UK & World News
How Forensic Experts Will Analyse Beheading Clip
The video of James Foley's murder is only two minutes and 28 seconds long and is shot against a barren desert backdrop. But it could still reveal important clues.
Security services are now combing the video using digital forensics. What will they be looking for?
There is no CSI-style 'voice print' that uniquely identifies a person. But there are a number of discrete techniques.
Joel Tobias says his company, CY4OR, would use patented software "to analyse the video and audio under laboratory conditions, in stages that would ascertain authentication for signs that the evidence has been edited or tampered with in some way".
This would determine whether the pictures match the audio.
Other elements of the video could be useful: It should be possible to estimate the height of the killer, known as 'John' according to a Guardian report; and the location and time, using the shadows in the video.
Analysts will also look at the clothing, shoes and weapons in the video, since "identifying where these items have been sourced may help determine where the perpetrators are located," according to Sam Type, director of Geek Forensics.
But the audio will be the focus of the investigation.
This is an automated process. Phonetic software will identify obvious characteristics in the voice of the killer, such as the pitch and intensity.
This will be matched against a database of voices of those thought likely to be associated with IS. This voice databank might have been built up from previous and ongoing investigations.
The software will sift the majority of voices, leaving 50-100 possible matches.
Experts phoneticians - a person who specialises in the study of speech sounds - will then comb through those.
They will look for patterns of speech - rhythm and timing. And most of all, voice quality - the timbre.
Elizabeth McClelland, a forensic speech analyst, told Sky News that 'John's' voice is "not the most routine voice, but not highly distinctive".
If the security services do have a reference sample, Ms McClelland said she would not be surprised if the security services could identify the killer with some confidence.
But it won't be a 'match' - merely strong evidence, as the field of forensic audio is as much art as it is science.
"Speech is a behaviour," Ms McClelland told Sky News. "It's not like measuring a piece of wood."