Computer system transcribes words users “speak silently” | Hacker News Reader

Computer system transcribes words users “speak silently”

news.mit.edu

69 pointsanonytrary17 comments

Researchers have created a potential keyboard competitor. A headset that is worn, which picks up sound (?) and electrical signals from the jaw area, with a resolution of 20 unique signals. Testers have been able to communicate chess game moves "silently" with an accuracy of around 92%. Researchers seem optimistic that N > 20 in the future.

This reminds me of a relevant HackerNews discussion about biological interfaces a few days ago: https://news.ycombinator.com/item?id=16727719

I am so happy to see progress in this area. While we're still some ways away from replacing keyboards, this seems like a huge step in the right direction. AFAIK, a vocabulary size of 20 for inconspicuous communication is unheard of, but if this has been done elsewhere, please share links!

I’m so happy to see this. Keyboards allow me to have private conversations in an office setting, which is great because I am shy in person :-)

It’s one reason I suppose that we will have keyboards around for a while longer, despite advances in voice recognition. That and I can’t imagine trying to code with voice...

>Keyboards allow me to have private conversations in an office setting

I'm sure you already know this (since it's old tech), but, one can deduce what you're typing just by listening to your keyboard typing sounds.

https://www.schneier.com/blog/archives/2005/09/snooping_on_t...

Yep I am aware. And there are even easier methods: A $30 keystroke logger could be inserted on my keyboard USB plug and I’d never notice.

Great. Several people have pointed how this can essentially be used to read minds.

So it WILL be built, to keep the homeland safe. :-(

Almost forgot: “I love Big Brother!”

I doubt this could ever be used to arbitrarily read minds. Even if you could do the measurements without the targer knowing, you would have to somehow correlate the target's real thoughts with those measurements. In the article, the targets voluntarily provide that correlation to measurements during a training phase. It might be the case that some neuro-physical patterns might be common across targets, but that seems very unlikely.

If the target is in your custody and you coerce him to give training data, then he could just give garbage data. You tell him to think "fish barrel neuton whale", but he really thinks "somebody once told me".

Do sub-vocalizations only occur for certain categories of thoughts? I suspect that facial memories do not induce sub-vocalizations.

how can you possibly claim this so confidently, but sure keep believing its impossible, until its not, but then its too late to worry about anyways. This seems like an incredibly foolish opinion on this matter.

Maybe a more intrusive setting would be something like a polygraph. The subject would be first tested to ensure that the training worked to a high accuracy level (like the polygraph baseline measurement). You could tell to some extent whether the subject was cooperating with this training because you could see the resulting accuracy rates, and maybe continue the training until they improved, or else declare the subject to have failed to cooperate with the test.

Probably, as with an present-day polygraph, people could specifically train themselves to produce inaccurate results, but also presumably most people wouldn't have done so.

On the other hand, I don't know what level of language this system operates at so I don't know if you're detecting phonemes, words, phoneme features (like +voice/-voice), or what. This seems significant for understanding how usable it would be for "mind reading" if the subject's thoughts included many words that hadn't been individually trained. If you're just detecting conditions that you've already trained for, it doesn't seem like there's really anything new here (from the mind-reading point of view, as opposed to the UI point of view) compared to existing biofeedback or polygraph systems because you're mainly just detecting when a known physiological state is or isn't re-elicited, which is something the earlier systems could already do.

Edit: other commenters said they can only distinguish about 20 words, so presumably the technology has a long way to go before it would be relevant to semi-adversarial mind reading.

If nothing else, I suppose you could also train yourself to think in code.

That word you keep using - I don’t think it means what I think you think it means.

That sort of thing.

  Dave, although you took very thorough precautions in 
  the pod against my hearing you, I could see your lips 
  move.

How cool - as an MIT undergrad back in the 80’s I spent quite a bit of time trying to design a device for this purpose using ultrasound to probe what’s going on in the throat/mouth during subvocalized thoughts. For me the genesis was realizing I couldn’t “say” sentences in my head any faster than I could say them out loud, and suspecting that my brain was trying to subvocalize the words (send below-threshold activation signals to the muscles in my mouth and throat, and accidentally introducing speech rate limiting in the process). I’ve no formal proof if that’s the case, but it does appear they were able to make the concept work.

As cool as this kind of tech is, there are a lot of really troubling interrogation style abuses that become possible if you can hear a person’s subvocalized thoughts.

That said, I can think much faster than I can speak or type, and both communication forms interact differently with my thought process.

Speech recognition will never be a good input UI for me as I think _while_ typing — they’re in a tight feedback loop. Speech does not work this way; I can’t easily delete the last 3 words and reform the same sentence.

This would be amazing for stutterers

This could be a great tool for use on the battlefield. Reminds me of the subvocal mics from the Shadowrun RPG, which was really a key piece of equipment to buy for any character working in a group.

Was anyone else greatly concerned with this? I was worried as soon as I read the headline, and as I read through the article I was simultaneously fascinated and distressed as my fears were confirmed.

This technology is absolutely fabulous and has so many potential uses (far beyond the accessibility and military use mentioned in the article). Think about it: wearing a headset that can respond to your thoughts. An Alexa/Bixsby/Siri/Google/whatever Voice Assistant right inside your head, performing tasks and answering questions all without ever needing you to whip out your phone or speak aloud. I already wear a Bluetooth headset 24/7 because of my job. I know some people think in concepts and visuals, but I form my thoughts by verbalizing my them into words and always have an inner monologue going.

You could quickly and (most importantly) privately take notes just by thinking of what you want to remember. You could quickly send messages in loud environments without losing focus - if you're at a concert and want to message your friends to see where they're at, you just need to think it; no need for shouting or getting your phone out. You could do things in the middle of a conversation without the other party being aware and without drawing attention to yourself - take notes on the resort they went to on vacation or that book they recommended, look up the definition of a word or fact you're unsure of so you don't look like an idiot, send a quick message to your boss or spouse without interrupting conversation, lookup movie showtimes or restaurant hours. Keep a journal of your thoughts and you would never forget anything ever again - record your thoughts as you're brainstorming, falling asleep, driving, or shopping.

The possibilities grow when you consider using this technology in the future with others (especially tech that isn't yet widespread, well-developed, or even thought up yet). I see this technology being combined with some sort of augmented reality that acts as an alternative to our current mobile device displays. An obvious contender would be [the AR 'smart' glasses currently on the market](https://www.tomsguide.com/us/best-ar-glasses,review-2804.htm...) and in research/development. These technologies would help integrate the internet and computing power into our daily lives more seamlessly and less obtrusively.

AR displays could be navigated more effectively using your thoughts rather than tracking eye movement. Also, in addition to your voice assistant responding via audio they could respond with visuals on your private AR display. You could see an acquaintance on the street and as you're about to pass them, quickly think "What's their name, and pull up their contact/profile". When you're in an important meeting you could silently and privately pull up relevant items, like a job posting if you're in an interview.

The ability and very real possibility for this technology to become a common, everyday household fixture is what really concerned me. People don't care enough about their privacy; just look at all of the technology out there today like social media and Google where we freely give away our data to make our lives easier. I'm guilty of it too. The only thing standing between us and chaos is our fragile and naive trust that these corporations will handle our data honestly, morally, transparently, and properly. I really hope we change as a society and start taking our privacy seriously before more technologies like this become commonplace. Otherwise, companies providing it will be able to store and sell your thoughts. I can see companies tracking and selling your thoughts as you walk through the grocery store or as you watch TV, and it only grows darker from there.

Alright, I'm done with my rant and fear of a dystopian future. I just had to share because I've been thinking about it all morning since I woke up!

Seems no more dystopian in an assistant setting than voice input now: listen for a wake-phrase and relay what's said after that.

As gargoyle tech, it's even more promising than that Tap chorded keyboard-cum-knuckleduster.

Wow. This has some seriously scary implications for uses by repressive regimes. Literal thought policing.

Reply on news.ycombinator.com