audio-composer
Generates genre-appropriate sound effects and looping background music for a generated game. When ELEVENLABS_API_KEY is set, calls the ElevenLabs text-to-sound-effects API for real MP3 audio; otherwise falls back to Web Audio oscillator synthesis. Produces sfx.json, music.json, and AudioManager.js. Run in parallel with art generation stages.
DeepseekModel
Curated skill
Quality Good · 48
v1.0.0
Get
https://deepseekmodel.com/api/download.php?id=ar9av-game-exa-skills-audio-composer-skill-md&format=skill
Download .skill
Standard format with system_prompt and model_config, ready for any agent framework
The actual content of the system_prompt field in the .skill file.
name audio-composer description Generates genre-appropriate sound effects and looping background music for a generated game. When ELEVENLABS_API_KEY is set, calls the ElevenLabs text-to-sound-effects API for real MP3 audio; otherwise falls back to Web Audio oscillator synthesis. Produces sfx.json, music.json, and AudioManager.js. Run in parallel with art generation stages. Audio Composer — SFX + Music Two-tier audio system: real AI-generated MP3s when ELEVENLABS_API_KEY is available, deterministic chiptune oscillator synthesis as a zero-cost fallback. The same AudioManager.js handles both transparently — no code change needed in Game.js. When to use Run in parallel with the art generation stages (sprite-artist, tile-artist, bg-artist), after GDD is finalized. Also run on an existing game when the user says "add sound", "add music", or "upgrade to ElevenLabs audio." Two modes Mode A — ElevenLabs (when ELEVENLABS_API_KEY is set) Calls POST https://api.elevenlabs.io/v1/sound-generation for each SFX action. Uses the eleven_text_to_sound_v2 model with a genre-flavoured text prompt (e.g. "2D action platformer game, retro 8-bit jump sound, quick upward pitch sweep" ). Saves MP3s to public/assets/sfx/<name>.mp3 . Adds mp3Path to each entry in sfx.json . AudioManager.js preloads all MP3s as AudioBuffer s via decodeAudioData and plays them on demand. Rate limit: requests are made sequentially with automatic retry on HTTP 429. Mode B — Web Audio oscillator (no API key, or --no-elevenlabs ) Writes oscillator parameters ( freq , dur , type , vol ) to sfx.json . AudioManager.js generates sound at runtime using OscillatorNode — no audio files, no network calls, works offline. Output contract public/assets/sfx.json — SFX descriptors (always written) public/assets/music.json — music loop: BPM + pentatonic note sequence public/assets/sfx/<name>.mp3 — real audio files (Mode A only) src/audio/AudioManager.js — fixed runtime template (never regenerated) game-state.json gains state.audio : { "audio" : { "sfxPath" : "assets/sfx.json" , "musicPath" : "assets/music.json" , "managerPath" : "src/audio/AudioManager.js" , "elevenlabs" : true , // true when Mode A ran "mp3Count" : 6 // number of MP3s successfully generated } } sfx.json shape Mode A (with MP3): { "jump" : { "freq" : 380 , "dur" : 0.14 , "type" : "square" , "vol" : 0.22 , "mp3Path" : "assets/sfx/jump.mp3" } } Mode B (oscillator only): { "jump" : { "freq" : 380 , "dur" : 0.14 , "type" : "square" , "vol" : 0.22 } } AudioManager.play('jump') checks for mp3Path first; if missing or load fails, falls back to the oscillator params. Per-sound graceful degradation — a failed ElevenLabs response for one SFX doesn't break the rest. Genre presets Genre BPM SFX keys action-platformer / platformer 130 jump, land, hit, pickup, death, win top-down-adventure / dungeon-crawler 100 hit, pickup, death, win top-down-rpg 88 pickup, hit, win shoot-em-up 160 shoot, explosion, hit, death, win beat-em-up 140 hit, death, win AudioManager usage (codesmith adds these) import AudioManager from '../audio/AudioManager.js' ; // In create(): AudioManager . init ( this ); // registers Safari AudioContext unlock // In event handlers / overlaps: AudioManager . play ( 'jump' ); // plays MP3 if available, oscillator if not AudioManager . play ( 'pickup' ); AudioManager . play ( 'hit' ); AudioManager . play ( 'death' ); AudioManager . play ( 'win' ); AudioManager . stopMusic (); // call on game-over // Optional: AudioManager . mute (); AudioManager . unmute (); Safari iOS compatibility AudioManager.init(scene) registers one-shot pointerdown and keydown listeners that create and unlock the AudioContext on first interaction. MP3 buffers are decoded at that point. This satisfies Safari iOS's user-gesture requirement without any extra plumbing in Game.js. Process Read game-state.json → gdd.genre . Run scripts/compose_audio.mjs <project-dir> . If ELEVENLABS_API_KEY is set: generate MP3s, fallback per-sound on error. Otherwise: oscillator-only mode. Patch game-state.json with state.audio . Codesmith reads state.audio and adds import AudioManager + AudioManager.init(this) to Game.js . Scripts scripts/compose_audio.mjs <project-dir> [--bpm N] [--no-elevenlabs] --no-elevenlabs — force oscillator mode even if the key is set --bpm N — override music tempo
Keywords that activate this skill. Click one to copy it.
This skill does not provide trigger words.
The downloaded .skill package contains the following fields.
| Field | Description |
|---|---|
| format | Format tag (skill/v1) |
| skill_id | Unique skill ID |
| name | Skill name |
| version | Version |
| description | Description |
| category | Categories (array) |
| trigger_words | Trigger words |
| tags | Tags |
| source | Source |
| source_url | Source URL (this page) |
| exported_at | Exported at (set per download) |
| system_prompt | System prompt body |
| model_config | Model config: provider / model / temperature / max_tokens / top_p |
| examples | Examples |
| install_guide | Import guide for Coze / Dify / Claude / custom frameworks |
The same skill can be exported in different platform formats.