March 14, 20269 min read

Creating Subtitles: A Step-by-Step Guide to Accessible Videos

Since 28 June 2025, the Accessibility Strengthening Act (BFSG) has been in force in Germany. In Austria, the Accessibility Act (BaFG) implements the same EU directive. Companies that direct digital products or services at consumers must make their content accessible – including videos on their own website. Subtitles are one of the central requirements here.

Even so, many companies have not yet made the switch. Often it is not the will that is lacking, but the knowledge: which subtitles are “accessible”? Which format do I need? And how do I create them efficiently, without editing every video by hand?

This guide walks you through the entire process – from the legal requirements to the finished subtitle file.

What the BFSG and BaFG mean for videos

In Germany, the BFSG has obliged companies with B2C offerings to make videos accessible since June 2025. In Austria, the BaFG implements the same EU directive. In concrete terms this means: subtitles for recorded videos, audio description for purely visual content, and transcripts as a full written version. The technical standard is WCAG 2.1 conformance level AA.

The Accessibility Strengthening Act transposes the European directive (EU) 2019/882 (the European Accessibility Act) into German law. In Austria, this is done through the Accessibility Act (BaFG). For companies, then, the abbreviation matters less than the practical implementation: digital services must be usable by people with disabilities. The technical standard behind it is WCAG 2.1 conformance level AA.

For videos, this means three obligations in concrete terms:

Subtitles for all recorded videos with spoken content (WCAG 1.2.2)
Audio description for visual content that cannot be inferred from the sound (WCAG 1.2.3)
Transcripts as a full written version of the spoken content (WCAG 1.2.1)

Affected are B2C offerings – that is, websites, shops and apps aimed at consumers. Small businesses with fewer than ten employees and a maximum annual turnover of two million euros are exempt. For everyone else, violations carry the threat of fines. For Austrian companies, the same substantive requirements apply via the BaFG.

Standard subtitles vs. accessible subtitles

Standard subtitles only reproduce the spoken text. Accessible subtitles (SDH) additionally describe relevant sounds, music, speaker changes and the emotional tone. Only accessible subtitles meet the WCAG requirements and thus the requirements for accessible video offerings.

Not every subtitle meets the requirements for accessible video offerings under the BFSG or BaFG. Platforms like YouTube generate automatic subtitles – these are generally not sufficient for legal compliance. The difference lies in the detail:

Standard subtitles only reproduce the spoken text. They help with low sound or a foreign language, but are incomplete for deaf people.
Accessible subtitles (SDH) additionally describe relevant sounds ([doorbell], [applause]), music ([suspenseful music]), speaker changes and the emotional tone. They make the entire acoustic context accessible.

WCAG requires accessible subtitles. That means: speaker identification, synchronous display, description of non-speech audio content and sufficient reading time per display.

Creating subtitles: step by step

The most efficient path leads through an AI-assisted transcription as the basis, which is then adapted for subtitle purposes. The workflow consists of five steps:

1. Prepare the audio: Extract the soundtrack from your video or upload the video file directly into a transcription tool. Common formats like MP4, MOV, MP3, WAV and M4A are supported by most services. The better the audio quality, the more accurate the transcription – background noise and reverberation significantly reduce the recognition rate.
2. Have the transcription created: Modern AI models transcribe audio in a few minutes with over 95 % accuracy for standard languages. Make sure the tool supports speaker diarization – for videos with several people, identifying “who is speaking” is mandatory for accessible subtitles. The result is a time-stamped transcript with speaker changes.
3. Post-process the transcript: Check the AI transcript for errors, especially with proper names, technical terms and abbreviations. Add descriptions for relevant sounds and music in square brackets. Make sure each subtitle block comprises no more than two lines and stays on screen long enough (rule of thumb: at least one second per five words).
4. Export the subtitle file: Export the edited transcript as an SRT or VTT file. Both formats contain timestamps and text and are supported by all common video platforms and players. Which format you choose depends on the intended use (see the next section).
5. Embed the subtitles: Upload the subtitle file to your video platform or burn the subtitles permanently into the video. For websites, the toggleable variant (closed captions) is recommended, so that users can activate the subtitles themselves.

SRT, VTT or burned-in: which format when

The choice of subtitle format depends on where and how you use the video:

SRT (SubRip): The industry standard. Supported by YouTube, Vimeo, LinkedIn, WordPress and almost all video editing programs. Contains a sequence number, start/end time and text. Ideal for most use cases.
VTT (WebVTT): The web standard for HTML5 video. It additionally supports styling options such as font color, positioning and CSS classes. Recommended if you embed videos in your own player on your website and want to control the presentation.
Burned-in (open captions): The subtitles are rendered permanently into the video and are always visible. Necessary for social media videos (Instagram Reels, TikTok), since these platforms do not support toggleable subtitles. Drawback: subsequent corrections require a new render.

For accessible video offerings on your own website, SRT or VTT as toggleable subtitles is recommended. For social media, burning in is the only option.

Common mistakes with subtitles

Even well-intentioned subtitles fail the accessibility requirements when typical mistakes creep in:

Missing speaker identification: For videos with several people, it must be clear who is speaking. Without identification, discussions, interviews and dialogues cannot be followed.
Display too fast: Subtitles that are visible for only half a second cannot be read. WCAG recommends a maximum of 20 characters per second.
Insufficient contrast: White text on a light background is illegible. WCAG requires a contrast ratio of at least 4.5:1. A semi-transparent black bar behind the text solves the problem.
No sound descriptions: If an alarm sounds or someone laughs and that is relevant to the context, it must be described – otherwise deaf viewers lack a crucial piece of information.
Automatic subtitles without correction: AI-generated subtitles contain errors. Proper names, technical terms and numbers are regularly misrecognized. Without human review, the subtitles are not reliable.

Data protection when creating subtitles

Corporate videos often contain confidential information – product announcements, internal training, customer testimonials or employee interviews. Anyone who hands the audio track over to an external service for transcription should check where the data is processed and stored.

The GDPR sets clear requirements for processing: a data processing agreement, a processing location within the EU and adequate technical safeguards. For confidential content in particular, client-side encryption offers the strongest protection – the transcription service never sees the plain-text content at any point.

Checklist: accessible subtitles under the BFSG/BaFG

All videos with spoken content have subtitles
Subtitles reproduce the spoken text completely and correctly
Speaker changes are identified
Relevant sounds and music are described in square brackets
Subtitles are synchronous with the audio (no noticeable offset)
Display duration is sufficient (at least 1 second per 5 words)
The text is legible (adequate size, contrast at least 4.5:1)
Subtitles are available as a toggleable SRT/VTT file or are permanently burned in
Automatic transcription has been manually reviewed and corrected
Data protection is ensured during transcription (DPA, EU processing, encryption)

Conclusion

The BFSG and BaFG make accessible subtitles a genuine compliance topic for many corporate videos – and that is more than a bureaucratic hurdle. Accessible videos reach a wider audience, improve the user experience and strengthen findability in search engines. The most efficient path leads through AI-assisted transcription with subsequent manual review and export as an SRT or VTT file. What matters is that the subtitles do not merely reproduce the text, but make the entire acoustic context – speakers, sounds, mood – accessible to everyone.

Note: This article serves general information purposes and is no substitute for legal advice in individual cases.