The track 'Ty moya...' by DJ NARODNII gained significant traction, appearing in over 150,000 videos within a week. The author intended it as a parody of generic AI-generated music, though it sparked debate regarding its content and the nature of AI-produced art.
Data from Yota indicates that in 2026, traffic to music-generating AI services increased by over 20%, with time spent on these platforms rising by 45% year-over-year. As of February 2026, Suno reported over 2 million paid subscribers and an annual recurring revenue of $300 million.
Modern music AI models typically utilize audio codecs like SoundStream or diffusion-based architectures. For instance, Stability AI's Stable Audio uses a diffusion model with 907 million parameters, trained on over 800,000 audio files, capable of generating 95 seconds of stereo audio in under a second on A100 accelerators.
Suno's current v6 family includes three models: the flagship v6, the v6-wild, and the free v6-mini. These models allow users to modify specific sections of a track or individual lines of text without regenerating the entire song, and support inputs including text, audio, images, and video.
AI music models often produce a 'plastic' sound due to their training on high-quality, clean audio data, which lacks the imperfections of human performance. Users can mitigate this by including specific prompts like 'imperfect timing' or 'slight rasp' to guide the model toward a more natural sound.
