The Hala Protocol

Glossary

Accountability mechanism

Accountability mechanisms include judicial systems, such as international or domestic courts and tribunals, whether via criminal prosecution or civil lawsuit; as well as non-judicial systems, such as truth and reconciliation commissions, treaty monitoring bodies, and ombudsmen and human rights commissions. They are designed to ensure that individuals, organisations, or institutions comply with and are held responsible for their actions and decisions. These mechanisms aim to make sure that parties’ actions align with established law, standards, or commitments and that there are consequences for failing to adhere to these obligations.

Admissibility (of evidence)

The quality of being acceptable or valid as evidence in a court of law.

Algorithm

A well-defined procedure or set of instructions that allows a computer to solve a problem or respond to a predetermined scenario.1

Algorithmic training

A process of using a data set to train a machine learning model’s parameters and omptimise its ability to analyse new data and perform requested outputs.

Anonymisation

The process of making it impossible to identify a specific individual.2 In the context of audio data, anonymisation may involve altering a voice to mask the speaker’s identity while leaving the content of the audio and other speech attributes intact.3 It may also involve altering attributes of the data that can be linked to, infer, or reveal an individual’s identity, such as background noise or the audio content.4 Audio anonymisation methods include noise addition, voice conversion, or ‘disentangled representation’ machine learning.5

Audio

Any sound that can be heard by the human ear within the acoustic range.6

Audio data

Raw or processed electrical signal that is captured and stored in the form of sound, including speech, music, or ambient sound.

Audio data file

A digital vessel for housing audio data. Each audio data file may include the original copy of the audio data, any duplicates, and the audio’s metadata.

(Audio) signal

An electrical representation of sound waves that carries the frequency, amplitude, phase, and other essential information of the sound, such that it can be perceived and reproduced. By converting the sound wave into an electrical signal, the audio signal can be captured, transmitted, stored, and processed by electronic devices.7

Audit trail

A reproducible step-by-step record of the collection effort and the data’s chain of custody.

Authenticity (of evidence)

Whether a piece of evidence is what it professes to be in origin or authorship.

Chain of custody

The chronological documentation of the sequence of custodians of a piece of information or evidence, and documentation of the control, date and time, transfer, analysis and disposition of the information.8

Collection effort

The entire process of data collection, processing, and data transfer.

Collector

A civil society entity engaged in the collection of data first-hand or from third party sources, where the data could serve as potential evidence of violations of international law.

Critical listening

The aural review of audio data to determine the audio contents and characteristics.

Cryptography

The practice of digitally encoding or decoding information, such as through encryption.9

Cryptographic hash value

A fixed-size numerical value generated by an algorithm and attached to a piece of data that provides a mathematical demonstration of the data’s uniqueness without revealing the original content.

Cryptographic signature

Two numerical keys (i.e. digital codes) attached to a piece of data, one private and one public, that are generated by an algorithm and mathematically linked. The public key is used to encrypt the data and the private key to decrypt it.

Data set

A collection of data.

Data subject

An identified or identifiable natural person whose personal data is collected. An identifiable natural person is one who can be identified, directly or indirectly, in particular by reference to an identifier such as a name, an identification number, location data, an online identifier or to one or more factors specific to the physical, physiological, genetic, mental, economic, cultural or social identity of that natural person.10

Deletion

The permanent and irreversible removal of the data, including all duplicates and backups.

Due diligence

The exercise of reasonable care and investigation. In a human rights context, it involves ‘assessing actual and potential human rights impacts, integrating and acting upon the findings, tracking responses, and communicating how impacts are addressed’.11

Duplicate copy

An identical (bit for bit) copy of an original digital asset.12

Encryption

The process of making data inaccessible without a decryption key.13

Enhancement

The processing steps taken to improve the overall quality and intelligibility of the audio, including reducing unwanted noise or increasing desired sounds, such as speech, without adding to or detracting from the audio’s content.14

Evidence

Information that has been tendered (formally submitted) to a court as part of a criminal or civil trial for the purpose of proving or disproving an alleged violation of the law.

Evidentiary value

The degree of quality or usefulness for the purpose of proving or disproving an alleged violation of the law. Information that will not be tendered as evidence may still have evidentiary value if it could be used to further an investigation.

Exculpatory

That which may indicate an alleged perpetrator’s innocence.

Equality of arms

A legal principle that requires all parties involved in litigation to receive an equal balance of procedural opportunity, as part of the right to a fair trial.

Inculpatory

That which may indicate the guilt of an alleged perpetrator.

The consent provided by a data subject regarding the future use of their data, once the data subject has received clear and understandable information, including: why the data is being sought; how the data will be used and foreseeable consequences; any risk related to the data’s future use, and any relevant safeguards in place; the data subject’s right to refuse consent and to revoke consent if granted; and, that their consent must be given voluntarily without coercion.15

Interoperability

The ability of different information systems, software, or devices to exchange, interpret, and use information seamlessly.

Labelling

The use of a labelling scheme to name a data file, denote the content of the data, and organise a data set to make it easier to search. A label may be a unique yet uniform identification number, or it may comprise the time, date, and/or location of collection. Labelling may also involve critical listening and analysis of the audio data’s content, e.g., ‘possible troop movements’, or ‘#possiblegunshot’.

Metadata

Data about data; metadata often includes a data file’s characteristics and history, and describe how, when, and by whom a digital file was collected, created, accessed, enhanced, modified, and/or formatted.16 Metadata can be either part of the data upon collection (known as embedded, internal, or application metadata), or added to the data (known as associated, custom, external, attached, or process metadata) after its collection. Metadata may include: the audio data file label; date and time of creation/of collection; details of the device used to record the audio data (i.e. make, model, serial number, source of power); details of any device(s) used to store the audio data; the technical specifications to which any collection tools are calibrated (a.k.a. ‘recorder’ metadata); the length of the audio data; the size of the audio data; the geo-location from where the audio data was emitted; the physical location in which the collection was made.

Non-State Armed Group (NSAG)

Also ‘armed non-State actors’: While absent a uniform legal definition, the definition relied upon by civil society group Geneva Call is: ‘[O]rganized armed entities that are primarily motivated by political goals, operate outside effective state control, and lack legal capacity to become party to relevant international treaties. This includes armed groups, de facto governing authorities, national liberation movements, and non- or partially internationally recognized states.’17

Original copy

Also ‘first copy’: The unprocessed form of the audio data, as collected. There may be multiple duplicates of the original copy of the audio data, verifiable in that all original copies should share the same content-based hash value or digital signature.

Personal data

Any information relating to an identified or identifiable natural person.18 A person can be ‘identified or identifiable, directly or indirectly, by means reasonably likely to be used related to that data, including where an individual can be identified from linking the data to other data or information reasonably available in any form or medium. If you are using publicly available data, note that this data can also be personal, and therefore may involve some of the same considerations as non-public personal data.’19

Personnel

Staff and contractors affiliated with the Collector.

Preservation

The policies and strategies required to manage and maintain digital information with enduring value over time, so that the digital information is accessible and usable by its intended users in the future.20

Prejudicial effect (of evidence)

The effect caused when the nature of a piece of evidence can have a negative effect on the fairness of the trial as a whole, and in particular, the rights of the accused.

Processing

The phase of the Collection effort that encompasses storage and preservation of the data, any duplication and enhancement of the data, and any analysis of the data.

Probative value

The quality or function of demonstrating or proving the existence of a fact.

Redaction

In the context of audio data, the act of obscuring data with a sound or removing/muting a portion of the audio.

Relevance (of evidence)

An item of evidence is relevant when it pertains to the matters considered at trial, i.e., when it can be used to show that the existence of a particular fact is more or less probable.

Reliability (of evidence)

An item of evidence is considered reliable if the veracity and accuracy of its content can be trusted.

Scraping

A method of extracting mass quantities of data from websites.21

Third-party recipients

Organisations or entities with whom the Collector shares audio data.

Third-party sources

Individuals, organisations, or entities that provide audio data to the Collector.

Weight (of evidence)

The relative importance attached to an item of evidence in deciding whether a certain issue has been proven or not.

Footnotes

  1. OHCHR, Berkeley Protocol, Glossary. ↩

  2. OHCHR, Berkeley Protocol, Glossary. ↩

  3. Natalia Tomashenko et al, ‘The VoicePrivacy 2020 Challenge: Results and findings’ (2022) 74 Computer Speech & Language, page 2. ↩

  4. Natalia Tomashenko et al, ‘The VoicePrivacy 2020 Challenge: Results and findings’ (2022) 74 Computer Speech & Language, page 2. ↩

  5. Natalia Tomashenko et al, ‘The VoicePrivacy 2020 Challenge: Results and findings’ (2022) 74 Computer Speech & Language, page 2. ↩

  6. Hasan Fayyad-Kazan et al, Verifying the Audio Evidence to Assist Forensic Investigation, (2021) 14(3) Computer and Information Science 25, page 27. ↩

  7. Workplace Technology Term Dictionary: Audio Signal (C&C Technology Group). ↩

  8. OHCHR, Berkeley Protocol, p. 61. ↩

  9. OHCHR, Berkeley Protocol, Glossary. ↩

  10. GDPR, Art 4(1). ↩

  11. OHCHR, Guiding Principles on Business and Human Rights, Guiding Principle 17. ↩

  12. NIST IR 8387, Digital Evidence Preservation Considerations for Evidence Handlers (2022), p. 6. ↩

  13. OHCHR, Berkeley Protocol, Glossary. ↩

  14. Scientific Working Group on Digital Evidence, SWGDE Best Practices for the Enhancement of Digital Audio (2020), page 3; Hasan Fayyad-Kazan et al, Verifying the Audio Evidence to Assist Forensic Investigation, (2021) 14(3) Computer and Information Science 25, page 29. ↩

  15. PILPG, Handbook on Civil Society Documentation of Serious Human Rights Violations: Principles and Best Practices (2016), page 9. ↩

  16. OHCHR, Berkeley Protocol, Glossary. ↩

  17. Pascal Bongard & Jonathan Somer, ‘Monitoring Armed Non-State Actor Compliance With Humanitarian Norms. A Look at International Mechanisms and the Geneva Call Deed of Commitment’, (2011) 93(883) International Review of the Red Cross 673, page 674, fn. 3. ↩

  18. GDPR, Art 4(1). ↩

  19. The UN Global Pulse Risk, Harms and Benefits Assessment Tool, page 3. ↩

  20. OHCHR, Berkeley Protocol, Glossary. ↩

  21. OHCHR, Berkeley Protocol, Glossary. ↩