schema/CHANGELOG.md

Annotation Schema Changelog

All notable changes to the annotation schema will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[1.0.0] - 2025-10-26

Added

Classification (31 categories)

• Communication: email_draft, email_reply
• Task Management: todo_list, reminder, shopping_list
• Content Creation:
  • Blog: blog_outline, blog_draft
  • Articles: article_outline, article_draft
  • Social Media: social_media_post, social_media_thread, linkedin_post, twitter_thread
  • Scripts: script_draft, presentation_outline
• Development:
  • Prompts: prompt_general, prompt_development, prompt_creative
  • Documentation: readme_dictation, documentation, code_comment
  • Engineering: bug_report, feature_request
• Personal & Professional:
  • Personal: journal_entry, memo, idea
  • Professional: meeting_notes, research_notes, project_planning
• General: question, other

Audio Defects (10 categories)

• significant_background_noise - High levels of ambient noise
• background_music - Music playing (copyright concern)
• background_conversations - Other people talking
• crying_baby - Baby crying audible
• traffic_sounds - Vehicle or traffic noise
• poor_quality - Low quality recording, distortion, clipping
• multiple_speakers - Multiple people speaking intentionally
• wind_noise - Wind interference on microphone
• echo - Noticeable echo or reverberation
• phone_ringing - Phone ringing or notification alerts

Content Issues (5 categories)

• side_conversation - Conversation not part of the note
• partial_content - Recording cut off or incomplete
• false_start - Recording started accidentally or restarted
• thinking_aloud - Stream of consciousness, unstructured
• self_correction - Speaker corrects themselves mid-recording

Languages (7 languages)

• english - English language content
• hebrew - Hebrew language content
• arabic - Arabic language content
• russian - Russian language content
• french - French language content
• spanish - Spanish language content
• german - German language content

Transcription Quality (5 levels)

• excellent - Near-perfect transcription
• good - Minor errors, generally accurate
• fair - Some errors but usable
• poor - Many errors, requires correction
• unusable - Transcription too inaccurate to be useful

Metadata

• notes - Free-form text annotations
• schema_version - Version tracking for annotations

Schema Structure

• Multi-select support for: classification, audio_defects, content_issues, languages
• Single-select for: transcription_quality
• Free-text field for: notes
• Automatic schema version embedding in annotations

Documentation

• Initial schema definition in JSON format
• Comprehensive description for each category
• Type specifications (multi-select, single-select, text)

Future Versions

Planned for 1.1.0

• Additional language support as needed
• Expanded audio defect categories based on real-world usage
• Performance metrics annotations (speaking rate, pause patterns)

Under Consideration

• Emotion/Tone annotations
• Topic/Domain classifications
• Urgency/Priority levels
• Speaker characteristics annotations

---

How to Update This File

When creating a new schema version:

1. Add a new section with version number and date
2. Use "Added", "Changed", "Deprecated", "Removed", "Fixed" subsections
3. List all changes with category IDs and descriptions
4. Update the schema file: annotation_schema_v{N}.json
5. Sync changes to Hugging Face dataset repository