Interactive learning · M3-T3-MLC2

Speech-to-Text, Text-to-Speech, Subtitling and Reading Support

One content, several channels – more ways to learn

Author: Zsolt LengyeliTStudy Hungary Training and Research Centre
Reviewer: Mária HartyányiiTStudy Hungary Training and Research Centre

A video explanation with a presenter and a caption bar displaying speech as text. — Captions make the spoken explanation in a video readable.

Learning objective

Converting speech into text and text into speech are among the most mature and useful applications of AI in education. Speech-to-text, text-to-speech, subtitling and reading-support tools can make learning materials more accessible and usable in everyday teaching.

Explore how these tools improve access for learners with different sensory, linguistic, literacy or concentration needs, while giving everyone in the group a choice of formats.

By the end of this course, you will be able to

choose suitable speech and reading tools for a learning situation;
create checked captions for videos, audio versions of texts and transcripts of explanations;
prepare simplified texts and glossaries, then check their accuracy;
offer the same content through several channels while protecting personal data.

One content, several channels

Who might find a single format difficult?

Most learning content reaches learners as printed text, a spoken explanation or a video. Each channel can create a barrier: a learner may not hear the explanation clearly, may struggle to read dense text quickly enough, or may lose concentration halfway through.

Several ways to convey the same meaning

AI-based speech and reading tools let learners listen to text, read spoken explanations and watch captioned video. The aim is to offer several accessible ways to engage with the same content.

01

Text

An explanation that can be re-read, searched and translated.

02

Audio

Content to listen to alongside reading or as an alternative.

03

Captioned video

A demonstration whose explanation can be followed without sound.

The same content, a choice of formats.Offer support to the whole group.

Speech-to-text: dictation and transcription

Speech recognition turns spoken explanations into editable text. It is also known as speech-to-text.

Dictation allows learners with writing difficulties or motor impairments to express their knowledge by speaking.

A transcript of an explanation can be re-read, searched and translated. This is especially useful for learners who are hard of hearing, were absent or do not learn in their native language.

A speaker, microphone and sound waves; speech is converted into editable text. — Speech-to-text turns spoken explanations into editable text.

Dictation on everyday devices

Examples in the source include Windows dictation (Win + H), macOS, Android and iOS keyboards, Google Docs, Word, and speech input in AI assistants such as ChatGPT.

Try a short paragraph from your subject. Check which languages your device supports.

Transcribing live speech and recordings

On Android, use Live Transcribe; in Word, use Transcribe. Meeting transcripts in Teams and Google Meet also support learning. Online services offer transcription too.

Availability and terms of use may vary by device, account and language.

Text-to-speech: audio versions of learning materials

Text-to-speech converts written text into an audio version. Its abbreviation is TTS.

Modern neural voices sound increasingly natural and are available in many languages, including Hungarian. Test the voice and the pronunciation of technical terms with your own learning material.

Audio versions support learners with dyslexia or visual impairment, learners reading in a second language, and anyone who prefers listening, for example while travelling to a work placement.

A model generates audio from a written document, which a learner listens to through headphones. — Text-to-speech turns text into an audio version.

01

On a computer

Read Aloud in Microsoft Edge and Immersive Reader in Word and OneNote read content aloud. Highlighting words as they are read can help learners follow the text.

02

On a phone

Android accessibility (Select to Speak) and iOS accessibility (Speak Screen): both read the text on the screen aloud.

03

Creating narration

The source gives ElevenLabs as an example of a voice-generation service for narration. Listen to the entire recording and check it before sharing.

Captions: follow a video without sound

Captions can be essential for deaf and hard-of-hearing learners. They also help in noisy workshops, when learning in a second language, and when learners need to see how a technical term is spelled.

YouTube, Microsoft Stream and other video tools can generate automatic captions. Some operating systems and browsers also offer live captions.

Automatic captions are a starting point. Technical terms, numbers and safety warnings must be checked and corrected.

Give learners control

Prefer closed captions that learners can switch on and off and resize, rather than captions burned into the image. Automatic translation can support second-language learners; check important technical content after translation too.

Try it without sound!

Watch the short demonstration with the sound muted. Can you follow the explanation through the captions? Are important sound cues and changes of speaker clear? Does the text obscure any important visual details?

Reading support: less clutter, more meaning

Reading-support tools reduce the effort of decoding text so that learners can focus on meaning.

Immersive Reader and browser reading views can remove distracting elements. Changing font size, spacing, background colour and line focus can make reading more comfortable.

Some functions split words into syllables or show pictures for common words. Check which options are available in the language and application you use.

Dense text becomes a clearer reading view that highlights the current line and offers read-aloud support. — Reading modes reduce visual clutter and help learners follow the text.

Simplification and summaries

A language model can rewrite a text at a lower reading level, split it into short sections, and add headings and a summary. Keep the technical terms and explain them.

A glossary for every unit

Simple explanations and pictures of key terms help the whole group, especially learners reading in a second language. The teacher checks the accuracy of the terminology.

Recipe: one worksheet, several channels

Practical workflow · about 20 minutes

Five steps to more accessible material

Choose one of your own worksheets and a short demonstration video. Tick each step when you have completed it.

1. Check the structure!Use clear headings, short paragraphs and readable lists instead of long sentences.2. Create an audio version!Use Read Aloud or a text-to-speech tool and check the pronunciation.3. Simplify and explain!Ask for a simplified version and a glossary of 8–10 key terms, then check both.4. Caption the demonstration!Add captions to the video, then correct technical terms, numbers and safety warnings.5. Offer choice!Make all versions available to the whole group so learners can choose.

The goal

The same content, more choices

The text, audio version, simplified material, glossary and captioned video all serve the same learning objective.

Share the versions with the whole group after checking their accuracy. Let learners decide which best supports their understanding.

Check meaning as well as spelling!

The teacher is responsible for shared content

AI transcripts, captions and simplifications can contain errors. Always check important technical words, numbers, conditions and safety information.

Background noise, overlapping speech, dialects and specialised terminology can increase errors. Clear, structured speech helps both people and machines.

A short word can change the meaning

An omitted negative or a misrecognised number can reverse an important technical instruction. Always compare the automatic text with the original explanation.

A fictional example for illustration

What would you do with this automatic caption?

Spoken sentence

“You must not remove the protective guard.”

Automatic caption

“You must remove the protective guard.”

Choose a step to see the explanation.

Data protection, choice and teacher support

01

Recording and consent

Make audio or video recordings only with the consent of those involved. Delete them after the exercise if they are no longer needed.

02

Accounts and data handling

Use school accounts or approved tools for transcription. Follow data protection rules when processing content in the cloud, and do not upload personal data.

03

A choice of support

Support tools give access to content while reading and writing skills develop. Teaching those skills remains the teacher’s responsibility.

Consider how sensitive material is processed

Some services process recordings and documents in the cloud. Tools that run locally or appropriately configured institutional accounts can reduce data-handling risks. Always choose a solution approved by your school.

Let learners choose what helps them

Offer choice rather than prescribing a format. Learners know best which format helps them. Make audio, captioned and simplified versions available to everyone.

Technology gives access; the teacher supports learning.

Takeaways for teachers

Five ideas for your next lesson

Offer the same content through several channels: text, audio and captioned video.
Dictation and transcription help learners who find writing or listening difficult.
Natural-sounding voices make audio versions of texts quick to produce. Check the voice and the pronunciation of technical terms.
Always check automatic captions and transcripts, especially technical terms and safety information.
Reading modes, simplified versions and glossaries reduce the effort of reading and support understanding.

Teacher exercise: one lesson, several channels

40 minutes · Individual work

Transform one of your own lesson materials into an audio version, a simplified version with a glossary, and a captioned video. Evaluate the quality of the AI outputs as you work.

Exercise steps

Choose a worksheet and a short, 1–2 minute demonstration video from your subject.
Create an audio version with Read Aloud and listen to it: are the technical terms pronounced correctly?
Ask the AI for a simplified version at B1 level, using short sentences and keeping and explaining technical terms. Also request a glossary of 8 key terms. Check both for accuracy.
Generate automatic captions for the video. Correct every error in technical terms, numbers and safety warnings.

Tools and lessons learnt

Edge Read Aloud or Immersive Reader; ChatGPT, Claude or Gemini; YouTube Studio or another tool with automatic captions.

What can you take away?

Producing several formats takes little time once the workflow is familiar.
Automatic captions and simplifications save time but always need expert checking.
Offering choices helps many learners, not only those with identified needs.

Download detailed instructions for both exercisesM3-T3-MLC2-EXC-EN.pdf

Classroom exercise: caption crew

45 minutes · Pair work

Learners record a short explanation of a workplace procedure, transcribe it automatically and correct the captions. The aim is to develop awareness of accessible communication and precision with technical terms.

Steps for pair work

Pairs choose a short procedure from their trade and plan a 60-second spoken explanation.
One learner explains while the other records with consent. Use an approved tool to transcribe or caption the recording automatically.
Pairs compare the transcript with what was said and correct every error, especially technical terms.
Pairs present interesting recognition errors and discuss how to speak more clearly for captioning.

Tools and lessons learnt

School tablets or phones, a transcription or captioning tool, and the textbook for checking terms.

Checking recognition errors is part of professional communication.
Clear, structured speech helps both people and machines.
Captions make explanations accessible, including for deaf and hard-of-hearing colleagues.

Notes for teachers

Ask for consent before recording. Delete the recording after the lesson if it is no longer needed. Use school accounts or tools approved by the school for transcription.

Check what you have learnt!

Each question has one correct answer. Answer all seven questions, then view your result and the explanations.

1Which tool helps a learner write down their ideas by speaking? Text-to-speech. Dictation, or speech-to-text. Only changing the video playback speed.

2Why is a transcript of a spoken explanation useful? It can be re-read, searched and translated. Because it is guaranteed to contain no recognition errors. Because it makes a single format compulsory for all learners.

3Which type of captions should you prefer? Captions that are always burned into the image. Captions that obscure important visual details. Captions that can be switched on and off and resized.

4What should you do before sharing automatic captions? Check spelling only. Compare them with the speech and correct technical terms, numbers and warnings. Accept them unchanged if they read fluently.

5What should an AI-simplified text preserve? Only the length of the text. Only easy, everyday words. Technical meaning, specialised terms, conditions and warnings.

6Who should be offered audio and simplified versions? The whole group, as a choice. Only learners chosen in advance by the teacher. No one until everyone learns in the same way.

7How should recordings be made in the Caption crew exercise? Without considering consent or school rules. With consent and an approved tool, deleting unnecessary recordings afterwards. By uploading learners’ personal data to any service.

References and further reading

References in the source document

W3C (2023): Web Content Accessibility Guidelines (WCAG) 2.2 – Success Criterion 1.2.2 Captions (Prerecorded).
CAST (2024): Universal Design for Learning Guidelines version 3.0.
Microsoft Support: Immersive Reader and Read Aloud – languages and voices.
Lengyel, Zs. (2026): Concrete applications for content generation. VITA training presentation.

Useful links

Language support, features and terms of use may change. Testing a tool on your own device with your own learning material shows what works in a particular learning situation.

According to the source, images without a specified source were generated by Claude for Mac. External links open in a new tab and require an internet connection.

Credits and copyright

Author: Zsolt LengyeliTStudy Hungary Training and Research Centre
Reviewer: Mária HartyányiiTStudy Hungary Training and Research Centre
Project: 2025-1-LT01-KA220-VET-000362456
Learning material: M3-T3-MLC2

Attachments

Source material · PDFM3-T3-MLC2-TEXT-EN.pdf Practical exercises · PDFM3-T3-MLC2-EXC-EN.pdf

This work may be reused and adapted under the terms of the Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA 4.0).

Funded by the European Union. Views and opinions expressed are those of the author(s) only and do not necessarily reflect those of the European Union or the National Agency.