With just 15 seconds of any voice, Fish Speech can reliably synthesize natural and fluent speech while maintaining the given timbre, style, and accent. Our open-source team, creators of So-VITS-SVC and Bert-VITS2, proudly introduces Fish Speech.