Introduction

At the turn of the 20th century, warring parties were introduced to an array of new technologies that would help them communicate with one another during conflict. These methods of communication have evolved quickly throughout the ages, advancing the way that warring parties communicate with one another and the coordination of their efforts in theater. More recently, audio information flowing from these channels has been collected and used to document war crimes and crimes against humanity by providing real-time on-the-ground information that can help establish the truth of specific incidents.

Over the past decade, active conflicts have brought the importance of audio data for accountability into sharp focus. Following the Russian invasion of Ukraine in February 2022, members of the international community quickly noticed that Russian forces were using unencrypted radio channels to communicate. By monitoring and analyzing these communications, they were able to corroborate allegations of widespread war crimes by Russian soldiers.¹

In the Netherlands, telephone communications helped aid the conviction of three accused in connection with the downing of flight MH17.² At the European Court of Human Rights, audio communications were used to support the allegations that Russia violated human rights on the territory of Ukraine.³ Audio data has also played a prominent role at the International Criminal Court (ICC). In the 2021 Ongwen judgment, intercepted radio communications were used to establish the dynamics of the armed group, including the role of the defendant within the group hierarchy and how and when attacks were reported to commanders.⁴ In the 2024 judgment in Al-Hassan, audio recordings helped show that the armed group intended to establish an Islamic State on the entire territory of Mali, which was an essential element of proving crimes against humanity.⁵

Civil society organisations (CSOs) are critical stakeholders in the accountability landscape and contribute substantially to the documentation of international crimes. Across recent conflicts, the documentation efforts of CSOs have helped ensure that audio data, which is ephemeral and highly volatile by nature, is preserved for accountability. A radio transmission or telephone call, if not intercepted and recorded in the moment, cannot be accessed later; a voice recording posted to social media can be removed by the platform or uploader without notice; and a voice note sent over an instant messaging app can be deleted by the sender within a certain time. As a result of this volatility, the copy of the data held by the CSO may be the only one in existence.

In recent years, a number of resources have been developed to help CSOs undertake documentation work effectively and responsibly. Examples include the Berkeley Protocol on Digital Open-Source Investigations, the GLAN/Bellingcat Methodology for Online Open Source Investigations, the WITNESS Video as Evidence guide, and the PILPG Handbook on Civil Society Documentation of Serious Human Rights Violations. However, these standard setting initiatives do not address audio data specifically, resulting in an information gap surrounding how to best collect this type of data.

Furthermore, despite often being discussed alongside (or subsumed within) video material, case law from international criminal courts and tribunals suggests that video and audio are not treated identically. In particular, audio is less likely than video to be used as standalone evidence. For example, in the Bemba et al case, the ICC Trial Chamber elaborated extensively on how they could establish that the voices on the telephone recordings were those of the accused.⁶ The Chamber did not rely solely on their own recognition of the voices, but rather looked at a number of other factors and corroborative evidence. This conforms with the approach taken at the ICTY, whereby intercepts were considered to be ‘a special category of evidence in that in and of themselves, they bear no prima facie indicia of authenticity or reliability, and as such these requirements must generally be fulfilled by hearing from the relevant intercept operators or the participants in the intercepted conversation’⁷ Video, in contrast, has been relied upon as a standalone form of evidence.⁸ This supports the proposition that it is harder to identify someone from their voice as compared with their (moving) image; and as a result, more evidence may be needed to corroborate standalone audio than video and photos.