A Vinyl Bar in Shibuya has raised $5.5 million in a pre-seed round to build small music apps that contain no generative AI and accept no prompts. The AI boom has produced a run of startups offering to write you a song in a chosen artist's style; this one, founded by Spotify's former head of innovation Mauhan M. Zonoozy, sells the opposite — a Chrome extension that warps the speed of whatever is playing in your browser tab, a site that turns a username into a melody, a stem player laid out on a 3x3 grid. Ten investors and Spotify's former content and advertising chief put money behind it.
The round drew Mantis VC, SV Angel, Boxgroup, Quiet Capital, ERA, Common Metal, Gold House, Liquid 2 Ventures, Breakers VC and Systemic Ventures, along with Dawn Ostroff, who ran Spotify's content and advertising business and has joined the company as an adviser.
Zonoozy led innovation at Spotify and left in 2024 to build something of his own again. Before Spotify he founded Bubbl, a startup for clipping and remixing video, which Cricket Media later acquired.
Source: techcrunch.com
His argument is that streaming solved access and stopped there. Reaching any song became trivial, while what people do with a song stayed consumption. Watching how younger listeners behave, he noticed they have grown up meeting culture through games, filters and creative tools, and saw room to offer music on the same terms: something to make, change and pick apart rather than only play.
The company ships its experiments as singles gathered into a single album. Each is built around handling one element of music and altering how it sounds.
Speed Surfer is a Chrome extension that changes the speed and filter of audio playing in the browser window.
Usersound is a site for building an audio username, each letter standing for a note.
Stacks is a stem player: mixes are assembled by switching bars on and off in a 3x3 grid.
Drops has the user arrange plates in a pattern to deflect falling balls, with every ball sounding a note.
Sampler turns YouTube videos into samplers built on the rhythm of different sections of a song.
Source: techcrunch.com
The larger product is bop, an iOS app. The user places instruments and effects on a grid to assemble mixes from a source song, changing the tempo and dropping in one-shot effects. Finished mixes export to TikTok or Instagram. A multiplayer version that lets people remix a song together is in development.
Source: techcrunch.com
Zonoozy says keeping AI out of the apps is a deliberate choice, while the company is willing to use AI in developing products and building reliable infrastructure. AI is already highly capable, he argues, but the future of music does not necessarily reduce to generating songs with it. If AI makes musical content more abundant and easier to reach, then taste, a human point of view, interaction and personal involvement matter more — and the company does not want those to become the scarce goods.
A music mixer-sandbox is due this year, letting people experiment with the components of making music, including constructing their own instruments.
The no-AI framing is thinner than it sounds. The company is not refusing AI; it is refusing generation at the surface the user touches, while using AI freely behind it. That is a product decision wearing the clothes of a principle. It also happens to be a good product decision — a grid you poke at holds attention in a way a text box does not — but it should be read as positioning rather than abstinence.
The more interesting question is where the rights come from. Sampler converts YouTube videos into instruments. bop remixes a source song. Speed Surfer reaches into audio already playing in the browser. Nothing in the announcement says who licensed what, and for a company built around Spotify alumni that silence is conspicuous: they of all people know the expensive part of a music product is not the interface.
Also absent is any account of how this makes money. Free browser toys and an iOS app with social export describe a distribution strategy, not a revenue one, and $5.5 million at pre-seed buys time to find one rather than evidence that one exists.
The sandbox due this year is the real test. Singles are cheap to try and cheap to abandon; a tool people use to build their own instruments is something they would have to come back to. Zonoozy is betting that once songs are abundant, the scarce thing is the person with taste making them — a wager that his users will want to do the work, at the moment the rest of the industry is selling them the chance not to.