Digital Audio Forensics Fundamentals

From Capture to Courtroom

James Zjalic

• Publisher: Focal Press
• Year: 2020
• Print ISBN: 9780367259129
• eBook ISBN: 9781000205718
• Edition: 1
• Topic: Computer Science
• Subtopic: Computer Science General
• Index: Computer Science

Digital Audio Forensics Fundamentals offers an accessible introduction to both the theory and practical skills behind this emerging field of forensic science. Beginning with an overview of the history of the discipline, the reader is guided through forensic principles and key audio concepts, before being introduced to practical areas such as audio enhancement, audio authentication, and the presentation of reports.

Covering all aspects of audio forensics from the capture to the courtroom, this book is pivotal reading for beginners entering the field, as well as experienced professionals looking to develop their knowledge of the practice.

Chapter 1

Introduction

In the modern world, the word ‘Forensics’ is defined as the application of science to the justice system, but the actual inception of this word harks back to 44 bc and the post-mortem examination of the infamous Roman leader, Julius Caesar. Upon performing his investigation, the physician, Antistius, arrived at the conclusion that Caesar had been stabbed a total of 23 times, and that it was the second piercing of the chest that inflicted the fatal blow. Once his analysis was completed, he was asked to present his findings ‘before the forum,’ an archaic version of the court system used across the world today. As such, the word itself is derived from the Latin term ‘forensis’ or ‘of the forum’ when translated into English (Wilkinson, 2015). Since the coining of its name, many sub-disciplines have since evolved, with the most recent addition being digital forensics. As it relates to the pertinent topic for this book, digital audio forensics is, therefore, the application of the science of digital audio to legal matters.

In recent years there has been an increase in the exposure of this very niche field of forensics, through real-life cases covered by the media, true crime podcasts and shows which grace our cinema and television screens. The reasons for this newfound publicity are twofold and can be surmised to be due to the proliferation of media outlets combined with the increased use of audio forensics within the courtroom. The latter is primarily a consequence of the now ubiquitous nature of digital devices consuming every waking hour of our lives, the majority of which are capable of capturing sound.

With this in mind, it is essential that accurate and reliable information exists on the subject, not only to allow the next generation of audio forensic practitioners to have the prerequisite information to continue to serve the justice system but also to provide those who are looking for further education in this discipline to have material presented in a non-biased, objective, scientific but accessible format. Although the vast majority of the information contained within this book is available from other sources, many require subscriptions to scientific journals, a large library, extensive in the field training or a rare educational programme.

Scope

This book is written with the intention of being both a primer for those interested in the topic, and as a reference guide for those already active within the field of audio forensics. As such, attempts have been made to ensure it contains the necessary information for a reader to understand all themes covered through the gradual cumulation of knowledge in a progressive manner. References are provided at the end of each chapter, should further knowledge relating to a specific concept be required. Those with an understanding of general audio concepts which relate to audio forensics may wish to advance to chapters which are of particular interest to them, but it is recommended that the book be read from cover to cover so concepts for which a priori knowledge is required can be reviewed to aid in understanding the preceding chapters. The presentation of information is designed to allow the book to serve as a reference for those within the field, and attempts have been made to minimise jargon. Where technical concepts are used, they are dissected to aid understanding.

Although the term audio forensics can be applied to both digital (data whose foundation is a binary representation of ones and zeros) and analogue (everything else which is captured in a non-digital format, for example, vinyl records and magnetic tape), there are as many differences between the two formats as there are similarities. These differences will be touched upon in Chapter 2 to enable the reader to gain perspective from the history of audio forensics and learn how the transition from analogue to digital audio came about, but to encounter analogue audio when working in forensics is now extremely rare as the majority of modern systems are digital. The only cases pertaining to analogue audio will, therefore, be appeals and cold cases. In light of such, it is solely the digital domain on which this book is focused. It appears that the digital element is here to stay, thanks to the lower costs, reduced physical size, increased storage capacities, and the ability to create perfect copies of data among other advantages.

There are also several areas in digital audio forensics which cross into other disciplines, including voice comparisons and analysis. This field lies somewhere between phonetics and statistics, and digital audio is the medium in which the pertinent data is stored. This area will be covered briefly for the sake of completeness, but for detailed information concerning such a topic, other sources by more suitably qualified experts should be sought. Acoustic analysis of signals is another area which is extremely broad and requires a deep understanding of both physics and various mathematical concepts. As the book is to be considered an introduction, an overview will be given with regards to the area, but, as with voice comparisons and analysis, further sources of information on the subject should be consulted.

When people are first introduced to audio forensics, comments are generally made which concern recordings captured via telephones or mobile recording devices, but this is only half the story. Video recordings are essentially a stream of images synchronised with an audio track, and although audio recordings in and of themselves are ubiquitous, the video format is probably the most popular multimedia format in the world as it provides both a visual and audible capture of events. There are, therefore, a phenomenal number of recordings captured every second of every day, and so it is somewhat inevitable that some will end up capturing evidence of an event which can be then used in a legal capacity.

Although scientific principles are the same throughout the world, the judiciary systems and protocols for presenting evidence are not (Ambos, 2003; Block et al., 2000; Froeb and Kobayashi, 2001). The non-technical information provided will, therefore, be agnostic to legal systems, regions, and countries, and will only include areas which pertain to the majority of the population.

Finally, although specific tools will be referenced where used to create demonstrative imagery in the form of figures, the book is software agnostic for two reasons. First, there are no analyses or processes within audio forensics which are only available by using a single tool or piece of software. In reality, there are numerous tools which can perform the same analysis and render the same results. It would also be a disservice to provide a ‘how-to’ guide on a piece of software without explaining the science behind the processes. Far more important is the understanding of concepts so a reader can use any tool they find to be suitable (or create their own tools), perform their own research, and present findings and answer questions in a scientific and logical manner. Understanding why a process is applied, and the impact of the said process on the audio signal is what separates forensic scientists from those who know which buttons to press but have no idea why they are pushing them, what is happening when they do, and what the output really means. A lack of understanding can lead not only to poor analysis through the loss of essential data (in the case of enhancement) but also to misinterpretation (in the case of audio authentication and analysis) and potentially wrongful releases or convictions.

Foundations

Just as a tree is only as strong as its roots, foundations must first be formed to provide a stable structure to all that follows. In order to achieve this, the audio forensics discipline must, therefore, be traced all the way back to the phenomenon to which every living (and non-living) thing is continuously exposed: Radiation (Lamb, 1910; Strutt and Rayleigh, 1877).

Radiation exists all around us in the form of frequencies which can be perceived directly by humans, such as light, heat, and sound, and those for which we require a medium to transpose it into a form which we can identify. For example, although dolphin chatter exists in a spectral region beyond the limits of our auditory system, these sounds can be transposed into those which we can discern. X-rays must be converted to a visual representation for analysis. Microwaves (the kitchen appliance) take their name from microwaves, which we cannot directly sense, but can feel as heat through the medium of food. When all the other forms of radiation are considered, we can hear only an extremely thin slice of the entire spectrum (Figure 1.1).

Figure 1.1The frequency bandwidth perceptible to humans within the electromagnetic spectrum

For thousands of years, sound existed as a transient event, ceasing once the vibrations perceived by the human auditory system had passed. Although the events heard in this manner were, and still are, used within the justice system as evidence in the form of utterances heard by witnesses or conversations heard between two parties, it is highly subjective. Perceived sounds are not only extremely susceptible to human bias (both conscious and subconscious), but also to other factors such as Post-Traumatic Stress Disorder. In addition to this, the sounds heard by a witness cannot be reproduced. This does not, therefore, lend itself to any form of analysis, and, as such, is of limited use and reliability when compared to other types of evidence. Once advances in recording had reached the point that sound could be captured and stored, it was in a form in which the aforementioned factors were no longer an issue and could be replayed and analysed in an objective and scientific manner.

For a sound to become potentially evidential, there are several conditions which must coincide. The first is the occurrence of an event (typically a crime or something relating to a crime). The second is the propagation of an acoustic wave pertinent to the event (for example, a verbal confession or a gunshot). Providing somebody is present, this has some potential as evidence in the form of an individual perceiving the sound and recalling this to a trier of fact, but as discussed above, this would be subjective, due to flaws in the human memory and potential biases. In light of this, there have been numerous studies which have shown the effects of stressful incidents on the mind, even when recollecting directly after the event (Clifford and Bull, 1978; Cook and Wilding, 1997, 2001; Mullennix et al., 2011; Pankkonen et al., 2017; Paunovic et al., 2002; Wells, 1995). When one considers that the witness may not be providing evidence within a courtroom until after an extended period following the incident, the memory is likely to have degraded even further, and subsequently, the reliability of the evidence is further reduced. Biases may also exist, either consciously or sub-consciously. If the witness is an acquaintance of the subject involved, biases may result in them giving an account which supports their friend’s position, or does not provide a full account of what really occurred. Subconsciously, they may also have inherent biases, all of which may distort the evidence. The biases essentially mean that this type of evidence is not too reliable. This is where the meeting of the third condition, that of the capture of propagated sound as an audio recording, becomes critical. If this condition is met, a snapshot of the acoustic events which occurred has been created (Figure 1.2). A witness statement can then potentially be replaced, verified, or challenged by the ironic ‘silent witness’ of a digital audio recording.

Figure 1.2The conditions required for audio forensics

The first two conditions can be reasoned to have a relatively high possibility of occurring based on Locard’s principle of exchange. Dr Edmund Locard (1877–1966) was a French scientist, who, in 1910, built a laboratory in the French city of Lyon, and is considered a pioneer of forensic science. His theory forms one of the two pillars of forensic science, expressed by (Miller, 2014) as:

The truth is that none can act with the intensity induced by criminal activities without leaving multiple traces of his path. The clues I want to speak of here are of two kinds: Sometimes the perpetrator leaves traces at a scene by their actions, sometimes, alternatively, he/she picked up on their clothes of their body traces of their location or presence.

Once this theory took root and was popularised in 1930s North America, it became simplified as ‘Every contact leaves a trace’ (Roux et al., 2012). P.L. Kirk, a professor at the University of California Berkeley Criminology School later expanded on this theory, writing:

Wherever he steps, whatever he touches, whatever he leaves, even unconsciously, will serve as a silent witness against him. Not only his fingerprints or his footprints but his hair, the fibers from his clothes, the glass he breaks, the tool mark he leaves, the paint he scratches, the blood or semen he deposits or collects. All of these and more, bear mute witness against him. This is evidence that does not forget. It is not confused by the excitement of the moment. It is not absent because human witnesses are. It is factual evidence. Physical evidence cannot be wrong, it cannot perjure itself, it cannot be wholly absent. Only human failure to find it, study and understand it, can diminish its value.

(Kirk, 1963)

Kirk also coined the term ‘Individuality Principle,’ which can be considered to be the second pillar of forensic science. The theory is based on the assumption that ‘every object in the universe is unique’ and that forensic science should focus on the source of two items (the questioned and the known). Further to this, one of his PhD students (Kwan, 1977) proposed that the agreement between properties measured from a trace (e.g., a fingerprint found at the crime scene) against a source (e.g., a fingerprint taken from a suspect) did not necessarily mean they were left by the source. The Bayesian approach proposed by the author for use in identification processes is still used across a number of forensic disciplines for the reporting of the results of an analysis to this day.

In the latter half of the twentieth century, the possibility of audio pertaining to a crime being captured became increasingly likely as law enforcement agencies focused on the recording of sound using wiretaps and bugs. In the present age, this possibility has further expanded due to the proliferation of audio recording devices, most notably in the form of smartphones, portable digital recorders (including those used covertly), and call centre recording systems.

As the capture of audio became more likely, developments in the field of audio analysis were also picking up steam. The first use of audio analysis was born out of necessity during the First World War when the US-based Bells Labs created a system to identify submarines. Once the war was over, a company named Kay Electric made the product commercially available, and in 1951 trademarked and marketed it as ‘The Sona-graph,’ producing graphs known as ‘Sonograms’ (Vale, 2019). It is this conversion of audio from the tim...

Table of contents

• Cover
• Half Title
• Series Information
• Title Page
• Copyright Page
• Contents
• Acknowledgements
• Abbreviations
• Chapter 1 Introduction
• Chapter 2 The history of audio forensics
• Chapter 3 Forensic principles
• Chapter 4 The human auditory system
• Chapter 5 Key audio principles
• Chapter 6 Audio encoding
• Chapter 7 Preparatory analysis
• Chapter 8 Audio acoustics
• Chapter 9 Speech forensics
• Chapter 10 Audio enhancement
• Chapter 11 Audio authentication
• Chapter 12 Forensic reporting
• Chapter 13 Research
• Glossary
• Index