# Stable Audio 3 Trainer API

> Trains a LoRA on Stable Audio 3 from a zip of audio clips with matching text captions

Fine-tunes a base checkpoint (medium-base, small-music-base or small-sfx-base) on an `audio_data_url` zip where each audio file has a same-named `.txt` caption. You can tune `number_of_steps` (up to 20000), `learning_rate`, `rank` (1 to 256), `adapter_type` and `batch_size`, or resume from a `.safetensors` checkpoint. Returns LoRA weights and a config file.

- Model id: `stable-audio-3-trainer`
- Made by: Stability AI
- Category: Training
- Output: file
- Price per run: from $0.003 / run (cheapest run $0.003 (3 API credits))
- Run: `POST https://api.doitong.com/v1/run/stable-audio-3-trainer`
- Quote: `POST https://api.doitong.com/v1/quote/stable-audio-3-trainer`
- Docs page: https://doitong.com/developers/models/stable-audio-3-trainer

## Pricing

Price: **from $0.003 / run** (the lowest tier; the quote prices your exact input). Price per run: the input files are measured before the run (length, resolution, frame rate). The quote measures the same files and is exactly what the run charges.

The full example input below costs $3.00 = 3000 API credits (1 run × $3.00).

Prices are the upstream price in USD, paid from the API balance in API credits (1000 API credits = 1 USD). `POST https://api.doitong.com/v1/quote/stable-audio-3-trainer` with the same JSON body returns the exact price for that input; the quote is exactly what a run charges. Invalid input is never charged and failed or canceled requests are refunded automatically.

## Input parameters

| Parameter | Canonical name | Type | Required | Default | Allowed values / range | Description |
|---|---|---|---|---|---|---|
| `audio_data_url` |  | string | yes |  |  | Dataset zip URL, with each audio/caption pair stored together under matching basenames: `clip.wav` pairs with `clip.txt`, for example. Captions must use the `.txt` extension. |
| `model` |  | enum (string) | no | `medium-base` | `medium-base`, `small-music-base`, `small-sfx-base` | Base checkpoint for fine-tuning Stable Audio 3. |
| `number_of_steps` |  | integer | no | `1000` | 1 – 20,000 | Step count for LoRA training. |
| `learning_rate` |  | number | no | `0.0001` | ≤ 0.01, > 0 | Learning rate applied by AdamW to LoRA parameters. |
| `rank` |  | integer | no | `16` | 1 – 256 | Rank of the LoRA adapter. |
| `adapter_type` |  | enum (string) | no | `dora-rows` | `lora`, `dora`, `dora-rows`, `dora-cols`, `bora`, `lora-xs`, `dora-rows-xs`, `dora-cols-xs`, `bora-xs` | Adapter family used for LoRA training. |
| `duration` |  | number \| null | no |  | 1 – 380 | Crop/pad duration in seconds, limited in all cases by the selected model's native training length. Supply a length or omit it to use the longest clip detected in the dataset, subject to that cap. |
| `batch_size` |  | integer | no | `1` | 1 – 8 | Batch size used during training. |
| `seed` |  | integer | no | `42` | 0 – 2,147,483,647 | Seed for randomization. |
| `base_precision` |  | enum (string) | no | `bf16` | `bf16`, `bfloat16`, `fp16`, `float16` | Precision used by the frozen base weights; LoRA parameters remain fp32. |
| `include` |  | array of string \| null | no |  |  | Module-name substrings that restrict where LoRA is added: only matching modules receive it. |
| `exclude` |  | array of string \| null | no |  |  | Name substrings used to exclude matching modules. |
| `lora_checkpoint_url` |  | string \| null | no |  |  | Optional URL of a `.safetensors` LoRA checkpoint from which to resume. |
| `pre_encode` |  | boolean | no | `false` |  | Pre-encoding of the audio archive into SAME latents before training the LoRA. |

## Example input

Required parameters only:

```json
{
  "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
}
```

A fuller example:

```json
{
  "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
}
```

## Run it

Set `DOITONG_API_KEY` to a key from https://doitong.com/developers/dashboard#keys. `?wait=20` returns the finished request in the same call when it is ready within 20 seconds; otherwise poll `GET /requests/{id}` (it also accepts `?wait=`), or pass `webhook_url`.

### curl

```bash
# Run the model; wait up to 20 s for the finished result in the same call
curl -X POST "https://api.doitong.com/v1/run/stable-audio-3-trainer?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
  }'

# Still "queued" or "running"? Poll with the id from the response:
curl "https://api.doitong.com/v1/requests/REQUEST_ID?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY"
```

### JavaScript

```javascript
const API = "https://api.doitong.com/v1";
const headers = {
  "Authorization": "Bearer " + process.env.DOITONG_API_KEY,
  "Content-Type": "application/json",
};

// Run the model; wait up to 20 s for the finished result in the same call
let res = await fetch(API + "/run/stable-audio-3-trainer?wait=20", {
  method: "POST",
  headers,
  body: JSON.stringify({
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
  }),
});
let request = await res.json();
if (!res.ok) throw new Error(request.error.code + ": " + request.error.message);

// Not finished yet: long-poll until it is
while (request.status === "queued" || request.status === "running") {
  res = await fetch(API + "/requests/" + request.id + "?wait=20", { headers });
  request = await res.json();
}
if (request.status !== "succeeded") throw new Error(JSON.stringify(request.error));
console.log(request.outputs.map((o) => o.url), request.cost.credits + " API credits ($" + request.cost.usd + ")");
```

### Python

```python
import os
import requests

API = "https://api.doitong.com/v1"
headers = {"Authorization": "Bearer " + os.environ["DOITONG_API_KEY"]}

payload = {
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3",
}

# Run the model; wait up to 20 s for the finished result in the same call
r = requests.post(API + "/run/stable-audio-3-trainer", params={"wait": 20}, json=payload, headers=headers)
request = r.json()
if not r.ok:
    raise RuntimeError(request["error"]["code"] + ": " + request["error"]["message"])

# Not finished yet: long-poll until it is
while request["status"] in ("queued", "running"):
    request = requests.get(API + "/requests/" + request["id"], params={"wait": 20}, headers=headers).json()
if request["status"] != "succeeded":
    raise RuntimeError(request["error"])
print([o["url"] for o in request["outputs"]], request["cost"]["credits"], "API credits", "$" + request["cost"]["usd"])
```

### MCP

The MCP server at `https://api.doitong.com/v1/mcp` exposes the tools `search_models`, `get_model`, `quote`, `run`, `get_request`, `cancel_request`, `list_requests`, `create_upload` and `get_account`. A `run` call for this model:

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "run",
    "arguments": {
      "model_id": "stable-audio-3-trainer",
      "input": {
        "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
      },
      "wait_seconds": 20
    }
  }
}
```

## Response

Every call returns a request object. `status` is one of `queued`, `running`, `succeeded`, `failed`, `canceled`. `outputs` lists the generated files with URLs on our storage (kept forever, or 7 days with `?storage=temporary`; `storage` says which); `output` is the model's own result object with the same URLs.

```json
{
  "id": "req_8k2m4q7x1z9c3v5b6n0p2r4t",
  "object": "request",
  "model": "stable-audio-3-trainer",
  "status": "succeeded",
  "created_at": "2026-10-20T12:00:00Z",
  "started_at": "2026-10-20T12:00:02Z",
  "completed_at": "2026-10-20T12:00:09Z",
  "queue_position": null,
  "cost": {
    "credits": 3000,
    "usd": "3.00",
    "charged": 3000,
    "refunded": 0
  },
  "input": {
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
  },
  "output": {
    "lora_file": {
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_name": "0-output.jpeg",
      "file_size": 1843200
    },
    "config_file": {
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_name": "0-output.jpeg",
      "file_size": 1843200
    }
  },
  "outputs": [
    {
      "type": "file",
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_size": 1843200,
      "stored": true
    }
  ],
  "error": null,
  "metrics": {
    "inference_seconds": 6.8
  },
  "urls": {
    "self": "https://api.doitong.com/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t",
    "cancel": "https://api.doitong.com/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t/cancel"
  }
}
```

### Output fields (`output`)

| Field | Type | Description |
|---|---|---|
| `lora_file` | object {url, content_type, file_name, file_size} | LoRA weights trained for Stable Audio 3. |
| `config_file` | object {url, content_type, file_name, file_size} | JSON metadata describing the training run and the compatible model for inference. |

## Related models

- [ERNIE-Image Trainer](https://doitong.com/developers/models/ernie-image-trainer.md) (`ernie-image-trainer`): $0.0012 / step
- [FLUX 2 \[klein\] 9b Base Trainer](https://doitong.com/developers/models/flux-2-klein-9b-base-trainer.md) (`flux-2-klein-9b-base-trainer`): $0.0043 / step
- [Flux 2 Klein 9B Base Trainer](https://doitong.com/developers/models/flux-2-klein-9b-base-trainer/edit.md) (`flux-2-klein-9b-base-trainer/edit`): $0.0139 / step
- [FLUX 2 Trainer](https://doitong.com/developers/models/flux-2-trainer.md) (`flux-2-trainer`): $0.0064 / step
- [FLUX 2 Trainer Edit](https://doitong.com/developers/models/flux-2-trainer/edit.md) (`flux-2-trainer/edit`): $0.0389 / step
- [FLUX 2 Trainer V2](https://doitong.com/developers/models/flux-2-trainer-v2.md) (`flux-2-trainer-v2`): $0.0064 / step
- [FLUX 2 Trainer V2 Edit](https://doitong.com/developers/models/flux-2-trainer-v2/edit.md) (`flux-2-trainer-v2/edit`): $0.0389 / step
- [Flux Kontext Trainer](https://doitong.com/developers/models/flux-kontext-trainer.md) (`flux-kontext-trainer`): from $1.25 / run

## Links

- HTML page: https://doitong.com/developers/models/stable-audio-3-trainer
- All models and guides for AI agents: https://doitong.com/developers/llms.txt
- OpenAPI 3.1 with this model's typed request body: https://doitong.com/developers/openapi.json?model=stable-audio-3-trainer
- Errors: https://doitong.com/developers/docs/errors
- Model details as JSON: https://api.doitong.com/v1/models/stable-audio-3-trainer