hHumanenessPERFORMANCENET

An integrated research framework

Human performance,
made legible.

From the inner logic of a script to the observable detail of a scene. Two systems and four complementary taxonomies for understanding, generating and curating performances.

Understand the system Read the system guide ↗

Text · Voice · Body · Emotion · Situation

Source materialRecorded / generated scene
01Script
AnalysisAuthor the performance.
02Performance
AnnotationDescribe the evidence.

A shared description space

SituationEmoNetVoiceNetPerformanceNet
EncodeFilterGenerateEvaluate

Conditions, observations and judgments remain distinct.

One connected vocabulary.

Ordinary encountersSubtle social dynamicsCinematic extremes

01 / The architecture

Two ways into
the same performance.

Start with what should happen, or with what has happened. Keep the requested inner state separate from what an observer can actually hear and see.

01 / SOURCE → PERFORMANCE

Humaneness
Script Analysis System

Turn a book, screenplay, play or audio script into coherent direction for a virtual AI actor.

  1. Character MakeupThe physical, emotional, intellectual and psychological role.
  2. Dramatic ObligationsSeven scene obligations, including the Emotional Obligation, interpreted through that character.
  3. Acting Beats + Experience ChunksExact words, actions and delivery; one authored inner stream per beat.
  4. Generated performanceAudio or audiovisual output, to be evaluated separately.
Read the complete system ↗
02 / PERFORMANCE → DESCRIPTION

Humaneness
Performance Annotation System

Describe recorded real-world, staged reference or synthetic scenes using explicit evidence and structured dimensions.

  1. Access and evidenceAvailable modalities, participants, context and time intervals.
  2. Descriptive annotationsWhat the voice, body, action and interaction make observable.
  3. Interpretation and task fitQualified readings, observer impressions and declared evaluation criteria.
  4. Model and data usesScene encoding, training labels, reward supervision and diverse data selection.
Read the complete system ↗

Inside Script Analysis

A character, a scene,
a moment.

The illustrated three-layer guide makes the hierarchy concrete. Select an image to inspect it.

Supplied illustrations retain their earlier Actor-State Atlas title. The current name is Humaneness Script Analysis System.

02 / The shared vocabulary

Four lenses.
One description space.

Use the taxonomies together. Each contributes different information; none is a substitute for the others.

01

Situation Taxonomy

The scene’s circumstances, task and constraints. Original domains plus explicit facets for specialized cases.

Context
Where and under what demands? 02

EmoNet Taxonomy

40 emotion and state labels, with 320 intensity–expression cells. Authored feeling, public expression and observer interpretation stay separate.

Emotion
What kind of affective account? 03

VoiceNet Taxonomy

A specialist vocabulary for audible delivery. Read every dimension’s actual range, anchor wording and direction.

Voice
How does it sound? 04

PerformanceNet Taxonomy

82 descriptive fields spanning action, body, attention, interaction and multimodal coherence. Eight observer impressions form a separate layer.

Performance
What is being done and conveyed?
Edge cases belong within the space.

The 266 acting edge cases and 600 physical-extreme premises are named reference points. Some specify situations; others specify an emotional or vocal demand and leave context open. Their crosswalk makes that distinction visible.

03 / Put the vocabulary to work

The description is shared.
The task determines the use.

Large-scale data bucketing

More coverage.
Less repetition.

Choose plausible combinations across situation, emotion, voice and performance. Retrieve or generate candidates, annotate the actual evidence, then measure accepted clips in the resulting buckets.

  1. Specify targets
  2. Sample across the space
  3. Inspect the output
  4. Count supported coverage
  5. Control duplicates

Catalog counts describe reference material. They are not counts of a training corpus or proof that a model can perform each case.

04 / Explore the reference points

Find a performance demand.
See where it belongs.

Search the retained situation, edge-case and physical-extreme catalogs. The mapping records explain the proposed coordinates and any missing context.

Loading the case index…

Download the complete crosswalk ↓

05 / The complete reading library

Start with the system.
Keep every detail within reach.

English reference volumes share the same typography, naming and evidence conventions. Read online or download the PDFs with linked contents and bookmarks.

Download the complete library ↓Shared style guide ↗

Start with the systems

Humaneness PerformanceNet · System Guide

The full architecture and its uses, with a worked example.

Read ↗PDF ↓

Humaneness Script Analysis System

Source text, Character Makeup, scene obligations, Acting Beats and Experience Chunks.

Read ↗PDF ↓

Humaneness Performance Annotation System

Recorded and generated scenes: evidence, description, interpretation and evaluation.

Read ↗PDF ↓

The four taxonomies

PerformanceNet Situation Taxonomy

Original situation domains, proposed contextual facets and auditable case mappings.

Read ↗PDF ↓

EmoNet Taxonomy

40 emotion/state labels and all 320 intensity–expression cells.

Read ↗PDF ↓

VoiceNet Taxonomy

The pinned 57-field vocal codebook, including its actual anchor ranges.

Read ↗PDF ↓

PerformanceNet Taxonomy

82 descriptive fields, with 8 separate observer-impression scales.

Read ↗PDF ↓

Complete appendices

Appendix A · Original dimensions

41 original PerformanceNet dimensions and their complete anchors.

Read ↗PDF ↓

Appendix B · Added dimensions & impressions

41 added dimensions and 8 observer-impression scales.

Read ↗PDF ↓

Appendix C · Emotions & states

The complete inherited emotion and state registry.

Read ↗PDF ↓

Appendix D · Situation catalog

351 retained situation records.

Read ↗PDF ↓

Appendix E · Acting edge cases

266 compound acting edge cases.

Read ↗PDF ↓

Appendix F · Physical extremes

600 physical-extreme scene premises.

Read ↗PDF ↓

Appendix G · Surface & underlying state

Surface and underlying state: 535 pairs and candidates.

Read ↗PDF ↓

Appendix H · Morris index & character questions

Morris concepts, techniques, exercises and character development.

Read ↗PDF ↓

Appendix I · Non-speech events

180 non-speech event types.

Read ↗PDF ↓

Appendix J · Sources & glossary

Source provenance, legacy companion availability and historical notes.

Read ↗PDF ↓

Design conventions

Humaneness PerformanceNet · Style Guide

Shared editorial language, typography and reading conventions.

Read ↗PDF ↓
Data files, source provenance and verification

The machine-readable files expose definitions, anchors, examples and the proposed case mappings. They are specifications and editorial data, not trained models.

Frameworks, taxonomy extensions and model uses are research proposals. Source definitions, exceptions and editorial changes are documented in the volumes. The original 2,931 catalog records remain represented; named taxonomy volumes are alternative views of their relevant source material.

Illustration

Original supplied artwork. Read the current operational definitions in the Script Analysis System.