Most AI music sites teach the wrong first habit: they train you to return to the marketing root when you want last week’s track. The root is a sales document. The place that is built to hold generated songs, covers, and videos is a different route. This is a short information-architecture note on that split, based on a live fetch of WAV AI and its media library page.
Route / is a brochure with a clear contract
The document title on the root is Free AI Music Generator & Music Tools. The visible h1 is Free AI Music Generator. The supporting line is the pitch: turn ideas into music instantly; create songs, beats, and tracks with Text to Music AI; no music skills needed. The primary call to action is Start free.
Below that, the page is still a brochure. Sample styles (Lo-Fi, EDM, Hip-Hop, Cinematic) live under Hear the Magic. Four feature blocks under Everything You Need to Make Perfect Audio describe a generator, an extender, a remixer, and an editor, each with a try-it prompt. Audience sections, testimonials, and a FAQ follow. If you are routing first-time visitors, this is the right surface. If you are routing someone who already generated a track and wants it back, this page has no collection to offer them.
Route /media is an Assets surface
In the site navigation, Media does not sit under Create (Music, Lyrics, Editor) or under Tools. It sits under Assets. That grouping is the IA clue. The Media Library page uses a different document title from the marketing root: Media Library | WAV AI. The meta description says to manage your generated songs, covers, and videos in the WAV AI media library.
The visible chrome on that page is collection chrome: All Creations, Upload, and Folders. A public fetch without a signed-in library still returns HTTP 200 and renders the page. It is not a login wall. The body shows an empty state, No assets found matching your criteria. That copy only makes sense on a filtered collection. It would be a strange sentence to put on a marketing homepage.
One thing to watch if you log titles for analytics: the Open Graph title on /media still reused the generic marketing title in this fetch, while the HTML <title> correctly said Media Library. If events or social cards key off og:title, the two surfaces can look identical in a report even though the routes and on-page contracts are different. Prefer the document title and the visible controls, not the social title, when you decide what the page is.
Recognition is the library’s job; recall is what the homepage leaves you with
Raluca Budiu’s Memory Recognition and Recall in User Interfaces (Nielsen Norman Group, 15 January 2024) is useful here as interface guidance, not as a score. The article distinguishes recall (retrieving a name or detail from scratch) from recognition (seeing a familiar item in a list). Recognition is easier because the extra cues raise the chance of retrieval. Histories, recently viewed items, and saved lists are the examples it gives of interfaces that let people resume work they have already seen.
A marketing root asks you to recall that a track exists and then start over from a prompt. A media library, if it shows the work, turns the same task into recognition: you scan songs, covers, and videos until the familiar one appears. That is the routing reason to keep / and /media apart.
The Library of Congress page Keeping Personal Digital Audio adds a practical, non-UX angle. Its Personal Digital Archiving tips are to identify audio files, decide which recordings have long-term value, organize them with descriptive names, tags, and folders, and keep metadata such as an export date. I am not treating the product library as a preservation archive, and I am not claiming it performs those backups. The transferable idea is smaller: generated audio is a collection, and collections need a place with folders and a way to find a file again.
A compact routing checklist
- First visit, or a visitor who still needs the pitch →
/. - Someone asking “where did that track / cover / video go?” →
/media, under Assets, with All Creations, Upload, and Folders on the page. - Do not collapse the two routes because their Open Graph titles look alike.
- Do not describe
/mediaas login-only; the public page loads, and an empty state is not the same thing as an auth wall.
Keep the brochure at the root. Put the library on its own path. The people who come back for work they already made should not have to recall a prompt from a sales page.
