EPUBCheck vs Ace by DAISY: What Each Tool Tests—and Why Publishers Need Both

EPUBCheck vs Ace by DAISY: What Each Tool Tests—and Why Publishers Need Both

An EPUB passes EPUBCheck.

Does that mean it is accessible?

No.

An EPUB produces no serious automated accessibility violations in Ace by DAISY.

Does that prove the title is fully accessible?

Also no.

This distinction is one of the most important things a publishing production team needs to understand.

EPUBCheck and Ace by DAISY solve different problems.

EPUBCheck primarily answers:

“Is this EPUB technically conformant with the EPUB specification?”

Ace by DAISY helps answer:

“Are there accessibility problems that can be detected automatically?”

And human accessibility QA answers another question:

“Can a real reader using assistive technology actually understand and navigate the publication?”

DAISY’s Accessible Publishing Knowledge Base recommends a validation sequence that uses EPUBCheck for EPUB conformance, Ace for automated accessibility checking, and manual evaluation for accessibility requirements that cannot be reliably tested by software.

That gives publishers a much stronger QA model than relying on a single green report.

Quick Answer: EPUBCheck vs Ace by DAISY

The difference is straightforward:

The strongest workflow is:

EPUBCheck → Ace by DAISY → Human Review → Assistive-Technology Testing → Correct → Retest

DAISY explicitly notes that Ace does not replace EPUBCheck and that passing automated Ace checks is not sufficient to establish WCAG conformance.

What Is EPUBCheck?

EPUBCheck is a validation tool used to identify technical problems in EPUB files.

DAISY describes it as checking areas such as:

• EPUB container structure
• OPF/package markup
• Content-document markup
• Internal references
• EPUB specification requirements

In practical publishing terms, EPUBCheck helps answer:

“Is this EPUB constructed according to the EPUB rules?”

This matters because a technically broken EPUB may:

• Fail retailer ingestion
• Render incorrectly
• Lose navigation
• Contain broken references
• Fail to open in some reading systems

Technical conformance is therefore essential.

But it is not the same thing as accessibility.

What Can EPUBCheck Detect?

Depending on the EPUB and specification version, EPUBCheck can identify problems involving areas such as:

• Package structure
• Invalid markup
• Missing required resources
• Incorrect manifest information
• Broken links or references
• EPUB navigation problems
• Invalid content structure
• Unsupported EPUB constructs

The tool is valuable because it catches technical defects before an EPUB reaches distributors or readers.

But consider this example:

An image has:

alt="chart"

That alt attribute may be syntactically valid.

The EPUB can therefore be technically well formed.

But if the chart communicates a significant research result, the words “chart” do not give a nonvisual reader the equivalent information.

EPUBCheck cannot make that editorial judgment.

What EPUBCheck Does Not Tell You

A clean EPUBCheck result does not automatically mean:

• Image descriptions are meaningful
• Reading order makes sense
• Heading hierarchy is logical
• Tables are understandable
• Navigation is usable
• Link text provides sufficient context
• Accessibility metadata accurately describes the publication
• Screen-reader output is understandable
• Complex content is accessible

That is why publishers should avoid saying:

“The EPUB passed EPUBCheck, therefore it is accessible.”

A better statement is:

“The EPUB passed technical validation. Accessibility evaluation is the next QA layer.”

What Is Ace by DAISY?

Ace by DAISY is a free, open-source accessibility checker specifically designed for EPUB publications.

DAISY says Ace runs automated accessibility tests on EPUB content and produces reports focused on areas including:

• Accessibility violations
• Metadata
• Document outlines
• Images

As of 2026, Ace remains actively maintained; DAISY released Ace 1.4.5 in April 2026 as a maintenance update covering both the command-line tool and app.

This makes Ace highly useful inside modern EPUB production and remediation workflows.

What Does Ace by DAISY Check?

Ace can automatically identify many accessibility conditions that software can evaluate.

Depending on the publication, this can include issues involving:

• Missing accessibility properties
• HTML accessibility problems
• Image alternatives
• Document structure
• Metadata
• Heading outlines
• EPUB-specific accessibility conditions

Ace's report also provides useful views of:

• Violations
• Accessibility metadata
• Document outline
• Images and their associated descriptions

This gives production teams significantly more accessibility-focused information than EPUBCheck alone.

But Ace Is Still an Automated Tool

This is where many production workflows fail.

A publisher runs Ace.

No critical automated violations remain.

The file is labelled accessible.

That conclusion is too strong.

DAISY explicitly states that passing Ace is not the same as meeting all WCAG accessibility requirements, because only some accessibility conditions can be tested automatically.

Consider an image description:

“Three people standing beside a chart.”

Software can determine whether an image has an alternative-text attribute.

It cannot reliably determine whether that description communicates the information the reader needs from the image.

That requires human judgment.

EPUBCheck vs Ace by DAISY: Side-by-Side Comparison

Easy way to remember it

EPUBCheck asks: “Is the EPUB built correctly?”

Ace asks: “Which accessibility problems can software detect?”

Human QA asks: “Does the publication actually work for the reader?”

Why Publishers Need Both Tools

EPUB production has at least two different quality layers.

Layer 1 — Technical Integrity

The EPUB must be structurally valid.

This is where EPUBCheck belongs.

Without technical integrity, a title can behave unpredictably across reading systems.

Layer 2 — Accessibility

The publication must also expose content in ways readers with disabilities can access.

Ace helps automate part of this assessment.

But accessibility contains many contextual decisions that automated systems cannot fully evaluate.

Therefore:

EPUBCheck without Ace leaves accessibility gaps.

And:

Ace without EPUBCheck leaves technical conformance gaps.

DAISY specifically notes that Ace is not currently a substitute for EPUBCheck.

Where Does Manual Accessibility Testing Fit?

After technical and automated checks comes the layer software cannot reliably replace:

human accessibility evaluation.

DAISY's published validation process identifies manual review as the necessary next stage after EPUBCheck and Ace. Its SMART tool exists specifically to guide evaluators through manual accessibility checks.

Human review is particularly important for questions such as:

• Is the alt text meaningful?
• Is the reading sequence logical?
• Do headings make sense as a hierarchy?
• Can complex tables be understood?
• Is link wording useful?
• Does navigation match reader expectations?
• Are accessibility metadata claims accurate?
• Does the content make sense with a screen reader?

These are not simply code-validation questions.

They are reading-experience questions.

Example: Why All Three QA Layers Matter

Imagine an academic EPUB containing:

• Chapters
• Tables
• Photographs
• Charts
• Footnotes
• Page navigation

EPUBCheck

May identify:

• Invalid markup
• Broken references
• Package problems
• EPUB specification violations

Ace by DAISY

May identify:

• Missing image alternatives
• Accessibility markup problems
• Metadata issues
• Automatable structural accessibility defects

Human QA

Can evaluate:

• Whether chart descriptions explain the important trend
• Whether table relationships remain understandable
• Whether footnote navigation is usable
• Whether reading order is logical
• Whether page navigation works effectively
• Whether screen-reader announcements make sense

That is why one green validation result should never become the entire accessibility strategy.

Automated Testing vs Human Accessibility QA

DAISY gives a simple example: software can detect whether an image has alternative text, but a human must determine whether that text is actually meaningful.

Key Point

Automation tells you whether a condition exists. Human review determines whether it works.

A Better EPUB Validation Workflow

For publishers, accessibility teams and production vendors, the QA process should be repeatable.

Stage 1 — Build the EPUB

Create or remediate:

• Semantic structure
• Navigation
• Images
• Tables
• Notes
• Links
• Language information
• Accessibility metadata

Stage 2 — Run EPUBCheck

Resolve technical EPUB conformance issues.

Do not move unresolved technical failures into later QA unless they are deliberately documented exceptions.

Stage 3 — Run Ace by DAISY

Review the accessibility report.

Investigate:

• Violations
• Metadata
• Outlines
• Image information
• Other automated findings

Stage 4 — Correct the File

Fix confirmed problems.

Then rerun validation.

Do not assume that one correction cannot affect another part of the EPUB.

Stage 5 — Perform Human Accessibility Review

Evaluate the conditions automated tools cannot reliably judge.

Focus on:

• Meaning
• Sequence
• Navigation
• Alternative-text quality
• Complex content
• Tables
• Notes
• Accessibility metadata accuracy

Stage 6 — Test With Assistive Technology

Where required by the project, test representative reading journeys using environments such as:

• NVDA
• JAWS
• VoiceOver
• Keyboard-only navigation
• Appropriate EPUB reading applications

The goal is not simply to hear whether text is spoken.

It is to determine whether the publication remains understandable and navigable.

Stage 7 — Retest the Final EPUB

Run the final file through the agreed validation process again.

The production flow becomes:

Build → EPUBCheck → Ace → Human QA → AT Testing → Correct → Retest → Release

Five Mistakes Publishers Make With EPUB Validation

1. Treating EPUBCheck as an Accessibility Checker

EPUBCheck is essential, but its main purpose is EPUB specification conformance.

2. Treating Ace as Accessibility Certification

Ace assists accessibility evaluation.

A clean automated result does not establish that every accessibility requirement has been met.

3. Skipping Manual Review

Context-heavy accessibility requirements cannot always be reliably automated.

4. Testing Only Before Remediation

Every correction should be followed by another validation cycle.

5. Testing One Chapter Instead of the Publication

Accessibility defects can appear in:

• Navigation
• Metadata
• Spine order
• Cross-document links
• Page lists
• Publication-level structure

QA should therefore evaluate the complete EPUB.

What Should Publishers Require From an EPUB Production Partner?

When outsourcing accessible EPUB production, ask how the vendor handles each QA layer.

Useful questions include:

• Is every final EPUB run through EPUBCheck?
• Is Ace by DAISY included?
• How are automated violations resolved?
• Who evaluates alt-text quality?
• How are complex tables reviewed?
• Is accessibility metadata checked?
• What manual accessibility testing is performed?
• Which assistive technologies are used?
• Are final corrected files retested?
• What validation evidence accompanies delivery?

The supplier should be able to explain:

technical validation + automated accessibility testing + human QA

as separate but connected stages.

Why Validation Evidence Matters at Scale

For one title, a production editor may remember every correction.

For 1,000 or 35,000 titles, memory is not a QA system.

Large-scale EPUB programmes need repeatable evidence such as:

• EPUBCheck results
• Ace reports
• Manual QA findings
• Assistive-technology testing records
• Exception logs
• Retest status

This creates traceability when multiple production specialists, publishers or distribution partners are involved.

How Gentize Approaches Accessible EPUB QA

Gentize's current Accessibility practice publicly lists:

• Accessible EPUB remediation
• WCAG 2.2 AA auditing and remediation
• NVDA, JAWS and VoiceOver testing
• Alt-text and structural tagging
• EPUB and ONIX accessibility metadata
• Accessibility documentation

Gentize describes its process as:

Audit → Remediate → Test + Sign-off → Document + Monitor

and states that its accessibility work combines automated and manual auditing with real assistive-technology testing.

The current Gentize site also publicly states:

• 50+ A11Y-certified testers with visual impairments
• A 35,000-title EPUB 3 backlist accessibility programme
• A 12,000-document accessibility-remediation programme

For publishers, the important point is not simply the size of those programmes.

It is the QA principle behind them:

Automated validation should begin the accessibility process—not end it.

Final Takeaway

EPUBCheck and Ace by DAISY are not competing tools.

They perform different jobs.

EPUBCheck helps determine whether the EPUB conforms technically to EPUB specifications.

Ace by DAISY identifies accessibility problems that can be tested automatically.

Human accessibility QA evaluates the meaning, usability and reading experience that software cannot reliably judge.

The strongest publishing workflow is:

EPUBCheck → Ace → Human QA → Assistive-Technology Testing → Retest

If your publishing team currently stops after receiving a “pass” from one automated tool, there is still another question to ask:

Does the EPUB actually work for the person who needs to read it?