# LTX-2.3 Text-to-Audio

> Generate sound effects, ambience, and spoken-style audio from a text prompt, with duration you control down to the frame.

## Overview

- **Endpoint**: `https://queue.modelrunner.run/lightricks/ltx-2.3/text-to-audio`
- **Model ID**: `lightricks/ltx-2.3/text-to-audio`
- **Category**: sound
- **Kind**: inference
- **Tags**: lightricks, ltx-2.3, text-to-audio, audio, sound

## Pricing

- **Price**: $0.1 per output second

## Request Lifecycle

This model runs on the ModelRunner **asynchronous queue API** — a single POST does not return the output. Every call requires an `Authorization: Key $MODEL_RUNNER_KEY` header. Run three steps:

1. **Submit** — `POST https://queue.modelrunner.run/lightricks/ltx-2.3/text-to-audio` with a JSON body holding the input fields at the top level. The body may also include a reserved top-level `metadata` object — a flat string map (max 16 keys, key ≤64 / value ≤512 chars) stored on the request for your own tagging. It is never sent to the model; filter your request history with `GET https://queue.modelrunner.run/requests?metadata=<url-encoded JSON>` (exact key=value matches, AND-ed). The response carries request handles only (no output yet):

```json
{
  "status": "IN_QUEUE",
  "request_id": "<21-char id>",
  "status_url": "https://queue.modelrunner.run/lightricks/ltx-2.3/text-to-audio/requests/<request_id>/status",
  "response_url": "https://queue.modelrunner.run/lightricks/ltx-2.3/text-to-audio/requests/<request_id>",
  "cancel_url": "https://queue.modelrunner.run/lightricks/ltx-2.3/text-to-audio/requests/<request_id>/cancel"
}
```

2. **Poll status** — `GET <status_url>` until `status` is `COMPLETED`. Possible values are `IN_QUEUE`, `IN_PROGRESS`, `COMPLETED`, `FAILED`, `CANCELLED`. A `FAILED` request responds with HTTP 400 and an `error` field.

3. **Read result** — `GET <response_url>`. Returns the finished request, including the generated `output`:

```json
{ "id": "<request_id>", "status": "COMPLETED", "output": ..., "input": ... }
```

The JavaScript and Python SDKs below perform steps 2–3 for you. In any language without an SDK (Swift, Go, Kotlin, etc.) you must implement the polling loop and the final result fetch yourself — see the cURL example for the full flow.

### Input Schema

- **`seed`** (`integer`, _optional_):
  Random seed for reproducible generation. Leave empty for a random result.

- **`prompt`** (`string`, _required_):
  The prompt to generate the audio from.

- **`num_frames`** (`integer`, _optional_):
  Audio duration in LTX video-frame units. Duration in seconds is approximately num_frames / frames_per_second. Default 121 frames at 24 fps is about 5 seconds; the maximum 481 frames is about 20 seconds.
  - Default: `121`
  - Range: `9` to `481`

- **`guidance_scale`** (`number`, _optional_):
  Classifier-free guidance scale; higher values follow the prompt more strictly.
  - Default: `1`
  - Range: `1` to `20`

- **`negative_prompt`** (`string`, _optional_):
  Qualities to steer the generated audio away from.
  - Default: `"pc game, console game, video game, cartoon, childish, ugly"`

- **`frames_per_second`** (`number`, _optional_):
  Frame rate used to derive the clip duration together with num_frames.
  - Default: `24`
  - Range: `1` to `60`

- **`num_inference_steps`** (`integer`, _optional_):
  Number of denoising steps. More steps can improve quality at the cost of speed.
  - Default: `15`
  - Range: `8` to `30`

- **`enable_safety_checker`** (`boolean`, _optional_):
  If true, run a content-safety check during generation.
  - Default: `true`

- **`enable_prompt_expansion`** (`boolean`, _optional_):
  If true, automatically elaborate the prompt before generation.
  - Default: `false`

### Output Schema

_No `Output` schema properties are available._

## Default Example

**Input**

```json
{
  "prompt": "A woman says: 'Welcome to ModelRunner.' Natural speech, clean studio audio.",
  "num_frames": 121,
  "guidance_scale": 1,
  "negative_prompt": "pc game, console game, video game, cartoon, childish, ugly",
  "frames_per_second": 24,
  "num_inference_steps": 15,
  "enable_safety_checker": true,
  "enable_prompt_expansion": false
}
```

**Output**

```json
"https://media.modelrunner.ai/3BXZR5YWKQW1iWymUVzRE.mp3"
```
## Usage Examples

### cURL

The queue API is asynchronous: submit the request, poll `status_url` until it is `COMPLETED`, then read the result from `response_url`. Requires `jq`.

```bash
# 1. Submit the request (returns request handles, not the output)
SUBMIT=$(curl --silent --request POST \
  --url https://queue.modelrunner.run/lightricks/ltx-2.3/text-to-audio \
  --header "Authorization: Key $MODEL_RUNNER_KEY" \
  --header "Content-Type: application/json" \
  --data '{
  "prompt": "A woman says: '\''Welcome to ModelRunner.'\'' Natural speech, clean studio audio.",
  "num_frames": 121,
  "guidance_scale": 1,
  "negative_prompt": "pc game, console game, video game, cartoon, childish, ugly",
  "frames_per_second": 24,
  "num_inference_steps": 15,
  "enable_safety_checker": true,
  "enable_prompt_expansion": false
}')

STATUS_URL=$(echo "$SUBMIT" | jq -r '.status_url')
RESPONSE_URL=$(echo "$SUBMIT" | jq -r '.response_url')

# 2. Poll until the request leaves the queue / in-progress state
while true; do
  STATUS=$(curl --silent --url "$STATUS_URL" \
    --header "Authorization: Key $MODEL_RUNNER_KEY" | jq -r '.status')
  echo "Status: $STATUS"
  case "$STATUS" in
    COMPLETED) break ;;
    FAILED|CANCELLED) echo "Request $STATUS"; exit 1 ;;
  esac
  sleep 1
done

# 3. Read the finished request, including the generated output
curl --silent --url "$RESPONSE_URL" \
  --header "Authorization: Key $MODEL_RUNNER_KEY"
```

### JavaScript

```javascript
import { modelrunner } from "@modelrunner/client";

const result = await modelrunner.subscribe("lightricks/ltx-2.3/text-to-audio", {
  input: {
  "prompt": "A woman says: 'Welcome to ModelRunner.' Natural speech, clean studio audio.",
  "num_frames": 121,
  "guidance_scale": 1,
  "negative_prompt": "pc game, console game, video game, cartoon, childish, ugly",
  "frames_per_second": 24,
  "num_inference_steps": 15,
  "enable_safety_checker": true,
  "enable_prompt_expansion": false
}
});

console.log(result.data);
```

### Python

```python
import asyncio
import modelrunner_ai

async def main():
    response = await modelrunner_ai.submit_async(
        "lightricks/ltx-2.3/text-to-audio",
        arguments={
  "prompt": "A woman says: 'Welcome to ModelRunner.' Natural speech, clean studio audio.",
  "num_frames": 121,
  "guidance_scale": 1,
  "negative_prompt": "pc game, console game, video game, cartoon, childish, ugly",
  "frames_per_second": 24,
  "num_inference_steps": 15,
  "enable_safety_checker": true,
  "enable_prompt_expansion": false
}
    )
    result = await response.get()
    print(result["output"])

asyncio.run(main())
```

## Additional Resources

- [Playground](https://modelrunner.ai/models/lightricks/ltx-2.3/text-to-audio)
- [OpenAPI Schema](https://modelrunner.ai/models/lightricks/ltx-2.3/text-to-audio/openapi.json)
- [LLM Instructions](https://modelrunner.ai/models/lightricks/ltx-2.3/text-to-audio/llms.txt)