An 8-bit console has four voices to write with: two pulse/square waves, one triangle and one noise channel. Every NES or Game Boy soundtrack was built from that set. Name the four voices by role (lead, harmony, bass, percussion) and the model returns a loop that can sit under a title screen or an overworld map. Asking for "8-bit music" alone tends to return a generic run of blips.
Every clip shows the exact prompt that produced it. The render script reads its prompts from the same file as this page. Copy one, change a word and generate your own.
Game Boy APU chiptune overworld theme, cheerful square wave lead synth melody, pulse wave harmony synth, triangle wave bass synth, simple noise-channel percussion, bright major key, 125 BPM, seamless loop
NES-style 8-bit title screen theme, bold square wave lead synth fanfare opening into a looping pulse wave melody, triangle wave bass synth, noise-channel snare hits, heroic major key, seamless loop
8-bit shop menu background loop, gentle square wave lead synth melody, soft triangle wave bass synth, sparse noise-channel percussion, relaxed mid-tempo major key, seamless loop
8-bit final boss battle theme, fast arpeggiated pulse wave lead synth, driving triangle wave bass synth, aggressive noise-channel drum pattern, dark minor key, 150 BPM, seamless loop
Game Boy APU chiptune victory loop, cheerful square wave lead synth melody, pulse wave harmony synth, triangle wave bass synth, bright major key, seamless loop
These models respond to physical description. Name the material, the action and the space.
Game Boy APU overworld theme, cheerful square wave lead synth, pulse wave harmony synth, triangle wave bass synth, simple noise-channel percussion, bright major key
Name all four voices by their job (lead, harmony, bass, percussion). Otherwise the arrangement collapses into one melody line with no supporting parts.
NES-style title screen theme, bold square wave lead synth fanfare opening into a looping pulse wave melody, triangle wave bass synth, noise-channel snare hits
A title theme opens with a fanfare and then loops. Say so in the prompt to get that intro-into-loop structure.
8-bit shop menu loop, gentle square wave lead synth, soft triangle wave bass synth, sparse noise-channel percussion, relaxed mid-tempo
Menu music has to sit quietly under UI sounds. The model fills every voice unless told to hold back, so say "sparse" and "relaxed".
8-bit final boss theme, fast arpeggiated pulse wave lead synth, driving triangle wave bass synth, aggressive noise-channel drum pattern, dark minor key, 150 BPM
Each square-wave channel plays one note at a time, so NES composers used fast arpeggios to imply chords. "Arpeggiated" asks for that.
8-bit hardware had a limit of four simultaneous voices, and that limit is what makes the genre recognisable. Ask for a lead melody, a harmony or counter-melody, a bass line and percussion by name, and the model renders something that reads as a composition. Ask for "8-bit music" alone and you tend to get one busy lead line with little underneath.
Stable Audio 3 SFX is built for one-shot sound design: coins, jumps, hits. A 15-second background loop needs melody, harmony, bass and drums that stay in key and in time for the full length. That is a composition task, and a music model (Stable Audio 3 Small Music or ACE-Step) holds it together better. Use the SFX model for the coin blip and a music model for the music under it.
State a BPM. Overworld and shop themes sound calmer around 100 to 125 BPM; boss and combat themes want 140 to 160. The model follows a stated tempo more reliably than a mood word like "energetic". A consistent BPM across a set also helps several tracks sound like the same game.
The victory loop above uses nearly the same instrumental brief as the overworld loop (same voices, same major key, same "seamless loop" instruction) but renders it on ACE-Step instead of Stable Audio 3 Small SFX. ACE-Step is mainly a lyrics-conditioned vocal model (the chiptune-music collection has a sung example). With `lyrics` left empty it renders a usable instrumental, which makes it a second option for loops like these.
The models run on your machine, which changes the cost, the privacy and what you can automate.
Weights are fetched once. After that, generation, the library and the API all work with the network off. Everything lives in one SQLite file and one folder of WAVs under ~/.soundbox, and nothing is uploaded.
Soundbox runs an MCP server in the same process, with tools for generating and listing sounds. Generated files come back as absolute paths, so an agent can copy them into a project.
Every model's terms are shown before you download it, and non-commercial weights need an explicit confirmation. Soundbox is MIT. The licence of the model that made a sound governs what you may do with it.
claude mcp add --transport http soundbox http://127.0.0.1:17494/mcpcurl -X POST http://127.0.0.1:17494/api/generate \
-H 'Content-Type: application/json' \
-d '{"model":"audiocpp-stable-audio-3-sfx","prompt":"8-bit coin pickup blip","duration_s":1}'