Video and audio as evidence

Video looks like stronger evidence than a photograph: it shows more, it is harder to misread. Evidentially, though, it opens a vulnerability a photo does not have — the cut. You do not need to manipulate the images to change the meaning of a recording: choosing where it starts and ends is enough.

The typical challenge shifts

With a photograph the argument is about retouching and date. With video the ground moves: was this a continuous recording or the result of editing? What happened before the start and after the end? Why that particular segment?

It is a harder challenge to rebut, because it is not about file integrity — a cut video is a perfectly intact file. It is about the continuity and representativeness of what it shows, and those are properties a hash alone does not certify.

A hash proves the file has not changed. It does not prove the video contains everything that happened.

What actually helps

The continuous position track. Where a photo records one point, a video records a sequence sampled at regular intervals throughout. A coherent, sealed route makes it considerably harder to pass off as continuous a recording that is not.

Extracted frames with fingerprints. Reference frames are extracted from the video, each with its own cryptographic fingerprint, included in the sealed package. They serve two purposes: allowing quick comparison without opening the full file, and providing verifiable anchor points distributed across the duration.

The timestamp on the whole file. As with photos: it attests that this video, in this form and with this duration, already existed at that instant. Later editing would produce a different file, with a different fingerprint and no matching timestamp.

A practical limit to know in advance

One operational constraint is better known before a site visit than after: maximum duration per file is around twenty minutes, due to a mobile platform limit rather than a product choice. Longer recordings must be organised as multiple segments, each sealed independently.

Not a problem in itself — several sealed segments with a continuous track remain perfectly usable — but it needs planning, because discovering it midway through an important recording is far worse than knowing beforehand.

Budget for the cost too: a video consumes two slots, a photo one. The combination that almost always works is a video for overall context and sealed photographs for the details that matter.

Audio: two different uses

Two things often conflated are worth separating.

A voice note attached to a photograph is supporting documentation: recorded on site to fix what the image does not say — conditions, circumstances, references. It stays inside the sealed package and is not transmitted. It is the quickest way to avoid trusting to memory details you will need months later.

A standalone audio acquisition is something else: a recording that is itself the exhibit. It follows the same treatment as video for fingerprints, signature and timestamp, with the waveform instead of the histogram as the representation sent to the server.

One technical point stated openly: for audio no provenance block is embedded in the file, because the track is raw and embedding would alter it. The guarantees remain the external ones — fingerprints, signed manifest, timestamp — which are in any case the only authoritative ones for photos and video too.

Legal limits: the part not to underestimate

Caution here is not a formality. Recording conversations and filming people and private premises are subject to strict and layered limits, depending on context, purpose, whether the recorder is party to the conversation, the location, and the forum where the material will be used.

On data protection, Regulation (EU) 2016/679 (GDPR) applies, with its obligations on lawful basis, minimisation and purpose limitation. On the evidential and procedural side the assessment differs substantially between civil and criminal proceedings, and between Member States.

That a tool technically permits a recording says nothing about its lawfulness, nor about its usability as evidence. These are legal assessments to be made beforehand, case by case, with a lawyer. An application cannot authorise what the law does not permit, and presenting the question as settled would be improper.

What remains true

With video and audio sealed at capture you demonstrate that this file, with this duration and this content, existed at that instant and has not been modified. Together with the position track and the reference frames, the coherence of the recording becomes checkable too.

You do not demonstrate that the recording contains everything that happened, nor that it was lawful. The first remains a contextual assessment; the second a legal one. Both sit outside what cryptography can do, and saying so is part of the job.

Continue with the chain of custody and how the opposing party verifies your evidence.