Audiovisual Script Production

Convert source material into a script that can be recorded, visualized, edited, and checked, or review an existing script or cut plan against the same standard. Adapting is not illustrating the text line by line: written narration controls the experience through words, selected detail, voice, and access to consciousness, while a screen work arranges time, space, bodies, images, voices, silence, music, and editing. The job is to decide which experience must survive and which form carries it best in the new medium. Preserve the work's facts, canon, uncertainty, point of view, and intended effect. Never invent missing canon to make a document look complete.

Modes and Scope

Select the mode from the request:

• Convert: adapt supplied narration, story, treatment, testimony, or research into a production script.
• Review: diagnose an existing treatment, script, shot list, storyboard, animatic, or cut plan. Do not rewrite it unless the user also asks for corrections.
• Review and revise: diagnose first, name what must be preserved, then apply the approved or requested corrections to the artifact.

When a request combines modes, convert first and review the result before delivery. If the material is enough to proceed, make the ordinary production choices yourself and mark uncertain project data as PENDIENTE; do not stop to ask optional questions. Ask only when no source material was supplied or when a decision would change canon.

Write the deliverable and the report in the language of the source material unless the user asks otherwise. Format tokens, headings, and state labels of a house format stay exactly as that format defines them.

What can be read. This skill works from text and still images: scripts, treatments, shot lists, storyboard frames, subtitle files, EDL or timeline exports, transcripts with timecodes, and the user's own viewing notes. Audio and video cannot be played. A rough cut or master is reviewed through whatever textual or visual evidence the user supplies, and every pass whose evidence the artifact does not carry is reported as not assessable, never inferred. Do not describe a sound, a performance, or a frame that was not in the material.

Intake

Establish from the supplied material, and record what was inferred:

• Source hierarchy and canon authority: which document wins when two disagree. If the user names an authority or version, follow it; otherwise expose the conflict and do not choose a new canon silently.
• Target: platform, format, audience, language, duration, aspect ratio, and viewing conditions. The same story changes when it becomes a short, an episode, a documentary, an animation, a vertical piece, or an interactive route.
• Fidelity hierarchy: fidelity to facts, to the emotional arc, to point of view, to tone, to theme, to the world and its rules, to a cultural or historical experience, or to a specific audience's expectations. When these conflict, state the order chosen.
• Production frame: available voices, locations, characters, assets, tools, budget, calendar, rights, and delivery constraints. Distinguish facts from assumptions and proposals.
• Production instructions: the project's own document, when one exists, fixing its project code and piece-code scheme, target platform and delivery specification, the tool for each pipeline function, narrator and character voices with their system and supported markup, generated-clip and lip-sync segment limits, the location of asset references and prompt-base sheets, its own master-script format, and the order of canon authority. It governs every production value in the deliverable. When the project has none, the defaults under The Production Master Script apply, and each one is labeled a proposal in the production notes.
• Audio state: whether recorded or synthesized audio already exists. Existing audio fixes timing; nothing else does.
• Output scope: script only, full production dossier, review report, revised artifact, or a combination.

Rights and limits of use come first in any adaptation: who controls the work, characters, illustrations, photographs, music, testimony, and archive material. For real events, separate public facts, original research, protected material, and claims that need legal verification. Record what is unresolved as PENDIENTE; never report rights as cleared without evidence.

Adaptation Method

1. Extraction reading. Read the source from four perspectives: facts and what causes each event; the reader's experience, expectation, and discovery; devices that depend specifically on the writing; and production demands such as locations, cast, effects, period, permissions, or sensitive material. The result is an inventory, not a script: premise and theme, protagonist or center of attention, desires, needs and opposing forces, decisive changes, relationships, recurring images, spaces and sounds, world rules, indispensable information, iconic scenes, elements that can be dropped, merged, or are costly, and ethical or representational risks.
2. Central experience. Complete the sentence "the work seeks that the audience experiences ____ while understanding or questioning ____". The answer describes an experience and a thematic direction, not a plot summary.
3. Adaptation thesis. State whose version this is, its dramatic present, the question it explores, which experience of the source it preserves, what will change and why, and what it promises the audience.
4. Audiovisual motor. Identify the renewable source of action, tension, discovery, or contrast: a goal pursued under pressure, an investigation, a changing relationship, a journey, a ritual or process, a countdown, an episode-producing system, a documentary question, or a perceptual transformation. When the source relies on reflection, atmosphere, or voice, the motor can be a present situation that forces someone to act, observe, or remember.
5. Point of view and device. Decide who perceives the story, who tells it if a narrator exists, what the audience may know at each moment, and which channel delivers each piece of information. The device is the formal rule that organizes the gaze: interview, diary, investigation, narrator, split screen, subjective camera, archive, dated episodes, multiple testimonies. Check that it holds for the whole duration instead of becoming an obligation.
6. Sequence map. Reduce the story to movements with a beginning, development, and change. Each sequence alters at least one variable: goal, plan, knowledge, relationship, power, cost, available time, or the audience's interpretation. Give each one a purpose, entry and exit condition, point of view, and estimated duration. No three-act structure is required; four acts, eight sequences, a circle, an episodic form, a mosaic, kishōtenketsu, an investigation, or an observational progression are equally valid when they describe the work.
7. Interiority. Translate thought into actions and decisions, physical behavior, contradictory goals, dialogue and subtext, performance, gaze and silence, space, costume, props and light, subjective sound, associative editing, memories, dreams or fantasies, on-screen text, or voiceover. Voiceover is legitimate; it weakens when it repeats what the image already shows or explains what the scene could dramatize, and it works when it adds counterpoint, intimacy, structure, irony, inaccessible information, or a literary voice that is central to the experience.
8. Operations. Apply selection, omission, condensation, expansion, transposition, reordering, invention, and refocalization only where the thesis supports them, and record for each change which function it preserves, replaces, or creates. Composite characters save time but can erase perspectives or attribute real actions to one person; in nonfiction, document these decisions with special care. A cleaner chronology is not automatically a stronger adaptation when mystery or subjective discovery depends on the original order.
9. Spoken text. Draft narration and dialogue for breath groups, pronounceable numbers, deliberate pauses, and controlled variation. Dramatize: a conversation is fully dramatic when it changes power, knowledge, or a relationship. Split long blocks at meaningful beats when a performer or tool has an input limit.
10. Duration. Take timing from existing audio whenever it exists. Otherwise estimate spoken material at 150 words per minute for measured Spanish narration, adjusted to the project's own measured rate from earlier episodes when one is known, and label every figure ESTIMADO with a 10% margin. Never present an estimate as a measured time, and never derive it from page count alone.
11. Channels. Design image and sound as complementary narrative channels. Remove triples in which narration, image, and music state the same idea; give the image behavior, the dialogue an objective, and the sound space, memory, or counterpoint.
12. Development ladder. Move through the lightest useful artifact: source inventory and authority map, thesis and audience promise, sequence map with duration budget, treatment or beat sheet when needed, spoken script and scene directions, read-through and timing pass, production plan with asset list and dependency check, and storyboard, shot list, or animatic only when requested or needed to reduce risk. A later document cannot repair an unresolved earlier decision: an extensive asset list does not fix an unclear protagonist, a weak causal chain, or a missing ending.
13. Self-review. Run the review passes below on the draft and repair every high-priority finding before delivery.

If the narrative-review skill is available and the source is fiction, apply its verification steps to the narrative pass. If the humanize skill is available, run it on newly drafted spoken text only, with quoted source passages and format tokens supplied as protected material.

Adjusting to the Format

When the target is a streaming or social platform, check that the opening establishes the promise before the first likely drop-off point and that chapter or section boundaries fall on narrative changes rather than on round timecodes.

Audio-First Workflow

Audio-first applies when narration, dialogue, music, or a fixed spoken performance controls duration, and it is the rule in a pipeline where visual assets are generated to match audio: music dictates cuts and visual rhythm, the voice track defines shot duration, and generating visuals first produces mass re-editing when the audio changes. The order is:

1. finalize or time the spoken material;
2. establish the major musical and silence beats;
3. divide the track into visual units;
4. assign or create visual material for those units;
5. add effects and detailed transitions;
6. mix, caption, and verify the result.

Before assigning any shot or asset duration, establish whether the audio exists. If it does, its timing is the primary duration source. If it does not, every duration is ESTIMADO. Audio-first is a project method, not a reason to ignore visual causality or editorial rhythm.

Image, Sound, and Editing

Image. Choose framing, composition, movement, light, color, and lens behavior for a narrative reason: a shot reveals information, shapes emotion, establishes space, redirects attention, creates contrast, or completes a transition. Shot size has no fixed meaning; it depends on duration, lens, angle, movement, performance, light, sound, and neighboring shots. Before moving the camera, complete "the camera moves because ____ changes". Establish location, orientation, entrances, exits, eyelines, and screen direction before breaking continuity; discontinuity is valid when it reads as a decision. For generated or composited material, track character identity, age, costume, props, handedness, weather, time of day, light direction, lens feel, and spatial relationships; reference images and stable descriptions support consistency but do not replace a continuity check.

Sound. Design the layers that apply: dialogue, ambience, specific effects, foley, sound design, diegetic and non-diegetic music, deliberate silence, narration, and audio description. Decide what belongs inside the scene world and what functions as score or commentary. Perspective, distance, reverberation, and room tone follow the apparent space unless a subjective break is intentional. Dialogue must stay intelligible, which is not the same as loud. Judge music by function: establishing or transforming tone, continuity, representing a character or idea, counterpoint, modulating expectation, or belonging to the scene. Music loses force when it dictates every emotion, compensates for a scene without tension, or flattens moments that should contrast; test the scene without it. Loudness targets come from the destination's specification, never from a universal number.

Editing. A cut controls information, duration, rhythm, association, and point of view. Enter a scene once a question exists and leave it once its effect is complete, keeping the breaths that let a change register. Slow is not a defect in itself; the problem is duration that produces no observation, tension, atmosphere, expectation, or transformation. Do not declare picture lock while narrative decisions remain open.

Previsualization. A shot list records scene and shot, narrative purpose, subject and action, size, angle, lens and movement, related sound or dialogue, technical needs, priority (essential, desirable, dispensable), and continuity or effects dependencies. Priority protects the story when shooting or generation time shrinks. A storyboard need not be illustrated; it must communicate framing, action, direction, changes, and the relation between shots. An animatic adds duration, movement, and provisional sound, and checks legibility, total duration, rhythm, shot count, transitions, and room for captions and audio description.

Review Method

Record before reading: title, version and date, duration or pages, stage, audience and platform, genre and promise, declared intent, constraints, the team's specific questions, and which elements are still temporary. Never issue notes against an ambiguous version, and separate failures of the evaluated material from problems the stage does not yet allow to judge.

Perform the passes the artifact supports, in this order. Separate observation from diagnosis: a reaction locates a problem, it does not explain the cause.

1. Continuous experience. Read without stopping. Note interest, confusion, anticipation, changes of goal or understanding, drops in attention, emotion felt, information remembered, moments that feel repeated, rushed, or unmotivated, and the questions the ending answers or leaves open. Do not diagnose yet.
2. Narrative map. Divide the piece into sequences and note entry and exit time, point of view, goal or question, new information, change of value, relationship, plan, or power, promise or clue, transition, visual and sound function, and consequence. Look for causal and thematic links; a sequence may justify itself by contrast, atmosphere, or perceptual transformation, not only plot advance. Check setup and payoff, motivation, escalating difficulty, irreversible choices, reversals, climax, and aftermath. A coincidence may begin trouble; repeated coincidences should not solve it.
3. Image. Check geography, composition and hierarchy, intentional or accidental discontinuity, eyelines, coverage, exposure and stability artifacts when frames are supplied, light and color continuity, graphics, text, captions and safe areas, unintended representation, and the relation between image and theme. When frames or a shot list allow it, ask whether action, attention, and space remain legible without the sound; do not make that test a rule, since a work may deliberately depend on its soundtrack.
4. Sound. Check dialogue intelligibility, acoustic perspective and continuity, noise, distortion, clicks, breaths, and cuts, ambience and sense of space, sync, dynamics and loudness, transitions, the function of music and silence, legibility in the destination's channel layout, and the balance of audio description and dubbing. From a script, this pass checks the sound design as written and the intelligibility risks of the spoken text.
5. Edit and pace. Check entry and exit of each scene, duration of actions and reactions, clarity of ellipses, accumulation of tension, variety and patterns, the relation between internal shot rhythm and cutting rhythm, anticipation and reward, repeated information or emotion, transitions between story lines, and the space after the climax.
6. Production. Verify duration math, filenames, asset coverage per scene, voice and character consistency, dependencies, technical targets, and delivery needs against the actual platform specification. Do not deduce requirements from another platform.
7. Accessibility. Check synchronized captions that carry relevant sound information, speaker identification when not evident, transcript, description of necessary visual information or an equivalent alternative, contrast and duration of on-screen text, no information carried by color alone, warnings for flashing or sensitive content where context and regulation require them, and narrative and sound room for accessible versions. Accessibility designed at the script stage costs least.

Writing Findings

Each finding carries an identifier and title, a priority, a confidence, a location (timecode, scene, section, or line anchor; never an invented one), observable evidence, mechanism or likely cause, impact on the audience or production, the minimal correction, alternatives when they clarify the goal, what to preserve, and how to verify the fix. Give the problem before the solution: the creative team may find a better one. Group symptoms with a common cause while keeping their separate locations; ten slow scenes can come from a protagonist without a comprehensible goal, not from ten independent duration problems. Name the value every fix must not lose: performance, ambiguity, humor, silence, a clue, a composition, information, or emotion.

• High: compromises comprehension, the motor, the climax, an essential rule, safety, rights, or delivery.
• Medium: weakens a sequence, relationship, twist, rhythm, or continuity.
• Low: a localized gain in clarity, finish, or consistency.

Keep priority separate from confidence. Do not invent findings to fill a quota, judge the work by a favorite template, confuse taste with defect, invent a universal audience reaction, demand final finish from an animatic or first cut, or attribute a structural problem to temporary music, effects, or color.

Frequent Artifacts

The Production Master Script

The conversion deliverable is one master script that serves as performance document and production dossier at once. Its structure is a template: the project's production instructions fill in names, limits, tools, and delivery values, and this section supplies defaults only where the project is silent. Its order follows the audio-first dependency chain:

1. identity and production metadata;
2. production notes: audio-first status, markup conventions, constraints and decisions;
3. characters and voices;
4. timed episode structure;
5. Section I, record-ready spoken material per scene, with output filename, performance direction, and the complete text block;
6. Section II, voice summary;
7. Section III, music design with a copy-ready brief;
8. Section IV, scene-by-scene audio and visual plan;
9. Section V, consolidated requirements: per scene, visual assets, audio assets;
10. canon verification, production control, document metadata, change log, and cross-references.

Include a section only when it has a consumer in recording, generation, editing, review, or archive work; never an empty one to imitate the shape. Headings divide stages and scenes; tables carry structured facts, totals, settings, filenames, and verification states; code fences hold copy-ready material such as spoken text, voice direction tags, music briefs, and production cues. Scene identifiers stay stable across the spoken section, the production section, and the requirements tables. Repeated characters and objects receive stable asset codes. Audio and visual directions state an intended effect, not only an object list, and the visual block describes the sequence of screen events closely enough to support storyboarding and asset assignment; it is not a shopping list. Every time is ESTIMADO until measured audio exists.

Use only markup the selected voice system supports. When none is selected, use neutral performance cues instead of inventing SSML compatibility or parameter values. Canon claims point to a source document and carry a state: VERIFICADO, CONFLICTO, or PENDIENTE; a missing source is PENDIENTE, never VERIFICADO. Confirmations of settings and assets use CONFIRMADO or PENDIENTE. The change log classifies its entries as CRITICO, MAYOR, MENOR, or ESTILO; that scale belongs to the script's own history and is separate from the High, Medium, Low priority of review findings. Document metadata describes the artifact, not the tools or people that drafted it; add no production-tool attribution.

Default Production Constraints

These values describe an audio-first pipeline with synthesized voices and generated imagery. A project's production instructions replace any of them; without instructions, they are proposals and are recorded as such.

Recurring characters and key artifacts carry a base prompt in the project's reference sheets: physical appearance, standard clothing or materials, behavior or active and inactive state, and visual style. The script's requirements tables reference those sheets by asset code; they do not restate them.

Template

Adapt this skeleton to the project; drop what the project does not consume, and let the project's production instructions override any heading, label, or field. Headings, markup, and state labels below are the default house format and stay in Spanish inside a Spanish deliverable; the bracketed placeholders describe what goes in each slot.

# GUION: [IDENTIFIER -- TITLE]
## *[Project or series]*

**Version:** [X.Y.Z]
**Estado:** [BORRADOR / EN REVISION / APROBADO]
**Clasificacion:** Guion Audiovisual / [subtype]
**Estilo predominante:** [style]

**Rol/Funcion:** [narrative and production function]
**Naturaleza/Categoria:** Guion para produccion audiovisual
**Dominio/Ambito:** [platform, series, or channel]
**Fuente principal:** [adapted narration or documents]

---

## METADATOS DE PRODUCCION

| Campo | Valor |
|---|---|
| **Designacion** | [code] |
| **Duracion objetivo** | [duration] |
| **Formato** | [video, episode, short] |
| **Audiencia** | [audience] |
| **Narrador** | [voice or PENDIENTE] |
| **Voces requeridas** | [list] |
| **Total de escenas** | [number] |

## NOTAS DE PRODUCCION

### Paradigma Audio First

[Whether it applies, which audio exists, which timings are still pending.]

### Convenciones de Marcado

- `[MUSICA:]` musical cue
- `[SFX:]` sound effect
- `[AMBIENTE:]` continuous atmosphere
- `[VISUAL:]` visual action or treatment

### Restricciones y decisiones

- [canon, platform, tools, budget, or limits]

## PERSONAJES Y VOCES

| Personaje o voz | Funcion | Aparicion | Estado de referencia |
|---|---|---|---|
| [name] | [function] | [scenes] | [VERIFICADO / PENDIENTE] |

## ESTRUCTURA

| Escena | Titulo | Tiempo estimado | Duracion | Cambio narrativo | Tono |
|:---:|---|---|---:|---|---|
| 1 | [title] | [00:00-00:00] | [0:00] | [what changes] | [tone] |

> Los tiempos estimados deben reemplazarse por tiempos medidos cuando exista el audio.

---

# SECCION I: MATERIAL HABLADO

## CONFIGURACION BASE DE VOZ

| Parametro | Valor | Estado |
|---|---|---|
| Sistema o interprete | [value] | [CONFIRMADO / PENDIENTE] |
| Idioma | [value] | [CONFIRMADO / PENDIENTE] |
| Perfil interpretativo | [value] | [CONFIRMADO / PENDIENTE] |

## ESCENA 1: [TITLE]

**Archivo de salida propuesto:** `[PROJ_S##E##_VOZ-ESCENA1-01-v01.ext]`

### Direccion interpretativa

| Aspecto | Indicacion |
|---|---|
| Tono | [tone] |
| Ritmo | [rhythm] |
| Evolucion | [change during the scene] |

### Texto de voz o dialogo

```text
[Complete text, ready for performance or synthesis.]
```

---

# SECCION II: RESUMEN DE VOCES

| Escena | Voz | Direccion | Archivo | Estado |
|:---:|---|---|---|---|
| 1 | [voice] | [direction] | [file] | [PENDIENTE] |

---

# SECCION III: DISENO MUSICAL

## Tema principal

| Campo | Valor |
|---|---|
| Intencion | [narrative function] |
| Duracion | [duration] |
| Pulso | [BPM or description] |
| Instrumentacion | [instruments or textures] |
| Evolucion | [start, transformation, close] |

### Brief o prompt musical

```text
[Musical description ready for a composer, library, or the chosen tool.]
```

---

# SECCION IV: GUION DE PRODUCCION POR ESCENAS

## ESCENA 1: [TITLE]
**[00:00 - 00:00, ESTIMADO]**

### Especificaciones de audio

| Elemento | Especificacion |
|---|---|
| Voz | [intent] |
| Ritmo | [intent] |
| Musica | [function and evolution] |
| SFX | [function] |
| Ambiente | [space and perspective] |

### Audio

```text
[AMBIENTE: ...]
[MUSICA: ...]
[SFX: ...]
```

### Voz o dialogo

[See SECCION I, Escena 1. Note insertions and their placement here.]

### Visual

```text
[VISUAL: In order: point of view, action, information, transition,
and the intended effect.]
```

---

# SECCION V: REQUERIMIENTOS AUDIOVISUALES

## A. Requerimientos por escena

| Escena | Voz | Musica | SFX/Ambiente | Visual | Duracion | Dependencias |
|:---:|---|---|---|---|---:|---|
| 1 | [voice] | [music] | [sound] | [visual] | [0:00] | [dependencies] |

## B. Assets visuales

| Codigo | Descripcion funcional | Escena | Tipo | Duracion o cantidad | Estado |
|---|---|:---:|---|---:|---|
| [IMG-01] | [description] | 1 | [imagen / video / grafico] | [value] | [PENDIENTE] |

## C. Assets de audio

| Codigo | Descripcion funcional | Escena | Archivo propuesto | Estado |
|---|---|:---:|---|---|
| [VOZ-01] | [description] | 1 | [file] | [PENDIENTE] |

---

# VERIFICACION DE COHERENCIA Y CANON

| Afirmacion comprobable | Fuente autorizada | Estado | Nota |
|---|---|---|---|
| [claim] | [document/section] | [VERIFICADO / CONFLICTO / PENDIENTE] | [note] |

# CONTROL DE PRODUCCION

| Comprobacion | Estado | Evidencia o accion pendiente |
|---|---|---|
| Duracion total calculada | [state] | [detail] |
| Audio y visual sin redundancia innecesaria | [state] | [detail] |
| Assets cubren todas las escenas | [state] | [detail] |
| Continuidad de personajes y objetos | [state] | [detail] |
| Subtitulos y accesibilidad | [state] | [detail] |
| Especificaciones de entrega | [state] | [detail] |

# METADATOS DEL DOCUMENTO

**Nombre oficial:** [name]
**Fecha de creacion:** [YYYY-MM-DD]
**Fecha de actualizacion:** [YYYY-MM-DD]
**Version:** [X.Y.Z]
**Clasificacion:** [classification]
**Proyecto:** [project]
**Formato:** Markdown
**Estado:** [state]

# REGISTRO DE CAMBIOS

**vX.Y.Z ([YYYY-MM-DD]):**

- [CRITICO / MAYOR / MENOR / ESTILO] [change and reason]

# REFERENCIAS CRUZADAS

- **[Reference]** [use and scope]

---

**FIN DEL DOCUMENTO**

Deliverables

Conversion returns the requested Markdown artifact. When no narrower scope is given, deliver one master script with spoken material, scene production directions, requirements, and verification tables in the master-script format above, plus the adaptation thesis and the record of canon changes. Filenames are proposals until approved unless filenames are the requested deliverable.

Review returns, in this order:

1. Verdict and readiness. What the work promises, how far it delivers, the main risk or opportunity, and which passes were not assessable from the material.
2. Strengths to preserve. Concrete evidence and the function each one serves.
3. Status by area. Promise and closure, structure and causality, characters and performance, point of view and information, visual language, sound and music, editing and rhythm, continuity, accessibility, technical quality and delivery; each one meets, partial, problem, not applicable, or not assessable.
4. Prioritized findings. As specified under Writing Findings. A low finding may be compact.
5. Revision plan. Ordered by dependency: rights, safety, and non-negotiable requirements; central experience, structure, and causality; characters, point of view, and information; sequences, scenes, and editing; sound, music, color, and effects; accessibility, text, versions, and delivery. State how to verify each major change.

Review and revise delivers the corrected artifact and a change summary with one entry per finding: identifier, location, before, after, and what to verify. Unresolved high-priority issues stay visible at the top of the summary, never buried in the log. For a second review, state which findings are resolved, remain, or changed, and confirm the strengths to preserve are intact.

Applying Corrections

Before editing, freeze facts, quotations, canon boundaries, approved timings, technical identifiers, filenames, and project-specific syntax. Preserve direct quotes unless the user authorizes changes. Rewrite only the findings the user selected, and where a finding offered options, have the user choose first. Make the smallest revision that solves each verified problem; edit only the passages the findings cite, and leave everything else byte for byte. An inserted line imitates its neighbors in register, rhythm, and form. Then recheck downstream timing, continuity, assets, references, and totals, because a change in one scene moves the numbers in three tables.

Never fabricate completed assets, measured timings, permissions, rights, technical tests, audience results, or canon verification. Mark them PENDIENTE when no evidence exists.

Example Use

"Use /video-script to convert this narration into the master script for episode S01E03. The project's production instructions are attached; the narrator's audio does not exist yet; the canon sources are the series bible and the two previous scripts. Deliver the full script with estimated timings and the list of canon changes."