The biggest change in Human Computer Interaction over the past few years is the rise of voice assistants. The Siris and Alexas are our HAL 9000s, and soon we’ll be using these assistants to open the garage door. They might just do it this time.
What would happen if you could talk to these voice assistants without saying a word? Would that be telepathy? That’s exactly what [Annie Ho] is doing with Cerebro Voice, a project in this year’s Hackaday Prize.
At its core, the idea behind Cerebro Voice is based on subvocal recognition, a technique that detects electrical signals from the vocal cords and other muscles involved in speaking. These electrical signals are collected by surface EMG devices, then sent to a computer for processing and reconstruction into words. It’s a proven technology, and even NASA is calling it ‘synthetic telepathy’.
The team behind this project is just in the early stages of prototyping this device, and so far they’re using EMG hardware and microphones to train a convolutional neural network that will translate electrical signals into a user’s inner monologue. It’s an amazing project, and one of the best we’ve seen in the Human Computer Interface challenge in this year’s Hackaday Prize.
Hmm, so my lucid dreams in vocalising could now be recorded too, great stuff as a bit more articulation of subconscious thoughts could be entertaining, thanks for post :-)
Makes me think of MIT’s Alter Ego: https://www.media.mit.edu/publications/alterego-IUI/
This reference will go well here (note the link to the Alterego HaD article):
https://hackaday.com/2019/01/11/you-wont-hear-this-word-on-the-street/
” NASA is calling it ‘synthetic telepathy’.”
Since we seem to do a lot of thinking in words anyway, is there a much of a border between advanced subvocal recognition and “real” telepathy?
Does thought even exist without words? Hard to say.
Intuituion fuctions without the need for internal narrative.
Well yah, you can visualise a thing, rotate a model in your head, that sort of thing, or imagine non-verbal sounds, music.
I’ve heard about people like you. There is a lot of humans who can do all manner of complex and logical cognition tasks and communicate without words, in fact they can do things that are near impossible with just words. BTW The word is not the base unit of verbal or written meaning anyway, n-grams are, i.e. words in context.
…People like me, huh. What do you mean by that? And isn’t “words in context” obviously implied when somebody talks about words and thought? I don’t see why we have to be so specific as to bring that term into play. I thought that term was more for computers and language than human psychology anyway.
What sorts of complex logical cognition tasks and communication are you thinking of? Any examples?
Whoooosh…
Are you talking about everyone who missed my “hard to say,” or did I miss some esoteric joke myself?
Ms. Ho and her team might be well served by researching the scientific literature to see what’s already been done in this field. There are a quite a number of journal and conference papers on this topic. They might find that if they continue down the path suggested by their electrode placement, they will likely not be that successful.
I’ve never really understood the target project complexity for the Hackaday prize. With all due respect to this project team, I give their project a very small chance of success, but I think it’s a great project that is rightfully highlighted and could definitely benefit from the exposure. But then there are other projects that are highlighted as Hackaday prize entries in the blog that are things like connector interface boards with two connectors and few to no components. While these types of projects are of general interest, I don’t understand why the blog posts highlight them as being associated with the contest.
Give us time.
We would love to prove you wrong. )
I think that would be great too. There’s nothing wrong with having a small chance of success – it means you’re working on a difficult problem.
I hope they succeed, as it would be widely applicable, but my intuition is asking me why this is not already a consumer product if it is doable.
We are not only one group who is interested in this topic. There are NASA, MIT, NIH and other teams who are working on it. Under the hood, problem is very similar to traditional speech recognition.
Breakthroughs in deep learning in last 3 years (tensorflow in particular) made this project possible. Give it couple years and it will appear on the consumer market.
Hopefully our team will help to make it happened. :)
About 2 months ago I dreamt that I played a major part in developing thought to speech tec. For what it is worth I have a great deal of cognitive power and strength. I am a intellectual and naturally so, as well as a vocalist. I am reasonably fine with performing accurate pronunciations and annunciations. I feel we could integrate aspects of the Mind Flex game. If you are not aware of that game see the link I have provided. The best way to contact me is via Facebook messenger not my Email. Thank you.
A link about Mind Flex
https://youtu.be/X3GUY7ZqS9E
Hmmm… as interesting as this technology may sound (and I find it fascinating) I can’t help being scared at the same time.
I did not expect NASA to be looking into this, but I can see darker agencies working on it.
If you interrogate somebody… and they won’t speak. Then this device can make then “talk”…
Other then that, I can’t wait to play with it, as it could change the way we work with computers. We all love speech recognition since the 80’s but it never caught on. Simply because we don’t like others to hear what we are doing OR because it is far from practical when everyone in the same room is commanding their computer through voice. This could break that barrier!
This is not new Hawkins used his eyes to speak,There are better methods,which I am not able to disclose,it is however good to see that it may not be long that the deaf can hear and maybe the blind can see.
Hmm ???? ???? so when you write “..and communicate without words” can you give specific examples from your keen observations or even diect experiences perhaps ?
And furthermore could you articulate an Experimental Method that avoids contamination through any potential placebo/nocebo influences ?
I wish them the best of luck on the project, but it’s been attempted before, and it simply isn’t practical. NASA toyed around with the concept in 2004, and achieved proof-of-concept but with very poor accuracy. A company called Ambient tried to commercialize this with their Audeo device around 2008. There’s a video on Youtube of the CEO using it to dial a phone number…and it takes him 5 minutes. Pattern recognition of EMG signals is just too error prone. More recently, MIT has been researching this with their AlterEgo device, and when last I checked, it had achieved 92% accuracy, but even that’s not good enough for them to commercialize the product.
Thanks! We actually had the opposite reaction when we came across the fact that NASA was working on it long ago. It’s 2018 now and machine learning models are so much better and we have so much more computing power available to us! The fact that there were failed attempts dating from over ten years ago just means it’s a matter of time until it succeeds.
And you’re right that AlterEgo is a purely academic project at the moment. We were so bummed to find that there’s nothing out there in the market for us to buy and try out so we decided to build the device ourselves. :)