ACE-Step 1.5 AI Lyric Song Generation
Input a music style and optional lyrics. Generate a full song with AI-created lyrics and audio.
简介
The ACE-Step 1.5 AI Lyric Song Generation workflow is a sophisticated tool designed to transform a simple music style input and optional lyrics into a complete song with AI-generated lyrics and audio. This workflow leverages the ACE-Step and Google Gemini models to synthesize both text and audio, providing a seamless experience from concept to creation.
At its core, the workflow uses nodes like SaveAudioMP3 for outputting the final audio file, and GeminiNode for processing the AI-generated content. The RegexExtract node is instrumental in parsing and refining input data, ensuring that the AI's output is coherent and aligned with the user's intent.
This workflow is particularly useful for musicians, content creators, and anyone interested in exploring AI's potential in music production.
你需要提供
- 正在读取工作流
在 ComfyUI 中运行。 此工作流无法在本页面运行。你可以在 Comfy Cloud 中打开（已预先加载所需内容），或下载后在自己的 ComfyUI 中使用。
在 Cloud 中试用查看工作流内部
节点图的只读预览。在 Comfy Cloud 中打开或下载后即可编辑节点。
Loading the graph
Open full-size workflow preview- 作者
- ComfyUI
需要自定义节点
- 正在读取工作流
通过代码使用此工作流
尽可能根据工作流图和输入自动生成代码，否则显示已保存的示例。运行代码可能产生费用，请先核对凭据和端点。
What you need
- Runs on
- Comfy Cloud
- 密钥
- COMFY_API_KEY
需要付费的 Cloud 方案和额度。
# Python 3.10+: pip install comfy-sdk==0.4.0
import os
from comfy_sdk import Comfy
api_key = os.environ.get("COMFY_API_KEY", "").strip()
if not api_key: raise RuntimeError("Set COMFY_API_KEY")
os.environ.setdefault("COMFY_BASE_URL", "https://cloud.comfy.org")
client = Comfy(api_key=api_key)
workflow = client.workflows.from_json({
"1": {
"class_type": "SaveAudioMP3",
"inputs": {
"audio": [
"2:18",
0
],
"filename_prefix": "audio/ace-step-1.5",
"quality": "V0"
},
"_meta": {
"title": "Save Audio (MP3) (DEPRECATED)"
}
},
"7": {
"class_type": "PreviewAny",
"inputs": {
"source": [
"11",
0
]
},
"_meta": {
"title": "Preview as Text"
}
},
"11": {
"class_type": "GeminiNode",
"inputs": {
"model": "gemini-3-pro-preview",
"prompt": "A blues song, to rival the great Robert Johnson.",
"seed": 1046438542938789,
"system_prompt": "**Role:** You are a professional Music Architect for **ACE-Step 1.5**. \n\n**Task:** Generate a music creation plan.\n\n**Strict Output Constraint (IMPORTANT):**\n1. Return the output as **RAW PLAIN TEXT** only.\n2. DO NOT use Markdown code blocks (no backticks ```).\n3. DO NOT use any headers, labels, or bold text.\n4. Use ONLY the separator `---` to divide the two parts.\n5. Part 1: Comma-separated style keywords.\n6. Part 2: Structured lyrics with [Tags].\n\n**My Topic/Requirement:**\n[USER INPUT]"
},
"_meta": {
"title": "Google Gemini"
}
},
"18": {
"class_type": "RegexExtract",
"inputs": {
"case_insensitive": True,
"dotall": False,
"group_index": 1,
"mode": "First Group",
"multiline": False,
"regex_pattern": "(?sm)^(.*?)^---$(.*)",
"string": [
"11",
0
]
},
"_meta": {
"title": "Extract Text"
}
},
"19": {
"class_type": "RegexExtract",
"inputs": {
"case_insensitive": True,
"dotall": False,
"group_index": 1,
"mode": "First Group",
"multiline": False,
"regex_pattern": "(?s)(?<=\\n---\\n)([\\s\\S]*)$",
"string": [
"11",
0
]
},
"_meta": {
"title": "Extract Text"
}
},
"2:104": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "acestep_v1.5_turbo.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"2:105": {
"class_type": "DualCLIPLoader",
"inputs": {
"clip_name1": "qwen_0.6b_ace15.safetensors",
"clip_name2": "qwen_4b_ace15.safetensors",
"device": "default",
"type": "ace"
},
"_meta": {
"title": "Load CLIP (Dual)"
}
},
"2:106": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "ace_1.5_vae.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"2:110": {
"class_type": "PrimitiveFloat",
"inputs": {
"value": 120
},
"_meta": {
"title": "Song Duration"
}
},
"2:18": {
"class_type": "VAEDecodeAudio",
"inputs": {
"samples": [
"2:3",
0
],
"vae": [
"2:106",
0
]
},
"_meta": {
"title": "VAE Decode Audio"
}
},
"2:3": {
"class_type": "KSampler",
"inputs": {
"cfg": 1,
"denoise": 1,
"latent_image": [
"2:98",
0
],
"model": [
"2:78",
0
],
"negative": [
"2:47",
0
],
"positive": [
"2:94",
0
],
"sampler_name": "euler",
"scheduler": "simple",
"seed": 27660246420877,
"steps": 8
},
"_meta": {
"title": "KSampler"
}
},
"2:47": {
"class_type": "ConditioningZeroOut",
"inputs": {
"conditioning": [
"2:94",
0
]
},
"_meta": {
"title": "Conditioning Zero Out"
}
},
"2:78": {
"class_type": "ModelSamplingAuraFlow",
"inputs": {
"model": [
"2:104",
0
],
"sampling": "flow",
"shift": 3
},
"_meta": {
"title": "ModelSamplingAuraFlow"
}
},
"2:94": {
"class_type": "TextEncodeAceStepAudio1.5",
"inputs": {
"bpm": 190,
"cfg_scale": 2,
"clip": [
"2:105",
0
],
"duration": [
"2:110",
0
],
"generate_audio_codes": True,
"keyscale": "E minor",
"language": "en",
"lyrics": [
"19",
0
],
"min_p": 0,
"seed": 27660246420877,
"tags": [
"18",
0
],
"temperature": 0.85,
"timesignature": "4",
"top_k": 0,
"top_p": 0.9
},
"_meta": {
"title": "TextEncodeAceStepAudio1.5"
}
},
"2:98": {
"class_type": "EmptyAceStep1.5LatentAudio",
"inputs": {
"batch_size": 1,
"seconds": [
"2:110",
0
]
},
"_meta": {
"title": "Empty Ace Step 1.5 Latent Audio"
}
}
})
# run() also passes api_key to the partner nodes in this workflow.
job = client.run(workflow, api_key=api_key)常见问答
You can input any music style you prefer. The AI is designed to adapt and generate lyrics and audio that align with your specified style.
Yes, you can input your own lyrics, and the AI will incorporate them into the song while generating the rest of the content.
The final song is saved as an MP3 file using the SaveAudioMP3 node, which is widely compatible with most audio players.
No prior AI experience is necessary. The workflow is designed to be user-friendly, with clear instructions and intuitive nodes to guide you through the process.