Cinematic sound effects
Trailer audio is built from a few basic shapes: risers, hits, drones and swells. Learn to prompt each one and you can score a whole cut.
Listen: 12 samples from the cinematic collection
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.
Full trailer buildup
via audiocpp-stable-audio-3-sfxcinematic trailer buildup, rising sub-bass synth drone, tense string section ostinato, sparse brass ensemble braam hits accenting the climb, orchestral hybrid, one final deep orchestral hit at the end
Epic trailer choir chant
via audiocpp-ace-stepepic orchestral trailer theme, dramatic choir vocals, deep strings and brass, rising tension, cinematic
Sub drop
via audiocpp-stable-audio-3-sfxcinematic sub bass drop, deep pitch falling into a long sustained rumbling low frequency bed
Metallic riser
via audiocpp-stable-audio-3-sfxmetallic cinematic riser, layered scraping metal harmonics accelerating upward into a hard cut
Whoosh to impact
via audiocpp-stable-audio-3-sfxlong airy cinematic whoosh accelerating into a huge distorted impact with a dark decaying tail
String tension
via audiocpp-stable-audio-3-sfxorchestral string section swelling on a dissonant sustained note, bows tightening, large hall
Taiko build
via audiocpp-stable-audio-3-sfxtaiko drums building from sparse hits into a fast driving pattern, huge room, epic and wide
Ticking tension
via audiocpp-stable-audio-3-sfxtense ticking clock bed with a low pulsing drone underneath, steady and claustrophobic, seamless loop
Heartbeat bed
via audiocpp-stable-audio-3-sfxslow heartbeat pulse under a dark evolving drone, cinematic tension bed, seamless loop
Reverse riser
via audiocpp-stable-audio-3-sfxreversed cinematic riser, sound sucking backwards into a sharp silence, wide stereo, dramatic
Choir braam
via audiocpp-stable-audio-3-sfxepic trailer braam with a dark male choir layered over a distorted brass blast, enormous hall
Title card boom
via audiocpp-stable-audio-3-sfxtitle card impact, deep clean boom with a bright metallic shimmer and a slow eight second decay
Prompts that work
These models respond to physical description. Name the material, the action and the space.
Trailer braam
single cinematic braam hit, deep brass-like impact with sub-bass underneath, hard attack, short decay, no reverb tail
"Short decay, no reverb tail" keeps the hit from smearing into the next shot. Trailer hits are edited on the frame, and a long tail fights the cut.
Slow build tension
cinematic tension riser, thin string-like texture climbing in pitch and volume over 2.5 seconds, no percussive hit at the end
"No percussive hit at the end" stops the model from returning a riser plus a braam when you only need the climb.
Sub-bass drone
cinematic low drone riser, sustained sub-bass tone with slow pitch climb, dark and tense, 3 seconds, subtle grit on the top end
"Subtle grit on the top end" adds enough high-frequency content to keep a sub tone audible on small speakers without turning it into a lead sound.
Reverse swell
reverse cymbal swell building into a sharp cutoff, bright metallic wash growing in volume over 1.5 seconds, ends abruptly with no tail
"Ends abruptly with no tail" makes it sound reversed. A real reverse swell has no natural decay; it stops.
Trailer boom
trailer hit boom, single deep sub-bass punch with a short metallic clang layered on top, tight low end, dry
Naming two layers (sub-bass punch and metallic clang) combines a felt low end with an audible transient, which a boom needs on small monitors.
Making cinematic sounds that fit
Separate the riser from the hit
A cinematic sting is usually two events edited together: a build and a hit. Prompt them separately, a riser that only climbs and a hit that only lands. "A cinematic sting" comes back as the model's own fixed version of both joined together. Separate clips let you time the hit to the frame you are cutting on.
Sub-bass needs a partner frequency
Sub-bass below roughly 60 Hz is felt more than heard, and on a laptop speaker or phone it disappears. When you want low-end weight, pair it with a word that puts something in the audible range too ("grit", "metallic clang", "brass-like") so the sound still works on small speakers.
Say what to leave out
Cinematic prompts often return a generic combo when you wanted one piece of it. If you want a riser with no hit at the end, or a swell with no tail, say so ("no percussive hit", "no reverb tail"). The model tends to bundle these events together, and a negation separates them.
Use ACE-Step for a sung choir
Stable Audio can render a wash that sounds vaguely choral, but it has no vocal path: no voice sings real syllables. ACE-Step is built for that. Give it a short lyrics block and its choir sings it. Use it when a trailer moment needs an intelligible chant or hook; use Stable Audio for a generic "epic choir" pad.
Why generate locally
The models run on your machine, which changes the cost, the privacy and what you can automate.
Offline after download
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.
Your agent can use it
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.
Licences stated up front
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}'Frequently asked questions
- Can I use these sounds in a commercial video or game?
- It depends on the model that made them. Soundbox shows each model's licence before you download its weights. The Stable Audio 3 models use the Stability AI Community License: commercial use is free below USD $1M annual revenue, and above that Stability requires an enterprise licence. ACE-Step 1.5 is MIT. MiniMax's models allow commercial use under a revenue cap and require an attribution notice. Some voice and stem-separation models are non-commercial only. Soundbox itself is MIT, but the audio is governed by the licence of the model that made it. Read that licence before you ship; this summary is not legal advice.
- Does this run offline?
- Yes, once the weights are downloaded. Soundbox runs the models locally and stores everything in one SQLite file and one folder of WAVs under ~/.soundbox. There is no account, API key or upload step. Only the initial model download needs the network.
- What hardware do I need?
- An NVIDIA GPU with 8 GB or more. CPU-only is supported, but each generation then takes minutes. That is fine for a few short clips and slow for a large batch.
- Can my coding agent generate these for me?
- Yes. While Soundbox is running it serves MCP at http://127.0.0.1:17494/mcp, with tools for generating and listing sounds. An agent can generate a sound as part of the task it is already doing and gets back an absolute file path to use in your project.
- How is this different from downloading a sample pack?
- A trailer sting pack gives you someone else's braam, in someone else's key and edit length. Generating gives you a riser or a hit at the duration your cut needs, with the low-end weight and grit set to match the rest of your mix. It costs time and a GPU.