Musicians: how reliable do you think your ear is at telling AI music from human-made music? [Built a 5-min test to find out]
Genuine question for yall musicians in here.
I've built a small anonymous listening test, 10 short clips, some AI-generated (Suno v5), some human-made (by independent musician without generative AI aid), rated on a 4-point scale. ~5 min, no signup.
What I've found running it so far:
• People who use AI music tools daily average ~67-70% accuracy
• General audience hovers around chance (~43%)
• Self-rated confidence does predict performance (r ≈ 0.38) but moderately: many "confident" listeners still score poorly
My hypothesis is that trained musicians (regardless of AI exposure) might land somewhere in between, sharper ear for production quirks but no specific Suno-pattern recognition.
Would love you to test that!
Try it out! https://parlor.shinyapps.io/audio-short-en/?src=reddit_askmusicians
Genuinely curious what cues you'd use to tell them apart, and whether being a musician helps or hurts.
The selected clips will be used in a follow-up academic research study on how streaming platforms should label AI content, not on training detection models or improving Suno or whatever other model. Skepticism is fair, but the outcome here is consumer protection.