Suno adds speech generation as questions over training persist

A feature with an unanswered question

The announcement describes what the tool produces, not how it learned to produce it. That distinction is hard to ignore for a company whose training practices are already under legal scrutiny.

Major record labels have sued Suno. More recently, a Munich court ruled against the startup, rejecting fair use as a justification for using copyrighted data.

I think the new speech feature makes the training question more pressing, not less. A capability that combines voice and music broadens what Suno can generate, while the company has offered no account of the data behind its model. The product is expanding faster than the public explanation of how it was built.