
A game over sound is the reverse of a power-up: descending, in a minor key, and slower than you might expect.
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.
dark orchestral-hybrid game over theme, slow dirge tempo, mournful choir vocals, deep strings and low brass, minor key, cinematic
comedic sad trombone wah wah wah, four descending muted brass notes, dry
game over fade to black, warm sustained chord slowly detuning downward into silence, roomy
8-bit death sound, fast descending square wave pitch spiral ending in a low noise thud, dry
two slow heartbeats then a long flat electronic monitor tone, clinical and close
failure sting, glass shattering into a low descending piano note and a dark reverb tail
orchestral defeat sting, low strings falling in a minor cadence under a struck timpani, hall reverb
retry prompt tone, neutral two-note synthetic chime, unresolved interval, dry
machine powering down, whirring fan and drive slowing to a stop with a descending pitch, close mic
small crowd groaning in disappointment, murmured aww falling away, medium room
VHS tape mechanism whirring and ejecting with a mechanical clunk, close mic, dry
These models respond to physical description. Name the material, the action and the space.
chiptune game over jingle, six-note descending minor line, square wave, slow and final, 1.2 seconds
"Slow" and "final" together stretch the note spacing near the end. A game over jingle has to sound like it is stopping as well as falling in pitch.
player death stinger, short descending synth sweep with a low thud landing, 700 ms, dry and abrupt
"Landing" on a low thud gives the sweep a definite endpoint.
orchestral-hybrid defeat fanfare, descending minor brass-synth line with a dissonant final chord, 1.5 seconds, light reverb
"Dissonant final chord" separates a defeat fanfare from a merely sad one. Unresolved dissonance reads as loss.
flatline defeat tone, single sustained low sine wave note fading out over 1 second, no vibrato, dry
"No vibrato" matters because any pitch wobble on a sustained tone sounds alive. A flatline has to stay flat.
boss defeated fanfare, dramatic descending-then-resolving synth line, 2 seconds, triumphant despite descent, moderate reverb
"Resolving" after the descent is the key word. The enemy went down but the player won, so this one should end on a stable chord.
dark orchestral-hybrid game over theme, slow dirge tempo, mournful choir vocals, deep strings and low brass, minor key, cinematic
This one uses ACE-Step, a music model. The prompt sets the genre and instrumentation, and a separate lyrics field ("[verse]\nGame over\nYou have fallen") carries the words a synthetic choir sings. A game over screen is one of the few places where a sung line sounds deliberate.
Descending pitch is the usual signal for game over, defeat and death. Ascending motion, even inside an otherwise falling line, reads as a win. For the player's loss (their own death, a timer running out), prompt for a clean descent that ends low. For an enemy's defeat, prompt for a descent that resolves upward or lands on a stable major chord.
Words like "slow", "final" and "landing" widen the spacing between the last notes and give the sound a definite stopping point, so it doesn't trail off. A game over sound that doesn't clearly end feels unresolved.
A time's-up buzzer should be blunt and harsh: a flat, static pulse with no melodic content. A story-beat death wants a melodic descent and maybe dissonance, because it should land emotionally. Use different prompt shapes for the two, and decide which one the moment is first.
Every sample above except one comes from a sound-effect model: synthesized tones and swept pitches, no voice. ACE-Step is a lyrics-conditioned music model. It takes a genre and instrumentation prompt plus a separate block of lyrics, and sings them. On a game over screen, a jingle implies the loss and a choir states it. Use the choir for a story beat such as a final boss run or a permadeath. For the twentieth retry on a platformer level, a half-second stinger is the right size.
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}'