Style reference: generate new songs based on a previous song's soundscape

The request:

Add the ability to point Songer at a previously generated song and say "create a new song with this same style and soundscape." The new generation would inherit the sonic character, production feel, instrumentation balance, and overall vibe of the reference track, while allowing new lyrics, themes, or structural changes.

The problem it solves:

Creators often land on a sound they love after multiple generations and prompt iterations. Reproducing that exact sonic identity later is difficult, even with the original prompt saved. Small variations in how prompts are interpreted across generations mean that "the sound you found" can be hard to return to consistently.

A style-reference feature would let creators:

• Build a cohesive body of work with a consistent sonic identity (useful for EPs, albums, artist projects)

• Iterate on lyrics or themes without losing the production character that made the original track work

• Use a successful generation as a creative anchor rather than relying entirely on prompt engineering to rediscover it

• Develop a recognizable "sound" across multiple tracks, the way human artists do

How it might work (open to product team):

• A "use as style reference" option on any previously generated song

• The reference contributes sonic and production characteristics, while the user supplies new lyrics, theme, or structure

• Optionally, a slider or setting for how closely to match the reference versus allowing creative variation

Why this matters:

Style consistency is core to how musicians build identity and how listeners recognize an artist. Right now, prompt-based generation makes every song a fresh roll of the dice in terms of sonic character. A style-reference workflow would give creators a way to develop and maintain a signature sound, which is one of the most-requested capabilities in AI music creation more broadly.