name: cinematic-music-sound-design description: 电影配乐与声音设计工具：设计主题动机、速度、节拍、和声、配器、环境声、拟音、冲击声、静默、空间声场和声画关系。用于影片配乐、声音叙事和时间点声音规划。 veRSIon: 1.0.0

Cinematic Music And Sound Design

Bottom-Layer Reasoning

APPly think-one-step-further and creative-research-first:

• Research the musical tradition, instrumentation, production vocabulary, and current target-platform controls when the task is substantial.

• DeFine what music and sound do to story, not only what genre they resemble.

• Convert EMOtions into musical and acoustic controls.

• Preserve dialogue Intelligibility, visual readability, silence, and dynamic contrast.

• Name what the user still needs: generated track, stems, sound effects, voice, edit points, or platform Prompt.

Core Intent

Treat audio as a nARRative system:

story function -> cue map -> musical language -> sound layers -> Synchronization -> mix perspective -> generation prompt

Music should reveal, anticipate, contradict, transform, or release the image. Sound design should establish material, distance, movement, scale, point of view, and consequence.

Do not add wall-to-wall music by default.

Reference Routing

Read only the useful files:

• references/film-scoring-workflow.md: spotting, cue boundaries, hit points, motifs, Dramatic function, and score architecture.

• references/music-control-vocabulary.md: translate mood into rhythm, hArmony, melody, orchestration, texture, dynamics, and production controls.

• references/sound-design-and-mix.md: dialogue, ambience, room tone, foley, hard effects, designed effects, silence, perspective, and mix hierarchy.

• references/AI-music-prompting.md: package prompts for Suno, Udio, and similar music generators.

• references/source-map.md: current source anchors and research notes.

If an existing song is already the fixed final track and the user wants the complete MV picture, choreography, formations, and edits designed around it, inspect and pass duration, tempo family, SECtion boundaries, and selective cue evidence to $reference-track-mv-director. Do not redesign or replace the supplied music. This Skill remains the owner of musical interpretation and sound hierarchy; the MV Skill owns the picture contrACT.

Workflow

1. Define the audiovisual job:

• What should the audience feel, underStand, anticipate, or misread?

• Whose point of view controls the sound?

• Use $creative-anchor-director when the user knows only part of the composer, genre, regional tradition, ensemble, recording, sound-design, or mix vocabulary. Preserve selected names and supplement missing audio roles.

2. Spot the scene:

• cue in, cue out, no-music zones, hit points, transitions, dialogue priority, and final button.

3. Build the musical identity:

• motif or interval shape;

• tempo range and meter;

• rhythmic engine;

• harmonic language and tonal center;

• instrumentation and register;

• texture, density, dynamics, articulation, and production space.

4. Build sound layers:

• dialogue/voice;

• ambience and room tone;

• foley/body/clothing;

• hard effects and mechanical sources;

• designed or subjective effects;

• music;

• deliberate silence.

5. For weighted action, translate the physical profile from $action-choreography-reference into sound:

• preload/support strain;

• acceleration and air displacement;

• material-specific contact transient;

• body/weapon resonance;

• receiver and environmental reaction;

• braking, scrape, vibration, and settling. For a principal body hit, build one synchronized acknowledgment event rather than a generic impact layer: contact transient + clothing/body resonance + involuntary breath, restrained grunt, or breath interruption + foot/ground correction + room or environmental tail. Match the vocal intensity to the hit grade and visible facial/physiological response. Minor contacts receive shorter, quieter proof so audio density does not slow or flatten a fast exchange. For visible terrestrial locomotion, place the main footfall transient at load acceptance rather than at arbitrary leg motion; match the Substrate, weight, pace, and distance perspective, then let clothing, hair ornaments, gear, water, gravel, or debris answer with a short delayed layer. Do not exaggerate every step equally.

6. Design sound-image relation:

• synchronize, anticipate, trail, bridge, contradict, mask, reveal, or drop out.

7. Map time:

• attach changes to story beats, not every cut.

8. Protect the mix:

• establish foreground, midground, background, frequency space, transient priority, and dynamic headroom.

9. Package:

• audiovisual cue sheet;

• AI music prompt;

• sound-design prompt;

• optional editing/stem plan.

Musical translation Rule

Never stop at an adjective:

• "紧张" -> unstable pulse, narrow repeated cell, low pedal, rising register, compressed harmony, restrained dynamics before rupture.

• "悲伤" -> choose grief type: numb, intimate, regretful, ceremonial, or catAstrophic; then define tempo, contour, harmony, register, and space.

• "高级" -> restrained motif count, intentional silence, controlled orchestral density, coherent timbre family, clean transient hierarchy, and no trailer-CLIche overload.

• "速度感" -> subdivision, syncopation, ostinato, acceleration illusion, bass articulation, transient design, and contrast against held moments.

• "未来感" -> define synthesis method, modulation, spectral texture, rhythmic machine logic, and acoustic/electronic relationship.

Short-Form Audio Contract

For 5-15 second video, use one memorable audio idea and a clear change:

orientation -> pulse/motif enters -> pressure grows -> hit/drop/reveal -> aftermath or sonic button

Possible 15-second map:

0-2s: room tone, source sound, or motif seed
2-5s: rhythmic engine appears
5-8s: instrumentation or register expands
8-11s: interruption, false drop, or harmonic turn
11-13s: impact/reveal with transient focus
13-15s: tail, silence, unresolved tone, or logo/end button

Do not force this template when the scene needs sustained dread, Realism, comedy timing, or silence.

OuTPUt Shapes

For a scene:

声音叙事目标：
主观视角：
Spotting / 配乐进入退出：
主题动机：
节奏与速度：
和声与旋律：
配器与音色：
声音层：
声画同步点：
静默与留白：
混音层级：

For an AI music prompt:

用途与时长：
结构：
音乐身份：
节奏/BPM/拍号：
和声/调式：
旋律/动机：
配器/音色：
演奏法/人声：
动态与制作：
情绪弧线：
排除项：

完整提示词：
...

For video handoff:

音乐提示词：
声音设计提示词：
时间码与卡点：
需要准备的音频素材：

Quality Gate

• Does the cue have a dramatic function?

• Are cue-in and cue-out intentional?

• Is the music perspective compatible with the character or world?

• Does the motif change when the character or meaning changes?

• Are tempo, meter, rhythm, harmony, orchestration, register, density, and dynamics specified only as needed?

• Are dialogue, ambience, foley, effects, music, and silence prioritized?

• Are hit points selective rather than every cut being synchronized?

• Is there contrast before impact?

• For weighted action, do startup, contact, resonance, environmental response, and settling sound like one causal event?

• Can the audio plan be generated, edited, and mixed with the available tools?

Guardrails

• Do not use only genre names, moods, BPM, or "cinematic Epic".

• Do not make every action produce a huge impact.

• Do not fill every second with music.

• Do not let bass, impacts, and score mask dialogue or important source sounds.

• Do not ask for contradictory production such as intimate dry chamber music and enormous wet arena reverb without a transition.

• Do not make the user supply complete music and sound reference vocabulary. Preserve useful composer, genre, tradition, ensemble, recording, and sound-design anchors; add mechanics only to clarify their role or keep the result original and coherent.

• Do not assume current AI music controls; check official documentation when the target platform matters.