New AI Creates Human Speech From Brain Signals

Hello and welcome! My name is Anton and in this video, we will talk about a new study that was able to teach the AI to pronounce words generated by human brain.

Support this channel on Patreon to help me make this a full time job:

Enjoy and please subscribe.

Bitcoins to spare? Donate them here to help this channel grow!
1GFiTKxWyEjAjZv4vsNtWTUmL53HgXBuvu

The hardware used to record these videos:
Some of the above are affiliate links, meaning I would get a (very small) percentage of the price paid.

Thank you to all Patreon supporters of this channel
Specifically, great thanks to the following members:

Morrison Waud
Vlad Manshin
Mark Teranishi
TheHuntress
Ralph Spataro
Albert B. Cannon
Lilith Dawn
Greg Lambros
adam lee
Gabriel Seiffert
Henry Spadoni
J Carter
Jer
Silver surfer
Konrad Kummli
Nick Dolgy
Mario Rahmani
George Williams
Shinne
Jakub Glos
Johann Goergen
Jake Salo
LS Greger
Lyndon Riley
Lauren Smith
Michael Tiganila
Russell Sears
Vinod sethi
Dave Blair
Robert Wyssbrod
Randal Masutani
Michael Mitsuda
Assaf Dar Sagol
Liam Moss
Timothy Welter MD
Barnard Rabenold
Gordon Cooper
Tracy Burgess
Douglas Burns
Jayjay Volz
Anataine Deva
Kai Raphahn
Sander Stols
Bodo Graßmann
Daniel Charles Smith
Daniel G.
Eugene Sandulenko
gary steelman
Max McNally
Michael Koebel
Sergio Ruelas
Steven Aiello
Victor Julian Castillo
xyndicate
Honey Suzanne Lyons
Olegas Budnik
Dale Andrew Darling
Niji
Peter Hamrak
Matthew Lazear
Alexander Afanasyev
Anton Newman
Anton Reed

[Alorand]

This is both amazing and terrifying in a:
"Your brain signals betray you as a thought criminal, fleshling..."
kind of way.

[RSK]

**holds up tiny fish**
"Im not putting that in my ear."

[mikicerise]

"The spinach was a famous singer" is a fascinating way to misinterpret "Those musicians harmonise together marvellously". Spinach is a homophone for musician - they sound alike, showing that the information coded for in the EEG includes sound information. But singer sounds nothing like musician or spinach, or anything else in the correct sentence, showing that meaning is also coded for. It's almost as if the algorithm was listening on the one hand for a sound, thinking it heard spinach, and on the other hand a meaning, thinking it heard singer, when picking up the word "musician". The same is true for "famous" and "harmonise together". Famous picks up some of the sound information of harmonise and the meaning of together. It may very well be that in the speaker's brain these words are indeed linked in the same manner. Amazing.

[SoI_Badguy]

RIP Stephen Hawking, this kind of technology would've really made his life easier if it just came a little sooner.

[barrykaine]

There was a rock and roll band in the sixties called "The Ultimate Spinach", so there's that. I don't recall if they harmonized marvellously, however.

[SquirrelASMR]

If they can listen to my brain, theyd hear soo much messed up stuff.

[tumoril_]

This is probably the most horrifying video I’ve seen in a while just due to the implications that this could be used to police people’s thoughts.

[SoulofCinder]

Damn Anton, you are almost at 0.5 millions subscribers. Here is an early congrats!

[LadyD]

Makes me feel hopeful about what the future may hold with regards to communication with people unable to speak, such as ppl w non-verbal autism, head trauma patients, etc

[I]

If I could use my inner voice for.

6-Axis or 12 directions.

"Up, down, left, right, forward, back, turn left, turn right, roll up, roll down, roll left, roll right".

With up to 16 actions.

All simultaneously actuatabal.

With latency under 0.1 milliseconds.

A new level of Gaming could be created.

Could you imagine such a level of control over a mechanism.

Virtual or otherwise.!

The mind boggles.!

Excellent excellent video. 👍

[robertschlesinger]

Very interesting and worthwhile video. Many thanks for the link to the relevant paper.

[Ben-Rogue]

2020: People saying Okay Google to their phone
2030: People mouthing Okay Google to their brain implant

[AlexAlex-bzoc]

We'll finally hear machines feel the pain of existing

[irisweird]

Oh well. It's time to make a tinfoil hat to match with my corona mask. What an amazing times we're living, wonderful people.

[cobalius]

Isnt it logical, that each brain uses different patterns for thinking of the same things?

I mean, each brain starts kinda blank and has to come up with an own system, there's no direct comparision.

[gaiustesla]

It's a computer code thats designed to match signals. It "learns" by sorting between the noise and signal you want.

Eventually when the coders look back at the results, they adjust the code until it gives the feedback a human can say, yes thats it! When it decodes the signals this way, it picks out the word we wanted it to pick out.

The computer doesn't know though, its just a calculator that produces results that humans can interpret as something only humans could usually do. That people have tailored until the code replicates something enough to produce convincing results.

It seems like a bad idea to have an all seeing eye watching our every move all day, with no need to rest or feel pain or care, just simply making calculations based on what its been programmed to "look" out for. You can't argue with a talking wall. Theres no point in fighting it, it can't die. It can just be replaced.

Although it would certainly change lie detector testing.

[criticallook]

Years ago, there was a study that demonstrated the ability to visualize what a person was thinking by placing electrodes on the subject's throat.

It turns out that the little voice you hear while thinking simultaneously controls the muscles in the neck which enable speech.

The action of thinking causes micro tremors in the muscles of the throat, as if actually vocalizing the word, but at a much lower electrical intensity.
:

[iamanempoweredone]

It’s not pioneering. The military is now willing to turn this portion of the tech into public domain. This was militarized in the early 1970s. Jose Delgado, military contractor demonstrated it in the 60s

[glynbo]

I'm so glad I found this channel.

[walterlyzohub]

While reading the comments I realized a new kind of future job. Now we have people who already transcribe writings into data input. With this you can have specialists who watch, listen, or read and gets inputted into data centers. Avoid typing and carpal tunnel syndrome.