# ACE-Step 1.5 | Text to Music

ACE-Step 1.5 is a diffusion and language model–based text-to-music system that generates music with vocals from natural-language prompts and optional custom lyrics. It supports Chain-of-Thought reasoning for higher quality, multi-output batches, multilingual vocals, and automatic detection of BPM, musical key, and time signature. Use markers like [verse], [chorus], [bridge] or [inst] and [instrumental] to structure songs. Outputs FLAC audio with user-defined duration and is billed per output second, with thinking mode charged at double rate.

## API Information

- **Model Slug:** ace-step-1-5-text-to-music
- **Branded URL:** https://www.eachlabs.ai/each-labs/eachlabs/ace-step-1-5-text-to-music
- **Provider:** each::labs
- **Category:** Music Generation
- **Output Type:** audio
- **Status:** active
- **Version:** 0.0.1
- **Base Cost:** Thinking-enabled (default): $0.0006 per output second (2x rate), multiplied by num_outputs
- **Estimated Processing Time:** 55 seconds
- **Last Updated:** 2026-07-13
- **Interactive Demo:** https://www.eachlabs.ai/ai-models/ace-step-1-5-text-to-music

## Pricing

- **Charge Type:** dynamic
- **Estimated Price (default example):** $0.0360
- **Pricing Details:** Thinking-enabled (default): $0.0006 per output second (2x rate), multiplied by num_outputs

### Pricing Rules

| Condition | Pricing |
| --- | --- |
| thinking == "false" | Non-thinking mode: $0.0003 per output second, multiplied by num_outputs |
| Default (fallback) | Thinking-enabled (default): $0.0006 per output second (2x rate), multiplied by num_outputs |

## Input Schema

| Parameter | Type | Required | Default | Constraints | Description |
|-----------|------|----------|---------|-------------|-------------|
| prompt | string | No | - | - | Music description or caption (style, instruments, mood). Max 512 chars. Drives the musical identity. |
| lyrics | string | No | - | - | Lyrics text. Use [verse], [chorus], [bridge] for structure, or [inst]/[instrumental] for a vocals-free track. Max 4096 chars. Multilingual. |
| duration | integer | No | 30 | - | Length of generated audio in seconds. Default 30. Billed per output second. |
| num_outputs | integer | No | 1 | - | How many distinct audio samples to generate per request. Default 1. Each additional output is charged at the full rate. |
| thinking | boolean | No | true | - | Enable the 5Hz LM Chain-of-Thought stage for higher quality. Default true. WARNING: 2x cost per output second. |
| vocal_language | string | No | unknown | - | Language code for vocals (en, zh, ja, ko, es, fr, de, etc.). Use "unknown" to auto-detect from lyrics. |
| bpm | integer | No | - | - | Beats per minute. Leave empty for auto-detect. |
| key_scale | string | No | - | - | Musical key and mode, e.g. "C Major", "Am", "F# minor". Empty for auto-detect. |
| time_signature | string | No | - | [{"label":"Auto","value":""},{"label":"2/4","value":"2"},{"label":"3/4","value":"3"},{"label":"4/4","value":"4"},{"label":"6/8","value":"6"}] | Rhythmic grouping. Empty = auto-detect. |
| infer_method | string | No | ode | [{"label":"ODE (Euler, deterministic)","value":"ode"},{"label":"SDE (stochastic, varied)","value":"sde"}] | Diffusion solver. ode (default) is deterministic; sde is stochastic. |
| num_inference_steps | integer | No | 8 | - | Number of denoising steps. Default 8 (turbo). Base: 32-64. |
| guidance_scale | number | No | 7 | - | Classifier-free guidance strength (base models only, ignored on turbo). Default 7. |
| shift | number | No | 3 | - | Timestep shift factor. Default 3.0 (recommended for turbo). |
| lm_temperature | number | No | 0.85 | - | LM sampling temperature when thinking is enabled. Default 0.85. |
| lm_cfg_scale | number | No | 1 | - | LM classifier-free guidance scale. Default 1. WARNING: >1.0 triggers ~10x slower step-by-step CFG. |
| lm_negative_prompt | string | No | NO USER INPUT | - | Negative prompt for LM stage. |
| use_constrained_decoding | boolean | No | true | - | Enable constrained decoding for reliable metadata parsing. Default true. |
| seed | integer | No | - | - | Random seed for reproducible results. Leave empty for random. |

## Example Request

```bash
curl -X POST https://api.eachlabs.ai/v1/prediction/ \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "ace-step-1-5-text-to-music",
  "input": {}
}'
```

## Output Schema

Response returned by `GET /v1/prediction/{id}` when the job completes:

```json
{
  "status": "success",
  "predictionID": "string",
  "output": "string (URL of generated audio)",
  "metrics": {
    "predict_time": "number (seconds)"
  }
}
```

## Polling

```bash
curl https://api.eachlabs.ai/v1/prediction/{PREDICTION_ID} \
  -H "Authorization: Bearer YOUR_API_KEY"
```

| Status | Meaning |
|--------|---------|
| `processing` | Still running — poll again |
| `success` | Done — read `output` |
| `error` | Failed — read `message` / `details` |

## Webhook (alternative to polling)

Pass `"webhook_url": "https://your.host/path"` in the create request. Eachlabs POSTs this payload when the job ends:

```json
{
  "exec_id": "prediction-uuid",
  "status": "succeeded",
  "output": "https://...",
  "error": ""
}
```

`status` is `"succeeded"` or `"failed"`. `exec_id` equals the `predictionID` from create. Return 2xx within 30 seconds.

## Errors

Error body: `{ "status": "error", "message": "...", "details": "..." }`

| Code | Meaning |
|------|---------|
| `400` | Invalid input |
| `401` | Missing / invalid `Authorization` bearer token |
| `404` | Unknown model or prediction id |
| `429` | Rate limit — 100 creates / min, 10 concurrent per key |
| `5xx` | Retry with backoff |

## Overview

**ACE-Step 1.5 | Text to Music Overview**

ACE-Step 1.5 | Text to Music revolutionizes music creation by transforming natural-language prompts into full songs with vocals, complete with custom lyrics and structured sections. Hosted on each::labs, part of the eachlabs family, this diffusion and language model-based system stands out with its Chain-of-Thought reasoning, enabling higher-quality outputs through step-by-step musical composition logic. Users can generate professional-grade tracks in FLAC format, supporting multilingual vocals and automatic detection of BPM, key, and time signature.

Ideal for creators seeking instant music without instruments or studios, ACE-Step 1.5 | Text to Music handles everything from verses to choruses using simple markers like [verse] or [chorus]. It offers multi-output batches for efficiency and user-defined durations, billed per output second—making it a cost-effective choice for each::labs music-generation workflows. Whether prototyping ideas or producing final tracks, this model delivers coherent, structured music from text alone.

## Usage Notes

- API Base URL: `https://api.eachlabs.ai/v1`
- Authentication: send `Authorization: Bearer YOUR_API_KEY`. Generate a key from the Eachlabs dashboard at https://www.eachlabs.ai/dashboard/api-keys.
- File-typed parameters (`*_url`, `image_url`, `video_url`, `audio_url`, etc.) accept publicly-reachable HTTPS URLs only. Upload your asset first (GCS / S3 / your CDN) and pass the resulting URL. Data-URIs and localhost URLs are rejected.
- For structured parameters (arrays / objects) send real JSON values, not stringified payloads.
- Monetary values are reported in USD; per-token / per-megapixel rates may be billed in micro-cents internally.
- Prefer `webhook_url` over polling for long-running predictions — see the Webhook Callback section.