ace-step/ACE-Step-1.5
Grade music generation under 4GB VRAM.
Access: Free or open-source project
Setup: self-host
Functional replacements with a real free or open-source path, described in plain terms.
Grade music generation under 4GB VRAM.
Access: Free or open-source project
Setup: self-host
Full-song generation foundation model.
Access: Free or open-source project
Setup: self-host
Meta's research library for music and audio generation, including MusicGen. It is a local model toolkit rather than a hosted Suno-style service, so hardware and model licenses matter.
Setup: local Python environment
Open music model family for generating songs from lyrics and tags, with multilingual support. Inference requires model weights and local compute; licensing should be checked for each weight release.
Setup: self-host
Open-source latent-diffusion model for generating full songs from lyrics and style prompts. Requires model setup and local compute; outputs and commercial rights depend on the published terms.
Setup: self-host
Open-source toolkit for training and inference with generative audio models. Run audio generation locally with the required weights and compute instead of relying on a hosted music subscription.
Setup: local Python environment