Insights · Audio evidence

What should lawyers check before relying on an audio recording in court?

On 28 August 2026 the Commercial Court in London gave permission for an inquiry into damages under a cross-undertaking, in a matter where the claimed losses are said to exceed £5 million. The decision itself is procedural. What sits behind it is one of the more instructive audio evidence failures to reach an English court in recent years: a freezing injunction obtained on excerpts of a covert recording that nobody presenting it had heard, and discharged once it was listened to.

In short

Listen to all of it, and put the complete original before the court. In Apollo XI Ltd v Nexedge Markets Ltd [2025] EWHC 1488 (KB), a freezing injunction obtained without notice on a covert recording was discharged after the court found that excerpts had been “cherry-picked” from different calls, no full transcript was provided, and counsel presenting it had not listened to the recording; listening took about 30 minutes. Before relying on audio, obtain the complete original file, have a named person listen end to end, produce a transcript with timings, speaker attribution and marked uncertainties, record how the file was obtained and handled, and be ready to say what is not relied on. How a recording was made can also create liability. Hong Kong courts apply their own rules; take advice.

What happened

The account below is taken from the two published judgments, both available on the National Archives Find Case Law service. Nothing here has been finally determined, and the underlying claims remain live. According to the 2026 judgment in Apollo XI Limited v Nexedge Markets Limited [2026] EWHC 2240 (Comm), on 11 April 2025 a recording was made of various calls made by Nexedge’s then chief executive designate, in Nexedge’s office, without him or Nexedge being aware at the time that he was being recorded.

On 15 April 2025, “in reliance on the Recording, Apollo applied without notice for the Freezing Injunction”. It was granted the following day. On 17 June 2025 Saini J discharged it, in Apollo XI Ltd v Nexedge Markets Ltd [2025] EWHC 1488 (KB).

On how the recording was obtained, Saini J concluded that “a strong inferential case” was available to Nexedge that the recording had been orchestrated as a form of “bugging”. That was an assessment of arguability, not a finding of fact, and it should not be read as one.

On how the recording was presented, the findings were more concrete. Saini J recorded that “it is not clear to me why a full transcript of the Recording was not put before the Judge”, and that “a decision appears to have been taken by Apollo to ‘cherry-pick’ excerpts from various different calls”. Counsel appearing on the without notice application “had not himself listened to the Recording and he did not invite the Judge to do so”. Listening to the material “was a relatively straightforward process and took no more than about 30 minutes”.

When it was listened to, the picture changed. The passage the applicant was apparently most concerned about involved a discussion of a “move over”. Saini J found that Nexedge had given notice on its office contract and was moving to a different building nearby in Canary Wharf, that Apollo knew about the planned move, and that it failed to disclose it. In his words, “the ‘move over’ was not offshore but a few yards up the road in Canary Wharf”. The 2026 judgment records that the threshold at this stage was “analogous to that for summary judgment”, and that Nexedge advances counterclaims including breach of confidence arising out of the use made of the recording.

The issue behind the headline

Two failures are stacked on top of one another here, and they belong to different disciplines. The second, the one the court dealt with, is evidential. Audio was treated as though it were a document. Passages were extracted, assembled into a narrative, and put before a judge in a form that could not be checked against the source. The remarkable detail is not that this was done, but how small the corrective step would have been. Thirty minutes.

The first failure comes before all of that, and no court has been asked to rule on it. Calls made by a chief executive designate, inside his own company’s premises, were captured without his knowledge. Whatever the eventual findings, the fact pattern alone should interest anyone responsible for what can be overheard in their own meeting rooms.

Practitioner analysis

A transcript is derived evidence. It is an interpretation of a signal produced by a person or a tool, and it inherits the assumptions of whoever produced it. Where audio is relied on, the primary exhibit is the audio file itself, in its original container, and the transcript is a working aid that must be capable of being checked against it. That is only possible if the transcript carries timings, identifies speakers with a stated basis for attribution, and marks inaudible or uncertain passages rather than silently smoothing them.

An excerpt is not the recording, and excerpts across several calls are not a recording at all. Assembling passages from different conversations produces a composite that never existed as an event. Even where each fragment is accurate, the whole is a construction, and it should be described as one. The disciplined approach is to disclose the complete source, identify each excerpt by its position within it, and state plainly what has been left out and why. Where an excerpt is relied on for its meaning, the surrounding context is not optional colour; in this case it was the entire answer.

Provenance and continuity are separate questions from content, and both need answering. How was the audio captured, on what device, in what format, and by whom? Is the file offered to the court the original as recorded, or an export, a re-encode or an extract? Container and stream metadata, encoder identification, sample rate, timestamps and any sign of editing at the boundaries of the material will often help answer this, at modest cost. A file that has been through a messaging application or an editing suite has lost much of what an examiner would rely on, and the party relying on it should say so before it is put in issue.

Somebody has to listen, and that person has to be able to speak to it. Where audio is the foundation of an application, particularly one made without notice where the court hears only one side, the material must be examined by the person presenting it or by someone answerable to them, and the court should be offered the means to examine it too. A hyperlink and a USB stick are an offer of access, not a presentation of evidence.

Covert audio carries its own liability. A recording obtained in these circumstances is not simply weak evidence. Here its use generated a counterclaim for breach of confidence, and it forms part of the background to a damages inquiry now permitted to proceed. Parties who regard a covert recording as a weapon should understand that it can become an exhibit pointing the other way, and that the manner of its capture will be examined alongside its contents.

What organisations should learn

For litigators and those instructing experts, the practical step is a short discipline applied before any audio is deployed: obtain the complete original, have it listened to end to end by a named person, produce a verified transcript with timings and marked uncertainties, record how the file was obtained and handled, and be ready to explain what is not being relied on. On a without notice application, where the duty of fair presentation is at its highest, that work is not a refinement. It is the minimum. Hong Kong courts apply their own rules on without notice applications and on recordings; take advice.

For boards, general counsel and those responsible for premises, the exposure runs the other way. Sensitive discussions happen in offices, in rooms with conferencing equipment permanently installed, and increasingly near devices that make recording trivial and invisible. The question worth asking is whether anyone can say with confidence who has access to the room, what equipment is in it, what that equipment is capable of, and who could place something in it and return later to collect it. Most organisations have a clear answer for their network and no answer at all for their meeting rooms.

For anyone weighing whether to record a counterparty covertly, the case corrects the assumption that the material will simply be usable. The court will look at how it was obtained, what was done with it, and what was withheld.

Conclusion

The most striking number in this case is not the £5 million. It is the thirty minutes. An order restraining a company’s assets was obtained on audio that nobody presenting it had heard, and the gap between the case as advanced and the case as it actually was could have been closed in less time than it takes to read the papers. When evidence is a recording, listening to it is not diligence. It is the first step of competence, and everything an expert would go on to say about provenance, continuity and completeness assumes that somebody has taken it.

See expert witness reports, mobile phone forensics and computer forensics.

Sources

General commentary on public reports, not legal advice and not a finding about any organisation or person named in those reports. Facts are as reported at the time of writing.

Frequently asked questions

What should lawyers check before relying on an audio recording in court?

Listen to all of it and put the complete original before the court. In Apollo XI v Nexedge a freezing injunction was discharged after excerpts were cherry-picked from different calls and counsel had not listened to the recording. Obtain the original file, listen end to end, produce a timed transcript and record how the file was obtained. Hong Kong courts apply their own rules; take advice.

Is a transcript the same as the recording?

No. A transcript is derived evidence, an interpretation of the audio. The recording itself, in its original form, is the primary exhibit, and the transcript should carry timings, speaker attribution and marked uncertain passages so it can be checked.

Can excerpts from several calls be relied on?

Excerpts from different calls form a composite that never happened as one event. Disclose the complete source, identify where each excerpt comes from, and say what has been left out and why.

How can you tell whether an audio file is the original?

A forensic examination of container and stream metadata, encoder details, sample rate, timestamps and signs of editing can help. Files passed through messaging apps or editing software may have lost much of that information.

Can using a covert recording backfire?

Yes. In Apollo XI v Nexedge the use of the recording led to a counterclaim for breach of confidence and forms part of the background to a damages inquiry. How a recording was made will be examined alongside what it contains.

Facing something like this?

The first discussion is confidential and free, and usually changes what you decide to do next. Call +852 5808 1071 or use the case review form.