Is the voice of the suspect the same as the one whispering a bomb threat on a police tape recording? How many people were arguing in the room held under audio surveillance, and what was the argument about? From which part of Germany is the person who made the anonymous blackmail call just the night before? Was the captain under the influence of alcohol while talking to the coast guard after he had run his ship into a bridge pier? These are just a few questions an expert in speaker recognition has to deal with every day. As soon as a case involves human speech or acoustic clues of any kind, the expertise of a forensic phonetician is required.

But, how does one become a specialist in forensic phonetics? In this case, there is no such thing as a "usual path," because the forensic application of phonetic sciences is a relatively recent development and involves a variety of disciplines--linguistics and physics in the first place, but also medicine, to name just a few. Therefore, the experts in this particular field may have gone through phonetic training with varying emphases. What is certain is that they all have a particular interest in languages and speech.

In my case, my initial interest in language was not scientific but rather playful. It was supported by interest in artificial languages from fantasy literature like Tolkien's Elvish or the Klingon vocabulary from the Star Trek Universe. Because my family moved around quite a bit, I soon began to learn not only the various dialects of German, but also some foreign languages.

Later on, I was introduced to the search for acoustic traces in the German Navy, where I started my duty as a sonar operator on a submarine. For 2 years, I spent my time listening to underwater sounds and noises. The idea of drawing conclusions about the entire tactical situation from just a few acoustic clues, putting the pieces of the "sound puzzle" together to compensate for the blindness of our boat, was most fascinating to me.

I then started to study law in Trier (Germany), which turned out to be a bad choice since the curriculum didn't leave any space for practical work or one's own ideas. After 1 year I took a short break and a close look around for a discipline that would better fit my interests. I quickly found out that the University of Trier had the leading faculty for forensic phonetics. Not only had they conducted joint projects with the federal police of Germany, but the professors and lecturers were also frequently called upon to give expert opinion in the criminal courts. What could be more suitable? It gave me the opportunity to combine my acoustical knowledge from the Navy with my interest in speech and languages.

Phonetics is about speech production, transmission, and perception. During my studies I consolidated my physical knowledge about sounds and speech. I learned about the anatomy of the human body from diaphragm and lungs upwards, and I was trained in the use of computer workstations for any kind of measurements concerning the human voice. Moreover, I practised transcribing any human utterance with a phonetic alphabet that can be used for any language of the world (the so-called International Phonetic Alphabet). In addition, I took courses in German philology and psychology to learn about language acquisition, speech recognition, and human decoding procedures. I finished my studies after 5 years with an MA degree.

My duties can be subdivided into four main areas:

  • Speech enhancement and decoding includes different techniques to eliminate all kinds of interference that disturb recordings of human voices. Due to noisy environments or to the use of regular or mobile telephones, law enforcement units are often unable to understand the words spoken by the offender. To improve intelligibility, we use a variety of filtering tools to extract the desired signal from the distorted material. If quality enhancement is impossible, we are left no other choice than to decode the questioned utterances with just patience and experience. The expert then familiarises himself or herself with every detail of the recording, a task that includes repeated listening to the relevant passages, sometimes for hours.

  • Tape authentication is all about the question of whether the given tape recording is the original or a copy. Thorough listening and analysis is necessary to find any possible trace of manipulation.

  • Speaker analysis becomes relevant when there is a speech recording of an anonymous offender and the police investigation has not yet come up with a suspect. A single voice can give a lot of information about the perpetrator, and after accurate analysis we may be able to judge more than just age and gender. Even background noises can give clues about the whereabouts of the wanted criminal.

  • Finally, speaker identification describes the task of comparing at least two voices: one recorded in connection with the crime, the other from a given suspect. Characteristics of each voice are carefully registered by auditory and technical analysis and then compared. The result is a statement about the identity or non-identity of the voices.

However, even though crimes nearly always involve acoustic evidence and the existing institutes are overloaded with orders from criminal courts and the police, the potential of speaker recognition is often misjudged. Sometimes the officials in charge of a case expect us to "press some buttons" to "convert" a disguised voice into its original condition, a process that in fact usually takes weeks, if it is possible at all. On the other hand, they are often unaware that just a few words from the perpetrator may reveal important details about his or her identity.

To be honest, there are not many job vacancies for a specialist in forensic phonetics, and I was very lucky to get one in the capital of Germany. But there are also a number of self-employed experts working for courts all over Europe. Hopefully the opportunities to work in the field of forensic phonetics will be expanded in the future to develop an even more effective speaker recognition network.