I find the overall audio output to sound very electronic and somewhat thin, resembling AM radio rather than the fuller FM sound I expected. I was unable to get realistic guitar tones.

The chord progressions in the songs are generally solid, but the melodies sometimes feel immature or out of place, which affects the overall flow.

The vocals stand out as a strong point and are as good if not better than any other AI music services I’ve used.
I love the ability to choose key and BPM, something I haven’t noticed elsewhere.
Song generation is very fast, and the number of credits are generous, which is great for quick experimentation.

The support for LoRA support for genre-specific models could significantly improve instrument realism and melody sophistication, which could eliminate these issues that are important to me. The addition of more generation presets and smarter AI tagging also sounds promising to help produce better-fitting melodies and sounds.

If the roadmap features are delivered effectively, this service could easily be a valuable alternative to the others.

When is the expected timeline for LoRA support implementation?