Sonus
Turn any idea into a complete original song with AI
7 followers
Turn any idea into a complete original song with AI
7 followers
Sonus turns a text prompt—or even an image—into an original, ready-to-share song in seconds. It handles composition, lyrics, vocals, and arrangement, so no music training is required. Unlike one-shot generators, Sonus supports the full creative workflow: complete songs, high-fidelity instrumentals, lyric assistance, and stem separation for remixing. It’s made for creators producing videos, games, podcasts, ads, or fast demos.
Would love to see a way to lock in a consistent vocalist across multiple generations, so a whole playlist or video series can feel like one artist. Right now every track sounds like a different singer, which makes it tricky for branded content where you want a recognizable voice tying everything together.
@tanerklnamvsz6 That’s really helpful feedback, Taner. Consistent vocalist identity across a playlist or branded video series would make the output feel much more cohesive. I’ve noted this for our product planning. Would reusable vocalist profiles, or the ability to preserve the voice from a previous generation, fit your workflow better?
Finally tried Sonus with a rough still from a short film I shot and it actually gave me a moody track that fit. The stem separation is the real surprise, saved me from digging through a session to clear a vocal.
@berk9j38 Thanks so much, Berk! It’s great to hear the image-to-song flow matched your short film’s mood. We built stem separation for exactly that handoff from inspiration to an editable session. Which stems or export format would make your video workflow even smoother?
Love that stem separation is built right in instead of bolted on, makes it feel like a real tool for people who actually want to finish a track rather than just play with a prompt.
@adabirvar Thanks, Ada! That was a core design goal: generation should be the start, not the end, of the workflow. We want creators to be able to shape and finish a track, not just listen to a one-off result. What other finishing controls would be most useful to you?