript
const apiKey = "YOUR_API_MARKET_KEY";
const baseUrl = "https://prod.api.market/api/v1/magicapi/emotionally-expressive-tts-lucataco-indextts-2";

async function createAndWaitForPrediction(input) {
  const response = await fetch(`${baseUrl}/predictions`, {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "x-api-market-key": apiKey
    },
    body: JSON.stringify({ version: "b219b0f22f95fd97cb2c8e3bbea6827a450a7fff05674c996d83171d70b3f685", input })
  });

  if (response.status !== 201) {
    throw new Error(`API error: ${response.status} ${await response.text()}`);
  }

  const prediction = await response.json();
  return pollForResults(prediction.id);
}

const delay = (ms) => new Promise((resolve) => setTimeout(resolve, ms));

async function pollForResults(predictionId) {
  const maxTimeout = 600000;
  const pollInterval = 1000;
  const startTime = Date.now();

  while (Date.now() - startTime < maxTimeout) {
    const response = await fetch(`${baseUrl}/predictions/${predictionId}`, {
      headers: { "x-api-market-key": apiKey }
    });

    if (!response.ok) {
      throw new Error(`API error: ${response.status} ${await response.text()}`);
    }

    const prediction = await response.json();

    if (prediction.status === "succeeded") {
      return prediction.output;
    }

    if (prediction.status === "failed" || prediction.status === "canceled") {
      throw new Error(`Prediction ${prediction.status}: ${prediction.error || "Unknown error"}`);
    }

    await delay(pollInterval);
  }

  throw new Error("Prediction timed out after 10 minutes");
}

createAndWaitForPrediction({
    "text": "Every long journey starts with one small step, and today is the day to take it. Start now with IndexTTS 2 on api.market.",
    "top_k": 30,
    "top_p": 0.8,
    "num_beams": 3,
    "temperature": 0.8,
    "emotion_scale": 1,
    "speaker_audio": "https://blog.api.market/wp-content/uploads/2026/10/indextts-2-input-api-market.mp3",
    "length_penalty": 0,
    "max_mel_tokens": 1500,
    "randomize_emotion": false,
    "repetition_penalty": 10,
    "interval_silence_ms": 200,
    "max_text_tokens_per_segment": 120
  }).then((output) => console.log(output));
```

#### Python
```python
import time
import requests

API_KEY = "YOUR_API_MARKET_KEY"
BASE_URL = "https://prod.api.market/api/v1/magicapi/emotionally-expressive-tts-lucataco-indextts-2"


def create_and_wait_for_prediction(input_data):
    headers = {"Content-Type": "application/json", "x-api-market-key": API_KEY}
    response = requests.post(
        f"{BASE_URL}/predictions",
        headers=headers,
        json={"version": "b219b0f22f95fd97cb2c8e3bbea6827a450a7fff05674c996d83171d70b3f685", "input": input_data},
    )

    if response.status_code != 201:
        raise Exception(f"API error: {response.status_code} {response.text}")

    return poll_for_results(response.json()["id"])


def poll_for_results(prediction_id):
    max_timeout = 600
    poll_interval = 1
    headers = {"x-api-market-key": API_KEY}
    start_time = time.time()

    while time.time() - start_time < max_timeout:
        response = requests.get(f"{BASE_URL}/predictions/{prediction_id}", headers=headers)

        if response.status_code != 200:
            raise Exception(f"API error: {response.status_code} {response.text}")

        prediction = response.json()

        if prediction["status"] == "succeeded":
            return prediction["output"]

        if prediction["status"] in ("failed", "canceled"):
            raise Exception(f"Prediction {prediction['status']}: {prediction.get('error') or 'Unknown error'}")

        time.sleep(poll_interval)

    raise Exception("Prediction timed out after 10 minutes")


print(create_and_wait_for_prediction({
    "text": "Every long journey starts with one small step, and today is the day to take it. Start now with IndexTTS 2 on api.market.",
    "top_k": 30,
    "top_p": 0.8,
    "num_beams": 3,
    "temperature": 0.8,
    "emotion_scale": 1,
    "speaker_audio": "https://blog.api.market/wp-content/uploads/2026/10/indextts-2-input-api-market.mp3",
    "length_penalty": 0,
    "max_mel_tokens": 1500,
    "randomize_emotion": False,
    "repetition_penalty": 10,
    "interval_silence_ms": 200,
    "max_text_tokens_per_segment": 120
}))
```

## FAQ
- **What audio works best as `speaker_audio`?**
  - A clean WAV recording of one speaker, sampled at 16 to 48 kHz. Only use voices you have permission to use.
- **How do I change the emotion without changing the voice?**
  - Keep the voice in `speaker_audio` and pass a different clip as `emotion_audio`, describe the emotion in `emotion_text`, or set 8 weights in `emotion_vector`. `emotion_scale` sets how strongly an emotion clip is applied.
- **Can I set the length of the speech?**
  - No field sets the duration directly. Long text is split into segments (`max_text_tokens_per_segment`) with `interval_silence_ms` of silence between them.
- **What format is the output?**
  - A WAV file, returned as one URL in `output`.
- **What if my prediction fails?**
  - Failed predictions are not billed. Read the `error` field, adjust the input and try again.

## Contact Us
If you have any questions, reach out to us through the chat on the bottom right of the page. You can also email our support team at hello@magicapi.com.


Generate speech in the voice of a reference clip with IndexTTS-2 and control its emotion from a second clip, a text description or 8 emotion weights.