LISTEN: AI recreates radio journalist's lost voice
A software company has brought a radio journalist’s lost voice back to the air with the power of neural networks.
American journalist Jamie Dupree suddenly lost his voice to a rare neurological condition in 2016.
The messages from his brain no longer reached his tongue, a condition so rare that there are currently no specialists to treat the condition.
Whilst searching for a cure the journalist, who covers American congress, has used a pen to suppress his tongue and turned to social media and written stories to make his voice heard.
Now artificial intelligence and neural networks have brought his voice back to life.
Cereproc, founded in 2005, reads out text with a human-sounding voice.
The technology is used by people who suffer from a range of speech conditions, but until recently users would have had to choose from one of the pre-created voices.
By harnessing neural networks, Cereproc’s software can now analyse the features of voice recordings to recreate it in the software.
The AI was given hours of Dupree’s previous work to analyse. In his latest blog, Dupree uploaded a demonstration of the results:
"AI techniques work quite well on small constrained problems, and learning to model speech is something deep neural nets can do really well," Chris Pidcock, CereProc's chief technical officer and co-founder, told the BBC.
Whilst the radio journalist continues the search for a cure, he plans to incorporate the technology into his broadcasts in the coming weeks.
“Yes, it will probably sound robotic to some of my listeners; but for the first time in two years, I will be back on the radio,” Dupree said in a blog.
“Jamie Dupree 2.0 is here – and I couldn’t be more excited about it!”