Angelika Braun is a forensic expert of a different kind. Instead of the traditional techniques of fingerprints or traces of blood, she analyses voice recordings of bank robbers, kidnappers, terrorists and other criminals to gain clues to their identity and other information.
The 55-year-old professor of phonetics at the University of Trier in western Germany says: "Every detail is important, even the background noises." She has been called in as an expert in Germany and abroad to work on serious criminal cases. "Sometimes it all has to be done very quickly," Braun says.
Abductions are cases of this kind. "A human life can in these cases depend on seemingly minor steps in the investigation," says Braun, who worked on voice recognition for criminal investigation departments in western Germany between 1986 and 2000.
One of her better known cases was the 1996 abduction of German multimillionaire Jan Philipp Reemtsma. "We phonetics experts have a well-tuned ear," says Braun, who took up her current post as professor in 2003.
"Forensic scientists are increasingly important for police work," Willy Thevessen, police spokesman in the city of Moenchengladbach, says. He points to DNA comparisons alongside voices and sounds.
Braun was also involved in a case involving the abduction and murder of a 10-year-old boy in western Germany last year. She investigated reports of a scream being heard near the scene in the town of Grefrath. Her report failed to lead to new evidence, but her expertise was nevertheless useful. "For us, Braun is the first person we call on," Thevessen says.
A voice can betray age, gender, origin and nationality, "even if the speaker whispers or changes their voice," the professor says. She investigates sounds and noises in her language laboratory and says that age can be assessed to within five or 10 years.
Origins can also normally be uncovered. "For this the school years are decisive," Braun says. If the subject has a foreign accent, the nationality can be determined on the basis of systematic language errors or certain sounds.
Braun also does voice comparisons, using specialised computer software. "It makes language visible and measurable," she says. The software assesses whether the subject has a low or high voice, and this can be done irrespective of the quality of the recording. But the computer can't do it all.
"There is no computer that can tell me where someone comes from or how old he is," she says. For this she has to use her own skills alongside those of the software.
When comparing voices, Braun might examine whether the voice on an Islamist video claiming responsibility for an attack corresponds to a voice recorded during a phone conversation. In a recent English court case she gave evidence on whether the voice of a controversial preacher was in fact that on a video.
Braun is also involved in less spectacular cases. In the case of a sinking on the River Elbe last year, she was able to show that the captain of a tanker carrying heating oil, who had a history of alcohol abuse, was in fact under the influence of alcohol at the time.
She used recordings of the radio conversations between the captain and the river authorities. "That was a clear-cut case," she says. A phonetic analysis showed that his speech organs did not reach "certain goal points" and that he spoke more slowly.
Braun moved to the university in 2000 on becoming increasingly interested in the research aspect of her work. "For example: How does speech change with age? How does the voice change when whispering or shouting?"
She now has around 60 students learning everything from articulation through transmission of the sound waves up to perception by the ear - and how they can track down criminals using their voices.