This morning, Ekot and the P1 program Kaliber reported on a major drug case in Helsingborg, where two police officers were set to listen to thousands of conversations, to determine which people were speaking.
The district court was strongly critical of this, and Fanny Carlström Plaza, speech analyst at NFC, says that with their method it takes upwards of two weeks to make a single comparison like this.
– As we work now, an analysis takes a very long time, and what we usually recommend is to sift through the material, says Fanny Carlström Plaza, speech analyst at NFC
That was in the beginning in 2018 as the police intercepted thousands of calls, which were then used as evidence in the more than year-long trial regarding drug trafficking in Helsingborg. Since several of the phones were wandering around among different people, the prosecutor had two police officers listen through the conversations to determine who was talking. They could sometimes do hundreds of analyzes a day.
In the judgment, the approach is criticized. The court questions, among other things, that the police have made voice analyzes regarding conversations that only contain a single word, and that they have performed analyzes regarding conversations in languages that they do not speak.
But Cathrine Bengtsson, chamber prosecutor and deputy head of the preliminary investigation, says that they had no alternative but to use the police, because it was such a large amount of material.
– Here you have two police officers who have worked with it for a long time, listened to a huge amount of conversations, you have had time to listen to all conversations. It had never happened at NFC, says Cathrine Bengtsson
– Even when I worked with the investigation, I could immediately say in the end that “now it is he who speaks”, “now it is the other who speaks”. So that for me then who has worked with this case for two years. Obviously who is talking in many cases.
Fanny Carlström Plaza at NFC explains that it is of course quite possible even for people who are not trained phonetics to recognize voices – but that there will be subjective assessments. What NFC does is, for example, by making acoustic measurements and looking at frequency content, check what it is that actually speaks for or against the voices on two recordings coming from the same speaker.
– What is it that says that it is the same speaker? What concrete measurements or information can you obtain that actually confirm your observations?
المصدر أخبار أيسلندا