Rosebud AI goes from silent games to audio by default with Eleven Music API

Rosebud AI is a platform for vibe-coding video games, with more than two million games created and thousands of active creators building daily. The team integrated Eleven Music and Sound Effects APIs so that every game generated on Rosebud ships with audio by default.

Audio as a default, not a power-user feature

Before the ElevenLabs integration, Rosebud had no native audio generation. A small number of pre-built templates included handcrafted ambient loops added by the Rosebud team, but most creators had to source their own music and sound effects from external platforms and upload them manually. Creators had to untangle licensing for every track, check usage rights, remove watermarks from previews, and confirm attribution requirements before a track was usable in a game. In practice, audio was used almost exclusively by experienced developers building complex projects. Nearly all games shipped silently.

After integrating ElevenAPIs, Rosebud added audio generation to its oneshot creation flow, the single-prompt experience where a creator describes a game and the platform builds it end to end. The platform now automatically produces a soundtrack and sound effects alongside the code and visuals of the new game. Creators who want more control can go into the assets tab and generate additional audio with their own prompts, adjusting the genre, style, mood, and length before adding tracks to the game. They can also change the sound direction entirely through the chat interface, prompting the platform to regenerate all audio assets without rebuilding the game.

Polish that used to require leaving Rosebud - licensing tracks or making music on a separate platform - is now built in. The quality of games just skyrocketed. We were always aiming for that one-click magic, and now every oneshot comes out sounding like a finished game.

— Andi Bucescu, Head of Marketing, Rosebud AI

The first generation sets the tone

Midnight Rider, a polished racing game on Rosebud's featured page, took its creator more than 300 prompts to build, but the first one defined its sound. Along with the score and a set of sound effects, the initial oneshot generated a sound direction, a creative brief that every later audio generation follows automatically. When the creator returned hundreds of prompts later to ask for start and pause sounds or layer in a passing car's "whoosh," each new effect arrived already matching the world built before it. The result is one coherent audio identity, with no manual audio work required.

1,500 games shipped with audio every day

The Rosebud team saw the impact of incorporating music and sound effects almost immediately. Average creator engagement (time spent building) rose 23% after launch, and the most active creators pulled even further ahead, with P90 engagement up 28%. Once audio generation was active, the quality of submitted games noticeably increased. Games that would have felt like rough prototypes started feeling finished and shareable, and the best of them now appear on Rosebud's featured pages with thousands of players.

The integration runs at that quality across the full platform. The audio pass adds only about four seconds to generation, with music and sound effects produced concurrently and no errors logged. At scale that adds up to roughly 1,500 games with audio a day and more than 150,000 audio files, with nearly 300 hours of music and sound effects since launch. Fewer than 1% of creators go back to manually change the default output, with many building on it by prompting additional music and effects as their games grow.

That momentum is starting to reach players too. Daily active users are up 8% since launch, as more audio-polished games get published and discovered on Rosebud.

More finished games, by default

ElevenLabs turns every Rosebud oneshot into a polished, production-ready experience that players do not just play, but hear.