Suno’s upgraded AI music generator is technically impressive, but still soulless
SMRTR summary
Fighting off record label lawsuits hasn't stopped Suno from releasing v5, its latest AI music creation tool. The upgrade brings cleaner audio and sharper instrument separation, but it's what the system can't do that reveals AI music's fundamental limitations.
During testing, even explicit instructions to create raw, imperfect vocals were completely ignored. Ask for "unprocessed emotional solo vocals with no reverb, no harmonies," and Suno still delivers perfectly pitch-corrected performances drowning in effects and backup singers.
The model struggles with musical authenticity in telling ways. Every rock vocal ends up sounding like Imagine Dragons, every R&B track resembles a sleepwalking Adele. When fed lyrics nearly identical to the Rolling Stones' "Gimme Shelter," Suno produced something with all the technical elements but none of the emotional devastation of Mary Clayton's voice-cracking performance in the original.
Product manager Henry Phipps acknowledges the vocals lack human imperfections, explaining that "models don't yet understand descriptions of specific effects and recording techniques." It's precisely those flaws that carry emotional weight in real performances, something no amount of technical polish can replicate.
SMRTR provides this summary for quick context. The original article belongs to The Verge.
Read the original article