What the world of fiction knows nothing of is the waiting, the time it takes for some processes to be carried out, the effort they require and technical deficiencies they entail. In a film there is always enough audio material available, of a sufficient, if not excellent quality. Background noise can barely be heard, simply because it can be filtered out at the click of a mouse. And uncertainties do not even feature.

"Voiceprint" is one term that makes phonetician Künzel’s ears prick up, for this invented word indeed promises incontestable identification, as though we were able to compare a human voice with a fingerprint. Voiceprints however are initially only visualizations, of frequency and volume for instance, and thus worse than the human ear’s power of resolution. Spectrographs such as these were still accepted as evidence in court in some US states until the late 1990s, although as early as the 1970s it had been established that they often led to wrong judgments. For voiceprints merge the voice and background noise as well as language and its transmission or recording. The result: Merely the same transmission path could suggest similarities, even though it is not the speakers who are similar, but their telephones. And anyone who has ever heard a voice imitator or been hoarse has reason to doubt whether voices really can be identified so clearly.

The pitch is not the only recognizable feature of a person’s voice. The frequency of errs, accent and dialect all have a role to play, provided the speaker does not put on a voice and there is no background noise. Hermann Künzel once identified a terrorist by way of a speech impediment. For whenever one of the guards holding a kidnap victim pronounced the sound "sch," he made a characteristic whistling sound, like that made by air escaping through a gap between incisor teeth. Today experts can use technological support to measure many distinctive features such as the exact "length of a plosive P" and match them with representative data.