跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
设计与 UI

claude-music-lora

LoRA and LoKr fine-tuning for ACE-Step 1.5. Trains custom styles from 3-10 songs, manages trained adapters, and applies them during generation. Uses ACE-Step's built-in training pipeline.

AgriciDaniel · 已有静态检查记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add AgriciDaniel/claude-music --skill claude-music-lora

技能说明


name: claude-music-lora description: > LoRA and LoKr fine-tuning for ACE-Step 1.5. Trains custom styles from 3-10 songs, manages trained adapters, and applies them during generation. Uses ACE-Step's built-in training pipeline. when_to_use: > Use when the user asks to train a LoRA, fine-tune, clone a voice, make a custom style, or use LoKr for personalized music models. allowed-tools:

  • Bash
  • Read
  • Write

claude-music-lora — LoRA Fine-Tuning

Overview

Train custom LoRA adapters to capture specific vocal styles, genres, instrument sounds, or production aesthetics. Requires 3-10 songs as training data.

Dataset Preparation

  1. Collect 3-10 songs in the target style (WAV/FLAC preferred, MP3 OK)
  2. Place in a directory: ~/Music/lora-datasets/<style_name>/
  3. Songs should be 30-300 seconds each
  4. Consistent style/genre across the dataset
  5. High audio quality (no noise, no clipping)

Training

cd "$(python3 -c "import json; print(json.load(open('$HOME/.claude/skills/claude-music/config.json'))['ace_step_dir'])")"

# LoRA training (standard, ~1 hour on RTX 5070 Ti)
uv run python3 -m acestep.training.train_lora \
  --checkpoint-dir ./checkpoints \
  --model-variant turbo \
  --dataset-dir ~/Music/lora-datasets/my_style/ \
  --output-dir ./lora_output/my_style \
  --rank 16 \
  --learning-rate 1e-4 \
  --steps 1000

# LoKr training (5x faster, ~12 min)
uv run python3 -m acestep.training.train_lora \
  --checkpoint-dir ./checkpoints \
  --model-variant turbo \
  --dataset-dir ~/Music/lora-datasets/my_style/ \
  --output-dir ./lora_output/my_style \
  --method lokr \
  --rank 16 \
  --learning-rate 1e-4 \
  --steps 500

LoRA vs LoKr

AspectLoRALoKr
Training time~1 hour~12 min
QualityHigher fidelityGood, slightly less detailed
VRAM~10GB~8GB
Use caseVoice cloning, precise styleGenre adaptation, quick experiments

Hyperparameters

ParameterDefaultRangeNotes
Rank (r)164-64Higher = more capacity, more VRAM
Learning rate1e-41e-5 to 5e-4Lower for voice cloning
Steps1000200-5000More data = more steps needed
Batch size11-4Limited by VRAM

Using Trained LoRA

After training, the LoRA is available in ACE-Step's generation pipeline. Refer to ACE-Step documentation at: <ace_step_dir>/docs/en/LoRA_Training_Tutorial.md (see config.json for path)

For detailed reference: load references/lora-training.md