Scientists can predict what a bird will sing next

By Mark Waghorn via SWNS

Scientists have predicted what a bird will sing next - by monitoring its brainwaves.

The breakthrough opens the door to a mind-reading device for those who have lost the ability to speak.

The AI (artificial intelligence) technique detected neuronal activity in zebra finches using electrodes.

Information was fed into a machine learning program that was trained to match sounds that follow.

The Team from the University of California, San Diego, expect it to lead to the development of vocal prostheses for humans.

Current state-of-the-art implants allow the user to generate text at a speed of about 20 words per minute.

Project co-leader Professor Timothy Gentner said: "Imagine a vocal prosthesis that enables you to communicate naturally with speech - saying out loud what you are thinking nearly as you are thinking it.

"That is our ultimate goal, and it is the next frontier in functional recovery."

Birdsong and human speech share many features. Both are learned from older members of the species.

They are also much more complex than other animal noises - such as the grunts of monkeys.

First author Daril Brown, a Ph.D. student in computer engineering, said: "Our work sets the stage for this larger goal.

"We are studying birdsong in a way that will help us get one step closer to engineering a brain-machine interface for vocalization and communication."

In the study, silicon implants recorded the firing of brain cells as male adult zebra finches went through their full repertoire.

In particular, a set of electrical signals called 'local field potentials' - necessary for learning and producing songs - were analyzed.

They've already been heavily studied in humans - and are being used to predict their vocal behavior.

Project co-leader Professor Vikash Gilja said: "Our motivation for exploring local field potentials was most of the complementary human work for speech prostheses development has focused on these types of signals.

"In this paper, we show there are many similarities in this type of signaling between the zebra finch and humans, as well as other primates.

"With these signals, we can start to decode the brain's intent to generate speech."

Different features translated into specific 'syllables' of the bird's song - showing when they will occur.

Brown said: "Using this system, we are able to predict with high fidelity the onset of a songbird's vocal behavior - what sequence the bird is going to sing, and when it is going to sing it."

The researchers even anticipated variations in the song sequence - down to the syllable.

It can be built on a repeating set of four for instance and every now and then change to five or three. Changes in the signals revealed them.

Gilja said: "These forms of variation are important for us to test hypothetical speech prostheses because a human doesn't just repeat one sentence over and over again.

"It is exciting we found parallels in the brain signals that are being recorded and documented in human physiology studies to our study in songbirds."

Conditions associated with loss of speech or language functions range from head injuries to dementia and brain tumors.

Project co-leader Prof Timothy Gentner said: "In the longer term, we want to use the detailed knowledge we are gaining from the songbird brain to develop a communication prosthesis that can improve the quality of life for humans suffering a variety of illnesses and disorders."

SpaceX founder Elon Musk and Facebook CEO Mark Zuckerberg are currently working on brain-reading devices that will enable texts to be sent - by thought.

The study is in PLoS Computational Biology.