New CFC emissions, Cannabis and the Environment, The Noisy Cocktail Party, Automated Face Recognition

28:27
 
Share
 

Manage episode 234276535 series 1301268
By BBC and BBC Radio 4. Discovered by Player FM and our community — copyright is owned by the publisher, not Player FM, and audio is streamed directly from their servers. Hit the Subscribe button to track updates in Player FM, or paste the feed URL into other podcast apps.
New CFC emissions Researchers say that they have pinpointed the major sources of a mysterious recent rise in a dangerous, ozone-destroying chemical. CFC-11 was primarily used for home insulation but global production was due to be phased out in 2010. But scientists have seen a big slowdown in the rate of depletion over the past six years. This new study published in the Journal Nature says this is mostly being caused by new gas production in eastern provinces of China. Dr Matt Rigby of the University of Bristol and the BBC’s Matt McGrath, who has also been following the trail, tell Gareth about the mystery. Yeast to make cannabinoids In California, where cannabis has become a major cash crop since legalisation there, researchers are trying to evaluate the environmental impacts of large scale agricultural planting. But, as Geoff Marsh reports, other researchers are finding other ways to produce various cannabinoids for potential future sale. Can humble yeast be modified to produce the active substances that some believe to have therapeutic benefits? Hearing aids for cocktail parties One of the most impressive properties of the human auditory system is the way most of us can overhear or eavesdrop on specific voices in an otherwise crowded room. Most hearing aids can’t help with that: they can sometimes filter out noises that are not human voices, but cannot do the very human trick of sorting one voice from a sea of others. Nima Mesgarani from Columbia University reports in the journal Science Advances a proof of principle for a device that might be able to do just that. Firstly, a new algorithm can separate out one voice from another. Then brain waves from the wearer could be used to recognise which of those voices they are trying to hear. Then it’s a simple case of turning that voice up, and lowering the volume of the others, all in nearly real-time. Automatic face recognition So called Neural Network computing techniques are revolutionising our lives. They are able to perform a host of tasks that not so long ago would be the preserve of human brains, and to process huge sets of data and “learn” very quickly. One of the things they are proving exceptional at is face recognition; being able to identify faces in a crowd, or on a street, from a set of images provided by a user. But with great computing power comes great computing responsibility. What are the implications for policing and personal privacy? Gareth discusses these issues with Stephanie Hare. Producer: Alex Mansfield

314 episodes