Hacker News new | ask | show | jobs
by Kenji 4152 days ago
Wow! Those screen readers are fast (his example https://soundcloud.com/freecodecamp)! I barely understood a word. I didn't know that.

It's a shame that so many programs don't follow the accessibility guidelines, but it's just too damn easy to forget about the disabled if you aren't. But this article was an eye-opener for me (no pun intended).

8 comments

I'm installing a screen-reader today ... and I'm going to listen to everything I create on it (at a slower speed - at least to start). We have stringent accessibility guidelines, tools to help find accessibility problems and a person in charge of making sure we adhere to those principles, but wouldn't it be more real to close your eyes and experience your creation yourself?
Thank you for doing this. And for saying this.
Since he has a whole occipital lobe to devote to parsing that, I don't feel so bad that my auditory processing isn't quite up to snuff. Reminds me of speed reading, though. What is the wpm on that? Maybe I'll ask him on freenode.
He informed me the clip is 525WPM. Impressive!
Is there evidence to suggest that a brain is that plastic? Genuinely curious. I feel like the "blind people have super-powered other senses" is an urban legend, like the 10% of your brain misconception.
The brain seems to be pretty darn plastic. There have been experiments with giving people whole new sensory inputs, and finding that the brain adapted to them quite well. Two examples:

- A grid of electrodes on the tongue, activating with a pattern fed from a camera. After a while, blind experimental subjects reported actually seeing what the grid displayed.

- A belt of buzzers hooked to a compass. The buzzer closest to magnetic north was always active. After a couple months, subjects weren't really conscious of the buzzing, but got a really good mental map of their environment and their position in it. They could navigate unfamiliar environments much better than before...and then much worse, when the experiment ended.

There are also some well-known cases of blind people using echolocation pretty effectively, without any special hardware.

Don't have links handy, unfortunately.

This American Life did an episode recently on Daniel Kish, one of the most famous echo location practitioners[0] and here's a short youtube video of the same guy.[1]

[0]http://www.thisamericanlife.org/radio-archives/episode/544/b...

[1]https://www.youtube.com/watch?v=A8lztr1tu4o

Echolocation is a skill that basically translates clicky sounds that bounce back into structural details of the obstruction the sound waves encounter. be it height, weight, distance and if you get good at it even material composition. Look at Dan Kish to see an example
This. Make any prediction of how advanced the brain is, then double that, at the very very least. It is barely short of magical. Even single neurons can do processing way more advanced than the artificial ones we use in machine learning. The rest of the body is honestly quite disappointing in comparison. Even the immune system and DNA/RNA mechanisms seem trivial and those are pretty dope compared to most other things.
Even single neurons can do processing way more advanced than the artificial ones we use in machine learning.

How does that work? A single neuron is a single neuron. My understanding is that the brain has lots of neurons and also they are assembled in certain "NN architectures" that are far more advanced than what we currently have. But I think that if you go the the single neuron level then they are pretty similar in terms of problem solving capabilities.

Well, a neuron can either fire or not fire; 0 or 1. For a neuron to fire, enough of its' synapses need to fire. How likely a single neuron is to do so based on an incoming action potential varies over time according to Hebbian learning. Once x synapses fire, the input is linearly summed to determine if the neuron as a whole fires. That's a neat abstraction and it is pretty descriptive. But it's also a little too neat to be true. The summation is, of course, not linear and neurons react to incoming signals even if they don't fire. How likely a neuron is to fire also depends on how much it has fired recently, as the synapses "get tired" (the concentration of certain molecules are temporarily exhausted), but for some neurons, likelihood increases as a result of previous firings before it decreases. Hebbian learning increases the strength of the connection to neurons that participates in successfully firing the neuron, but connections under a certain strength threshold decay with time, while those over it are stable (inability to create these stable connections is connected to alzheimers).

That's what I remember and I just took some neuroscience courses. There are multiple books out there attempting only to describe behaviour mathematically, let alone describing the underlying mechanisms, which is true "here be dragons" territory.

"Neuropsychologists compared the brain activity of people who can see and people who were born blind, discovering that the part of the brain that normally works with our eyes to process vision and space perception can actually rewire itself to process sound information instead."

http://www.sciencedaily.com/releases/2011/03/110316104123.ht...

the WPM on that is aproximatley 525
I actually thought it was a corrupt file or something - it doesn't sound like anything I'd be able to pick out and understand.

If someone can understand at that speed, listening to audiobooks must be a very quick activity...

That's a very interesting point! I would consider that a superpower. Would this also imply that humans are capabable of way more than one would normally think, given proper training?
It's just skimming. If you were reading a book, you would just scan across it to find what you want. You don't have to read everything on the page to know what it is talking about.
I would not say that. When you are skimming, you actually skip parts of the text. Since an audible text is a constant stream of data, just skipping parts would make it incomprehensible
In speed reading, you can read sentences backwards and the brain will reorder them in real time. You can learn to decouple image (symbol/word) acquisition from comprehension. Step one om this path is to stop subvocalization, saying words out loud in your head.

It's the same difference between Sync (O_DIRECT) and Async disk writes.

>Step one om this path is to stop subvocalization, saying words out loud in your head.

I've heard people say that before but I still don't understand how to do it.

You still see the text, even if you don't understand it. Your eyes are collecting a constant stream of data, too.
SoundCloud is evil and started playing bad songs right after this audio stopped.
Yeah, shocking after turning up the volume a bunch to hear the screenreader.
I have no idea if there is a way to stop it from doing that either
Good introduction to screen readers from Victor Tsaran of Yahoo [1] also shows his normal speed for browsing.

[1] http://yuiblog.com/blog/2007/05/14/video-intro-to-screenread...

I would compare this to using something like Spritz (http://www.spritzinc.com/) for a sighted person. At first the higher speeds feel too fast but your brain quickly adapts, especially with daily practice. Given that he's used a screen-reader his entire life when dealing with computers, I'm sure his brain has deeply entrenched pathways for parsing audio like this.

I found that it was easier to catch a few words from that file when I closed my eyes to shut out other distractions. I still couldn't understand much, but I could see how there would be a progression to build to that level.

Indeed, I used NVDA for a while when I was working on accessibility for a web app and after a week I could turn it up pretty fast, but nowhere near that fast.

(I'm not blind)

No kidding. The speed of that is incredible.