Ace Step 1.5 XL Turbo and SFT - TEXT to AUDIO model with Ollama — v1.6 AceStep 1.5XL

Use this ComfyUI workflow to make up to four Ace Step 1.5 or 1.5 XL songs from a song or artist reference, another description, or tags and lyrics you provide.

Ollama can generate tags and lyrics from a song, artist, or description; you can also enter your own tags and lyrics.

Choose Turbo, SFT, Turbo XL, or SFT XL tracks, and switch each option on or off.

This version includes a step to save audio as MP3.

To use Ollama for tags, lyrics, or song titles, install Ollama, download a model, select it in Ollama Connectivity, then reconnect.

Place diffusion models, text encoders, and the Ace Step VAE in their ComfyUI model folders: diffusion_models, text_encoders, and vae.

The workflow lists these model files: ace_1.5_vae.safetensors, acestep_v1.5_turbo.safetensors, qwen_0.6b_ace15.safetensors, and qwen_1.7b_ace15.safetensors.

It lists comfyui-easy-use 1.3.5, comfyui-ollama 2.1.0, and rgthree-comfy 1.0.2512112053.

You can randomize key, BPM, and song duration, and use dynamic prompts.

For a LoRA, enter its trigger word in the Pretext input; the workflow description says a note includes around 25 trigger words.

The author suggests starting with one Turbo or Turbo XL model if you want to try it before downloading more models.

Suggestion · not verified

The title identifies v1.6, but the description also includes v1.8 notes; check which version your download contains before relying on version-specific features.

Suggestion · not verified