Audio Course Materials for Neurodivergent Students: A Guide for Universities

Universities across the UK already convert course documents into audio. What almost none of them produce is audio anyone would choose to listen to. The automated conversion built into virtual learning environments generates a single synthetic voice reading one file aloud, and institutions say so plainly in their own guidance. This article sets out what the statutory duties actually require, what the published evidence does and does not show about audio for neurodivergent students, where existing provision leaves a gap, and how a university team can scope a pilot of conversational audio course material without committing to anything first.

What are alternative formats, and where does audio already fit?

Alternative formats are versions of course material produced in a form other than the original: tagged PDF, HTML, EPUB, DAISY, electronic braille, enlarged print, and audio. In UK higher education they are usually delivered through two routes at once.

The first route is automatic conversion inside the virtual learning environment. UWE Bristol explains that “Blackboard automatically creates a number of alternative formats based on the original content”, audio among them, “playable on any devices that can play MP3 files”, and that “all students are encouraged to explore and use the alternative formats”. The University of York and the University of Reading describe the same MP3 output on their own systems.

The second route is a self-service or staffed conversion service. The Bodleian Libraries at Oxford convert documents “into a wide range of alternative formats: including audio books (MP3 and DAISY)”. The University of Glasgow tells students that the same service “allows you to convert lecture notes, articles and slides into alternative formats, including MP3s and e-books”.

So audio provision is not a new idea in the sector. It is an established, widely deployed capability. The question worth asking is what kind of audio it produces.

What does the Equality Act 2010 actually require?

It requires accessible formats. It does not name audio, and any supplier who tells you otherwise is overstating the law.

Section 91(9) of the Equality Act 2010 applies the duty to make reasonable adjustments to the responsible body of a higher education institution, and Schedule 13 applies the first, second and third requirements to those institutions. The third requirement, in section 20(5), covers auxiliary aids. Section 20(6) then adds that where the first or third requirement relates to the provision of information, “the steps which it is reasonable for A to have to take include steps for ensuring that in the circumstances concerned the information is provided in an accessible format”.

The planning consequence comes from how that duty is timed. The EHRC technical guidance on further and higher education states that the duty “is anticipatory in the sense that it requires consideration of, and action in relation to, barriers that impede all disabled people prior to an individual disabled student seeking to access education”, that providers “should therefore not wait until a disabled person approaches them”, and that it applies “regardless of whether the education provider knows that a particular person is disabled”. The same guidance notes the duty relates to “all disabled students and any disabled applicants, not just those entitled to Disabled Students’ Allowances”.

Leeds Beckett University puts the same point in its own words in its staff guide: the university has “a legal and moral responsibility under the Equality Act 2010 to provide all information in an alternative format if requested by a disabled person”, and “a duty to anticipate the needs of disabled people by ensuring all information is produced in the most accessible way possible”.

For web content specifically, UK universities are in scope of the Public Sector Bodies (Websites and Mobile Applications) Accessibility Regulations 2018 — the carve-out for schools and nurseries does not reach higher education. Which version of WCAG applies is worth stating precisely, because the two authorities do not currently say the same thing: the GOV.UK guidance points at WCAG 2.2 AA, while EN 301 549, the harmonised standard the regime leans on, reproduces WCAG 2.1 AA verbatim. Plan against 2.2 and know why the instrument says 2.1. The same GOV.UK guidance anticipates that “somebody might ask for information in an alternative, accessible format, like large print or an audio recording”. EN 301 549 is also the EU baseline, specifying requirements “in a form that is suitable for use in public procurement within Europe” and underpinning Directive (EU) 2016/2102.

In the United States the position splits by institution type, and this is routinely got wrong. The Department of Justice’s 2024 web accessibility rule sets WCAG 2.1 Level AA under ADA Title II, which covers public universities only. Its compliance dates were then extended by an interim final rule published on 20 April 2026 (91 FR 20902), to 26 April 2027 for entities serving 50,000 or more people and 26 April 2028 for smaller entities and special districts — so any source still quoting the original 2026 date is stale. Private universities fall under Title III, which has no adopted web technical standard at all.

One clarification worth making, because it is widely got wrong: WCAG’s “text alternatives” guideline requires text for non-text content, not audio for text. It runs in the opposite direction. The relevant W3C material is technique G79, “Providing a spoken version of the text”, a sufficient technique for the Reading Level criterion, which helps “users who have difficulty sounding out (decoding) words in written text”. That criterion sits at Level AAA, so no regulation requires it. It is a recognised technique, not an obligation.

A second point saves an argument later. WCAG’s prerecorded audio-only criterion normally requires a text alternative for audio content, but it carries an explicit exception “when the audio or video is a media alternative for text and is clearly labeled as such” — a media alternative for text being media that “presents no more information than is already presented in text”. A clearly labelled audio version of a course pack that adds nothing beyond the source document therefore does not need a separate transcript, because the document already is the transcript.

The flip side of that is not optional, and it is the sentence to keep in mind through the rest of this article: audio is an additional modality, never a substitute for accessible text. Audio alone excludes deaf and hard-of-hearing students. The accessible source document has to exist and stay available regardless of what else you offer alongside it.

Is there evidence that audio helps neurodivergent students?

Yes — but the honest headline is parity and access, not superiority. The evidence supports audio as an equally viable route into the same content, which is a sufficient reason to offer it and a far more defensible position internally than a claim that listening beats reading.

The largest comparison is the meta-analysis by Clinton-Lisell (2022) in Review of Educational Research, which pooled reading-versus-listening comprehension studies and found the two broadly comparable, with a small advantage to reading under some conditions. Audio is not a shortcut to better comprehension. It is a second door into the same room.

The study closest to this specific population is a pre-registered randomised controlled trial by Jafarian and Kramer (2025), published in Computers and Education: Artificial Intelligence. It reports that “the audio-learning modules increased student motivation and reading engagement”, that those increases “boosted academic achievement”, and — directly relevant here — that “students with greater ADHD symptom severity particularly benefited from the audio-learning modules, as they played a crucial role in determining course success”. That is a single trial, and it speaks to ADHD symptom severity rather than to every neurodivergent profile.

The broader literature on text-to-speech is more cautious. The meta-analysis by Wood, Moxley, Tighe and Wagner (2018) found a small-to-moderate average effect on reading comprehension for students with reading difficulties, concluding that these technologies “may assist students with reading comprehension” while “more studies are needed to further explore the moderating variables”. The authors also report that “moderator effects of study design were found to explain some of the variance”, the included studies predate modern speech synthesis, and the population studied was students with reading disabilities — not ADHD, autism or dyspraxia.

For the institutional framing rather than the effect size, the systematic review by Gunderson and Cumming (2023) in Innovations in Education and Teaching International examines podcasting in higher education specifically as a component of Universal Design for Learning — which is the argument this article is making, made by authors with nothing to sell.

Three honest limits follow:

• No published body of evidence demonstrates improved outcomes from audio course material across every neurodivergent profile. Anyone claiming otherwise is extrapolating.
• The Office for National Statistics confirms it does not hold clinical or diagnostic records for neurodivergent conditions and does not capture dyslexia, dyspraxia or other specific learning differences in its datasets, so widely circulated national prevalence figures should be treated with care.
• Audio is best understood as compensatory — it helps a student access material now — rather than as an intervention that builds reading skill over time.

The case for offering audio does not rest on it being better. It rests on it being comparable for comprehension, available in moments when reading is not, and cheaper in effort for students who find decoding text costly.

Why read-along matters more than audio alone

A common instinct is to treat audio as a replacement for reading. The cognitive-load literature does not support that framing. The redundancy principle described by Mayer and Moreno (2003) warns specifically against duplicating on-screen text as speech, and the modality effect concerns narration paired with graphics — not audio-only versus text-only. Citing “dual coding” to argue that listening should replace reading misrepresents the research.

What the accessibility guidance actually describes is audio alongside text. W3C’s Making Content Usable for People with Cognitive and Learning Disabilities records the user need directly: “I need text-to-speech support, with synchronized highlighting, so I can follow along as words are read aloud.” W3C’s perspective on text-to-speech names “people with dyslexia and other cognitive and learning disabilities who need to hear and see the text to better understand it”, and its guidance on cognitive and learning disabilities covers dyslexia, ADHD and autism spectrum disorder explicitly.

CAST’s Universal Design for Learning Guidelines make the design case rather than the empirical one, and the specific reference matters. Version 3.0, released in 2024, retired the old checkpoint numbering; the relevant guidance is Consideration 2.2, “Support decoding of text, mathematical notation, and symbols”, which suggests “Allow the use of text-to-speech” and “Use digital text with an accompanying human voice recording”. That is the decoding consideration, not the perception one — the perception guidance runs the other way, toward text alternatives to audio, so citing it to mean “UDL says provide audio versions of text” is a misreading that a disability-services reader will catch. UDL is a design framework rather than evidence of outcomes, and it is worth citing as such. (Cite as CAST (2024), CAST Universal Design for Learning Guidelines version 3.0.)

For dyspraxia and developmental coordination disorder, the reported barrier is often production rather than reception. A 2024 study in the European Journal of Investigation in Health, Psychology and Education found handwriting to be a considerable difficulty for over half of its participants with DCD, with students describing losing the thread of a lecture while concentrating on writing notes. Material available in a form that does not have to be transcribed in real time addresses that directly. (Note that the Dyspraxia Foundation closed in 2024; the British Dyslexia Association confirmed the closure, and Movement Matters is a current UK point of reference.)

Where does current provision fall short?

Three gaps are visible in universities’ own published pages.

The gated route is slow, and the fast route is unchecked. UWE Bristol runs both. Its alternative formats service is open to “any UWE Bristol student who has a visual impairment, is neurodivergent or has any other condition that makes it difficult to access text”, and states that it aims “to produce accessible PDFs within three weeks of receiving a request”, with “other formats, such as Word and Audio” taking longer. Its automated route returns audio immediately to everyone. A student choosing between those is choosing between a considered format that arrives after the topic has been taught, and an instant one that, as City St George’s, University of London says of automated formats generally, is “not checked by any staff, such as lecturers” and should be treated as “a supplementary resource”.

The onus still sits with the student. The Office for Students’ review Beyond the bare minimum found that “the onus remains on disabled students to ask for inclusive teaching practices, rather than such practices being the norm”, and reported that while the vast majority of surveyed providers recorded some of their lectures, only a small minority recorded all of them. That review was published in 2019 on 2017 data, so treat the specifics as historical — but the structural point it makes is the same one the anticipatory duty makes.

Students have already said what they want. Open SU, the Open University Students Association, has published a policy position asking institutions to “provide materials in multiple formats wherever possible (text, audio, video, printable)” and to ensure materials are “designed with neurodivergent students in mind by default, rather than relying on disclosure or delayed adjustments”. That is a written, public, unsolicited statement of the requirement from the student side.

One terminology note, since it signals whether a supplier has read the sector: use neurodivergent for individuals. As the University of Glasgow’s neurodiversity resource hub sets out, “neurodiverse” describes a group spanning the spectrum of neurodiversity, while neurodivergent describes individuals. “Neurodiverse students” is the wrong construction.

What does conversational audio change?

The distinction is between a document read aloud and a subject explained.

Automated conversion takes one file and narrates it in one synthetic voice, in the order the file happens to be written. Podhoc restructures material into a discussion between voices, which changes what listening feels like and what it is useful for. Concretely, and verifiable on this site:

• Source material in .pdf, .docx, .doc and .txt, as well as pasted text and links, becomes a multi-voice episode rather than a single-voice narration.
• Eight audio styles — including Feynman Technique, Critique, Debate and Simplified Explanation — determine how the material is treated, so a dense methods paper and an introductory reading list do not receive the same handling.
• Read-along highlights words and the current sentence in sync with the audio, which is the pattern W3C’s cognitive-accessibility guidance describes, rather than audio replacing text.
• Playback speed controls let a listener slow difficult passages down or move quickly through familiar ones.
• Output is available in 73 languages, which matters for institutions teaching international cohorts.
• A REST API with token authentication allows conversion to be driven from existing systems instead of by hand.

Which of these capabilities matter for a given deployment, and how each is configured, is exactly what a pilot establishes — the list above describes the platform, not a preset institutional configuration.

What it does not do is remove the need for a considered accessible PDF, a braille transcription, or a human-checked format where the material demands one. It is one more format in the alternative-formats toolkit, and it should be evaluated as such.

You can hear the underlying approach on the pages for students and researchers, and the reasoning behind the format is set out in why audio learning works.

What should a university ask any supplier before running a pilot?

These questions apply to every vendor in this space, including us. They are the questions a data protection impact assessment will ask anyway, so it is faster to ask them at the start.

The first row is the one to settle before any student coursework is processed. Publication is on by default on the Podhoc platform, and that default is appropriate for an individual sharing something they made. It is not automatically appropriate for institutional use, and it is a configuration question rather than an assumption.

How can a university pilot audio course materials?

Keep the first pass small enough that it does not need a business case.

1. Pick one module with a heavy reading load and a cohort that already generates alternative-format requests.
2. Choose a small set of representative documents — a core reading, a lecture handout, a methods chapter — rather than a whole module.
3. Agree the publication and data-handling settings in writing before anything is processed.
4. Decide in advance what you will measure. Engagement with the material and student feedback are realistic within a term; attainment effects are not, and a pilot that promises them will disappoint.
5. Compare the output honestly against what your virtual learning environment already generates automatically. If it is not meaningfully better, that is a useful result too.

This mirrors how UK institutions already trial assistive technology — as a loan or a trial before any commitment — so it should fit an existing process rather than requiring a new one.

Request a pilot for your institution, company or employer →

Related reading

• Why audio learning works — the research behind learning by listening.
• University e-learning audio partnerships — how audio conversion fits alongside an existing VLE and e-learning estate.
• Audio learning for training providers — the same argument for professional and vocational training.
• The best passive learning tool — layering audio over time already spent on other things.
• Podhoc for students — the student-facing workflow.
• Podhoc for researchers — reading-heavy academic workflows.
• The eight audio styles — matching the treatment to the source material.