Voice analysis helps track criminals

A little-known voice analysis technique is being used to recognise and distinguish criminal suspects during court proceedings.

Forensic Speaker Identification is used to establish whether different recordings are of the same person, confirming or eliminating people as suspects.

Dr Bernard Guillemin, a senior lecturer in the Department of Electrical and Computer Engineering at Auckland University, has been used as an expert witness in high and district courts cases.

He said speaker identification was useful when dealing with large amounts of recordings, and cellphone networks had had a strong impact on the technique as voices over mobile phones are of poorer quality than landlines.

Lisa Wong, a PhD student in the Engineering Department at Auckland University said while forensic speaker identification has many advantages in the crime field, it is by no means perfect.

"It works best when you already have a sample of the voice for comparison to the recording," she said.

Lawyers in New Zealand are aware of the technique but it is not yet widely used. Marcus Martin, director of member services at the Auckland District Law Society, said it was up to barristers whether they used it.

But he added: "We actually know very little about it. We had a couple of seminars last year and we had to fly guys in from Australia to do it."

Overseas, members of the International Association of Forensic Linguists have used the technique in legal cases.

The Police Forensics Department was not aware of forensic speaker identification being used within the system here and Constable Rob Bailey of Auckland central police said officers needed a search warrant in order to tape phone conversations.

He said they would have to have good reason to suspect criminal activity to do so.

* Karen Tay is a journalism student at AUT university.