Generator hub
AI music generators, and what each one means for detection
Every generator builds audio differently, and those differences decide which measurements carry information. These pages describe how each platform works and how much a detection reading on its output is worth — including when the honest answer is “not much”.
One thing they all have in common: this tool does not perform generator attribution. It estimates whether a recording looks machine-generated, not which service produced it.
Analyse a trackPlatforms covered
- SunoFull-song generation with vocals, lyrics and a finished master.Read the detection notes →
- UdioPrompt-driven songs with extension and inpainting workflows.Read the detection notes →
- ElevenLabs MusicMusic from a lab best known for synthetic speech and voice cloning.Read the detection notes →
- Stable AudioDiffusion-based generation for instrumentals, loops and sound design.Read the detection notes →
- RiffusionSpectrogram diffusion that grew into full song generation.Read the detection notes →
- MubertGenerative production music for royalty-free background scoring.Read the detection notes →
- Seed MusicResearch-lineage end-to-end song generation.Read the detection notes →
- MiniMaxMultimodal model family with a reference-conditioned music mode.Read the detection notes →
- Mureka (Sonauto)Prompt-to-track generation with vocal control and style transfer.Read the detection notes →
Why attribution is not offered
Naming the tool behind a track requires a classifier trained on labelled output from every platform in question, kept current as each of them ships new model versions. Nobody has published such a model with credible held-out results, and we are not going to imply one exists by putting platform names on a result screen.
What these pages give you instead is context: knowing that a generator maximises loudness by default, or stitches sections together, tells you which parts of a report to weigh and which to discount.