echoflow for video accessibility
The European Accessibility Act didn't make subtitling a video hard: it made subtitling three hundred of them by hand impossible. That part we automate — and we say precisely which part stays yours.
14 days · 60 minutes · 10 videos · no card
- Synchronised subtitles in SRT and VTT, or burned into the video, from the same upload
- Editable transcript and subtitles: correct the text and its download updates
- Built for the archive, not the single video: the real cost is volume
- Storage in the EU; transcription and AI also outside the EU, sub-processors named one by one
- No declaration of compliance: we can't give one, and no tool can
- Review, speaker identification and publishing stay your work
⚠️ This page describes what a tool does. It is not legal advice and it does not establish whether an obligation applies to you: that answer is built with your own advisor against the text of the law, not on a product page.
Subtitling a video was never the hard part. What changed with the European Accessibility Act — Directive (EU) 2019/882, applicable since June 2025 — is scale: for many organisations the requirement isn't about the next video, it's about every one already published.
Three hundred videos are three hundred transcripts. That's where the problem stops being technical and becomes a matter of budget and calendar.
What the tool does
- Transcribes the speech and generates synchronised subtitles, in SRT and VTT, from the same
upload.
- Produces, once the job is done, the video with subtitles burned in for the channels where a
separate file can't be uploaded.
- Leaves the text editable: correct a proper noun in the subtitles and the SRT and VTT files
update. Transcript and subtitles are corrected separately, and texts already generated don't change on their own.
- Works at volume, which is the whole point: the cost is measured in minutes of content, not in
hours of somebody's time.
- Returns the rest as well — summaries, chapters, texts — because it comes out of the same transcript
and costs no second pass.
What stays with the organisation
This is the part the page exists for, and we write it out rather than leave it implied:
- The review. The accuracy the guidelines call for is guaranteed by no automatic system: results
shift with audio quality, overlapping voices, accents and domain terminology. The time you save is the transcription, not the proofreading.
- Speaker identification, which echoflow does not do.
- Relevant non-speech sounds, which a transcript of speech does not contain.
- Audio description, which is a separate requirement and which we do not produce.
- Publishing to your channels, with subtitles correctly associated to the video.
- Final verification and the accessibility statement, which concern what you publish.
Why we don't say "compliant"
Because we can't, and because a word like that on a sales page is a handle against whoever writes it.
Compliance depends on the accuracy of the text, on speaker identification, on non-speech sounds and on how the content is published: it concerns the published content, not the tool that produced it. The European technical reference — standard EN 301 549, which points to the WCAG guidelines — describes requirements on the result, not a badge a tool can issue.
No automatic generator can guarantee it, ours included. Anyone promising otherwise is describing a product that doesn't exist, and on the day someone checks, they won't be the one answering for it.
How to approach an archive, in practice
Not by starting with the oldest video and working forward.
Start with a dozen representative pieces — one with clean audio, one badly recorded, one with several voices, one dense with technical terms. The point is to measure the only variable that matters for a quote: how much review will actually be needed. On a uniform archive it can be very little; on multi-voice interviews it can be a lot, and that's worth knowing before a budget is committed.
After that you work in batches, giving priority to the videos that sit in front of a service — the ones a user cannot avoid — rather than the ones with the most views.
Where your data is
Files and content stored in the European Union. Transcription and AI generation also use non-EU providers, under the safeguards of Article 44 and following of the GDPR. They are listed by name on the public sub-processors page, which can be read before you sign anything — which, when your privacy advisor asks for the list, saves you a round of emails.
The source file you upload is deleted within 24 hours of processing; on the Pro plan retention can be governed with a configurable workspace policy. Generated media files stay downloadable for 7 days; text outputs — transcript and subtitles included — don't expire while the account stays active.
The trial runs 14 days, includes 60 minutes and asks for no card. The right way to use it is to upload your hardest videos, not your easiest: it's the only way to know how much review work will really be left to you.
Frequently asked questions
Does the European Accessibility Act apply to my organisation?
We can't tell you, and be wary of anyone who tells you from a sales page. Directive (EU) 2019/882 has applied since June 2025 to a defined list of products and services — among others e-commerce, consumer banking, e-books, passenger transport and audiovisual media services — not to everyone who publishes video, and it carries exemptions, including one for microenterprises providing services. If a service on that list depends on your videos, the answer has to be worked out with your legal advisor against your country's transposing law, not here.
Are automatic subtitles enough to be compliant?
No, and anyone telling you otherwise is selling you a reassurance they can't provide. The technical guidelines call for accurate subtitles: they have to render speech correctly, identify who is speaking where that matters, and flag relevant non-speech sounds. An automatic transcript that has been read and corrected can meet that; one published unchecked typically doesn't. Compliance concerns the content you publish, not the tool that produces it.
Do you produce audio description?
No. Audio description — the track that tells someone who can't see what is happening on screen — is a requirement distinct from subtitles, and echoflow does not produce it. If your obligation includes it, that part has to be covered elsewhere: we'd rather tell you now than have you find out halfway through an archive.
Do the subtitles say who is speaking?
No. In a multi-voice conversation the text renders the speech but doesn't mark the speaker, and when voices overlap the transcript makes more mistakes. If your content is panels or multi-person interviews, test it on a real recording before planning the whole archive.
Which subtitle formats do you produce?
SRT and VTT, synchronised and downloadable, plus the video with subtitles burned in, which you can generate once the job is done. Subtitle files are text and don't expire while the account stays active; the subtitled video is a media file and stays downloadable for 7 days from creation.
Where are our videos processed?
Files and content are stored in the European Union; transcription and AI generation also use non-EU providers, under the safeguards of Article 44 and following of the GDPR. They are named one by one on the public sub-processors page, which you can read before you sign anything.
Can we use it on an archive of hundreds of videos?
That's the case it's built for. Consumption is measured in minutes (one minute of video, one minute of balance), so the arithmetic runs on the archive's total duration; for workspace-scale volumes, with configurable retention and separate spaces, the right size is Pro with 1,000 minutes. Our advice is still to start with a dozen representative videos — good audio, difficult audio, several voices — to measure how much review will really be needed before committing a budget.