The result is an album with a song for every success criterion in WCAG 2.2. That’s 87 tracks: all 86 current criteria, plus a farewell to 4.1.1 Parsing, which was removed in WCAG 2.2.

More than 80 genres, six languages

Each song has its own musical style, and many of the pairings were chosen on purpose:

  • 4.1.1 Parsing, retired from the standard, gets “Last Post for Parsing,” a bugle call in the style of Taps.
  • 3.1.1 Language of Page is sung in German as industrial metal (“Sprachcode”).
  • 3.1.2 Language of Parts is J-pop that moves between Japanese, English, Spanish, and French (“Language Blossom”).
  • 1.4.7 Low or No Background Audio is an a cappella track called “Twenty Decibels Down.”
  • 2.1.4 Character Key Shortcuts became “The Accidental Polka.”

Across the album you’ll hear lyrics in English, Spanish, Italian, German, French, and Japanese.

How we made it

For each criterion, we:

  1. Picked a real song as musical inspiration.
  2. Turned the W3C “Understanding” page into a plain language summary.
  3. Asked Anthropic’s Claude to name the genre and write lyrics from that summary.
  4. Generated the music with Suno.
  5. Created cover art with OpenAI’s image model.
  6. Published each song with captions taken from the actual lyrics, not auto-generated ones.

Where people stepped in

AI did a lot of the work, but it needed people at every step:

  • Pronunciation. Suno sang “3.3.5” as “3.4.5,” so every song now spells its number out (“three point three point five”).
  • Accuracy. Even counting is hard for AI. Asked how many success criteria WCAG 2.2 has, one model said 86 and another said 78. We checked every song against the source.
  • Cultural sensitivity. We reviewed lyrics that borrowed language from the cultures behind a genre, and rewrote some of them.
  • Do-overs. Some songs were regenerated. Our first heavy metal version of 1.3.1 was so loud its captions wouldn’t sync, so it was replaced. We remixed another in a studio app to remove drums Suno insisted on adding.
  • Prompt revisions. We revised the songwriting prompt six times. Each song’s page lists the prompt version used to make it.

What’s new in 2026

We’re revisiting the album with newer AI models and our own notes from the first round. We’re also writing a new set of 13 songs, one for each WCAG guideline, such as 1.1 Text Alternatives and 2.5 Input Modalities.

Tell us what you think

This album was made with AI, so mistakes are possible. If a lyric is inaccurate, confusing, or misses the mark, we want to hear about it. Every song page has a thumbs up and a thumbs down, and a thumbs down asks what didn’t work. For anything else, use our contact page.

Start listening

Full playlist

Technology behind the music

Creating these accessible musical pieces involved a blend of cutting-edge technology and tools, including:
– Python — The core programming language used for this project.
– OpenAI Python Library — For interacting with OpenAI’s GPT models to summarize information in plain language and generating images for album art.
– Anthropic Python SDK — For interacting with Claude models from Anthropic to generate lyrics.
– Suno — AI-powered platform for generating original music and audio content.
– Ableton Live — Digital audio workstation (DAW) used for music editing of specific issues in generated songs.
– TechSmith Snagit — Image annotation tool used for updating visuals.
– Deepgram — AI-powered speech recognition platform for improving the accuracy of syncing lyrics as captions.
– YouTube API — API for programmatically accessing and managing the songs on Equal Entry’s YouTube channel.
– gspread — For reading and writing Google Sheets from Python to manage the 86 tracks.
– Pillow (PIL) — For image processing, resizing, and overlaying branding onto generated videos.
– WordPress REST API — For programmatically updating Equal Entry website content.
– Markdown — For converting Markdown text to HTML.
– pyperclip — For cross-platform clipboard operations.
– Beautiful Soup — For parsing and extracting information from HTML, namely the WCAG success criteria.
– ffmpeg — For combining images and audio into video files for posting to YouTube.

WCAG 2.2

Rate a song

0:00 / 0:00

Learn the guidelines behind the songs

Every song on this album is a WCAG success criterion. Our self-paced course, Understanding the guidelines, teaches the guidelines themselves. A complete introduction to WCAG Level A and AA, organized around its four principles.

Take Understanding the guidelines