filmov
tv
Neuroscientists recovered speech from brain signals
Показать описание
Researchers reconstructed songs of birds by recording the brain activity from a sensory-motor nucleus.
#Neuroscience #Brain #YuriNeuro
Timecodes:
0:00-Introdudction
0:28-Original publication
0:36-Biomechanical model and FFNN
1:30- Results of the song reconstruction by biomechanical model and FFNN
2:47- Song reconstruction directly from the brain activity using LSTM neural networks
3:21- Results of the direct prediction from LSTM. Comparison to biomechanical model
3:58- Further advantages of the combination of FFNN and biomechanical model
4:27- Potential applications to humans
5:47- Share your ideas
Neurobiologists from University of California, San Diego recorded neuronal activity in the sensorimotor part of the brain (HVC) of songbirds, which controls the muscles responsible for singing. Also, scientists employed a biomechanical model of the vocal organ that allowed to link the brain activity to the ultimate changes in pressure and tension in the vocal organ of birds during singing. By using used a shallow feedforward neural network (FFNN) with one hidden layer neuroscientists estimated parameters of the biomechanical model and used this model to reproduce the song of the bird.
The spectrograms of the songs produced by the birds and predicted by the model exhibited a close similarity. Also, neuroscientists used recurrent, long-short-term memory neural network (LSTM) to reconstruct the song directly from the brain activity. In summary, the best result was achieved by combination of a biomechanical model and FFNN to estimate model parameters.
This techonolgy provides an opportunity to capture the voice with its pitch, volume and timbre in real time from the analysis of the brain activity, which is feasible for humans. This scientific work makes an important step toward the development of the vocal prosthesis, that in the future may allow people who lost their voices to get back an ability to speak without a temporal delay, as well as to exhibit emotions.
Explore other intersting facts about the brain by checking the other neuroscience videos
Train your brain by solving logical and math puzzles
Original publication:
#Neuroscience #Brain #YuriNeuro
Timecodes:
0:00-Introdudction
0:28-Original publication
0:36-Biomechanical model and FFNN
1:30- Results of the song reconstruction by biomechanical model and FFNN
2:47- Song reconstruction directly from the brain activity using LSTM neural networks
3:21- Results of the direct prediction from LSTM. Comparison to biomechanical model
3:58- Further advantages of the combination of FFNN and biomechanical model
4:27- Potential applications to humans
5:47- Share your ideas
Neurobiologists from University of California, San Diego recorded neuronal activity in the sensorimotor part of the brain (HVC) of songbirds, which controls the muscles responsible for singing. Also, scientists employed a biomechanical model of the vocal organ that allowed to link the brain activity to the ultimate changes in pressure and tension in the vocal organ of birds during singing. By using used a shallow feedforward neural network (FFNN) with one hidden layer neuroscientists estimated parameters of the biomechanical model and used this model to reproduce the song of the bird.
The spectrograms of the songs produced by the birds and predicted by the model exhibited a close similarity. Also, neuroscientists used recurrent, long-short-term memory neural network (LSTM) to reconstruct the song directly from the brain activity. In summary, the best result was achieved by combination of a biomechanical model and FFNN to estimate model parameters.
This techonolgy provides an opportunity to capture the voice with its pitch, volume and timbre in real time from the analysis of the brain activity, which is feasible for humans. This scientific work makes an important step toward the development of the vocal prosthesis, that in the future may allow people who lost their voices to get back an ability to speak without a temporal delay, as well as to exhibit emotions.
Explore other intersting facts about the brain by checking the other neuroscience videos
Train your brain by solving logical and math puzzles
Original publication:
0:06:04
Neuroscientists recovered speech from brain signals
0:14:24
After watching this, your brain will not be the same | Lara Boyd | TEDxVancouver
0:11:05
You can grow new brain cells. Here's how | Sandrine Thuret | TED
0:04:50
How a Brain Implant and AI Gave a Woman with Paralysis Her Voice Back
0:01:33
Scientists create speech using brain recordings - TomoNews
0:53:33
Your Brain: Who's in Control? | Full Documentary | NOVA | PBS
0:20:49
A critical window for recovery after stroke | John Krakauer | TEDxJohnsHopkinsUniversity
0:00:51
3 Ways to Grow New BRAIN CELLS.
0:08:01
I Found My True Self by Learning Another Language
0:03:01
Speech From Brain Signals
0:02:10
He Found His Voice—and His Brain Map
0:01:00
How a Brain Implant and AI Gave a Woman with Paralysis Her Voice Back
0:13:03
Wendy Suzuki: The brain-changing benefits of exercise | TED
0:00:11
🧠 The Brain in Detail: Exploded View Anatomy 🧠 #anatomy
0:00:37
Neuroscientist: How To Self-Regulate | Andrew Huberman #neuroscience #shorts
0:09:45
Restore your brain with nature | David Strayer | TEDxManhattanBeach
0:00:31
Neuroscientist: How To Remember Better | Andrew Huberman #neuroscience #shorts #hubermanlab
0:10:09
REWIRE YOUR BRAIN - Neuroscientist Explains How To Control Your Mind in MINUTES!
0:17:53
How mindfulness changes the emotional life of our brains | Richard J. Davidson | TEDxSanFrancisco
1:19:13
The Brain and Recovery: An Update on the Neuroscience of Addiction
0:12:15
Never Enough: The Neuroscience and Experience of Addiction | Judy Grisel | TEDxPSU
0:01:58
How Does Your Brain Decode Garbled Speech?
0:00:33
Neuroscientist: How To Stop Being Lazy | Andrew Huberman #joerogan #neuroscience #shorts
0:00:56
Neuroscientist: Why Do You Feel Burnt Out? | Andrew Huberman #neuroscience #hubermanlab
Комментарии