Ace-Step 1.5 XL SFT LoRA - Multi-Genre Reggaeton / Afrobeat
This LoRA was trained for Ace-Step 1.5 XL SFT and is designed to enhance generation of urban Latin music styles, especially Reggaeton and Afrobeat / Afro-pop inspired sounds.
The same LoRA can be used with different genres by changing the first keyword in the Style prompt.
Recommended usage
At the beginning of the Style prompt, specify the desired musical direction:
For urban reggaeton:
Reggaeton, [your instructions here]
For afrobeat / afro-pop inspired generations:
Afrobeat, [your instructions here]
After selecting the genre, add your own instructions describing the desired vocals, instrumentation, mood, structure, or production style.
Example:
Reggaeton, energetic male vocals, classic dembow drums, deep sub bass, catchy melodic hooks, commercial production
or:
Afrobeat, melodic female vocals, tropical groove, layered harmonies, emotional hooks, modern pop production
Training approach
This LoRA was trained using a small curated dataset with structured musical metadata, lyrics, and song sections (Intro, Verse, Pre-Chorus, Chorus, etc.).
The dataset focuses on improving:
- Melodic coherence
- Vocal phrasing
- Harmonic variation
- Catchy hooks
- Urban Latin and Afro-pop textures
The model performs best when the Style prompt clearly specifies the intended genre and musical characteristics.
Base model requirement
This LoRA was trained specifically for:
Ace-Step 1.5 XL SFT
Make sure you load it with the correct base model for best results.
- Downloads last month
- -