'I Tried Listening To Podcasts at 3x and Broke My Brain' (medium.com) 82
'Podfasters' listen to their favorite pods at 1.5x, even 2x speed. But how fast is too fast? From a report: Bumping the speed up to 1.5x was initially jarring. People were talking so quickly that I had to stop what I was doing and focus on the audio to keep it from falling into background chatter. After about 20 minutes of this intentional listening, however, it felt like my brain had adjusted. What at first felt rushed and slightly wrong, now felt natural. Once I found that I could go back to doing the things I normally do when I listen to podcasts -- brush my teeth, do the dishes, fold the laundry -- I bumped up the speed another notch to the 2x barrier. Like the previous jump in speed, the first 15 to 20 minutes required an additional level of focus to get my brain to match the cadence of the conversation. But once I was there I felt like I didn't have to strain to understand what was being said -- my brain just "learned" how to listen to this accelerated pace. In our discussion of breaking 2x, Uri Hasson, director of Princeton's Hasson Lab, brought up one population that handles sped-up speech much better than the rest of us: the visually impaired. A 2018 University of Washington study attempted to quantify human listening rates by measuring the intelligibility of audio from a text-to-speech generator played at increasingly faster speeds. Researchers found that the average sighted person could comprehend around 300 words per minute, or about double the average talking speed of an American English speaker. Visually impaired subjects, however, vastly outperformed sighted subjects at speeds past 2x, demonstrating comprehension at rates even approaching 3x.
The researchers hypothesized that this difference between sighted and visually impaired listening rates was attributed to one group being more familiar with synthesized text-to-speech voices. At 2x, the experience of listening to audio began to change: Though I could understand the words, they seemed to have less emotional resonance. At these high speeds, my brain seemed to shift away from assessing people's feelings towards baseline comprehension. At the end of each sentence, I'd feel a little twinge of joy, not because of anything happening in the podcast, but just because I had understood the words. Hasson points out that single word comprehension is really only one dimension of comprehension. Our brains do not work like computers. We can recognize words very quickly, but to integrate them into a sentence, a sentence into a paragraph, and a paragraph into a larger narrative takes time. Feeling competent in my base-level comprehension at 2x, I crossed the threshold into 3x. It took every ounce of concentration to just register what was being said. After 20 minutes, my brain couldn't settle into the rhythm of the conversation. I sat there for an hour, with my eyes closed, hoping that my brain would eventually "click" like it did before, but it refused.
The researchers hypothesized that this difference between sighted and visually impaired listening rates was attributed to one group being more familiar with synthesized text-to-speech voices. At 2x, the experience of listening to audio began to change: Though I could understand the words, they seemed to have less emotional resonance. At these high speeds, my brain seemed to shift away from assessing people's feelings towards baseline comprehension. At the end of each sentence, I'd feel a little twinge of joy, not because of anything happening in the podcast, but just because I had understood the words. Hasson points out that single word comprehension is really only one dimension of comprehension. Our brains do not work like computers. We can recognize words very quickly, but to integrate them into a sentence, a sentence into a paragraph, and a paragraph into a larger narrative takes time. Feeling competent in my base-level comprehension at 2x, I crossed the threshold into 3x. It took every ounce of concentration to just register what was being said. After 20 minutes, my brain couldn't settle into the rhythm of the conversation. I sat there for an hour, with my eyes closed, hoping that my brain would eventually "click" like it did before, but it refused.
No problem here (Score:2)
Just sounds like any french waitress in Paris.
Re:No problem here (Score:5, Informative)
There's actually a paper looking at different languages and their information bandwidth. It turns out that some languages (like French) involve speaking much faster, but they are also less succinct. The speed at which languages transmit information tends to be pretty similar.
Turned around, any student will wryly observe that it seems to take the French twice as many words to say anything, but they say the words twice as fast, so it cancels out.
Re: (Score:2)
There's actually a paper looking at different languages and their information bandwidth. It turns out that some languages (like French) involve speaking much faster, but they are also less succinct. The speed at which languages transmit information tends to be pretty similar.
It wouldn't surprise me if different languages inherently have different average or maximum rates of intelligible information transmission. However, this thread relates to the maximum intelligible increase relative to the average rate. In my personal experience, the sped-up speakers I understand more easily are naturally slow speakers and speakers who have accents that match what I'm used to.
Re: (Score:2)
They don't.
It would be quite surprising if that were not true.
Re: (Score:2)
Italian, Japanese, Chinese, Thai have a higher information yield per second/minute than e.g. english or german.
For Jap. and Chinese it is even more frappant when reading it.
Re: (Score:2)
The only thing I think about it. Train you brain to listen twice as fast and now enjoy the sadistic pleasures of listening to the really slow in normal speed, in real life interactions, the thought makes my skin crawl, the desire to scream "get to the fucking point" running around over and over in my brain. I already spend much time intuit the conclusion of slow speakers to arrive to the point well before the slow thoughts get there. Imagine training myself to make it ever worse.
Re: (Score:3)
Re:No problem here (Score:5, Insightful)
There is one situation where running the audio at 3x speed is useful, and there's no loss of information: Any Youtube HOWTO video where some wanker records a ten-minute-long video containing around ten seconds of actual useful information hidden somewhere in it. And then Google indexes it and returns a ton of this crap when you try and find information on how to solve a particular problem.
Come to think of it, you need a 10x speed play option for those, along with a Cattle Prod button next to the like/dislike one.
Re: (Score:1)
I tried reading slashdot at 3x (Score:5, Funny)
When I first read Slashdot, I used 1x, and words had meaning and intonation and emotion.
Then I went to 2x. Not for any reason, mind you, but because I could. It required that I actually pay attention to what I read.
It was amazing. I found if I actually thought about things, they meant more! Amazing!
Then I sped it up to 3x and found that really I was just skimming words, doing that Evelyn Wood
speedreading thing, and nothing meant anything. Finally I could read slashdot without caring
what I was reading about, understanding any of it, and commenting while ignoring what anyone
else had ever posted.
There's a moral here somewhere.
E
Re: (Score:2)
Re: (Score:2)
The moral is that a story summary should not be an enormous wall of text.
But, actually summarizing an article, rather than lifting multiple paragraphs right out of the text, takes work.
Depends on the podcast source (Score:5, Funny)
If it's one of those shock jocks who talk real fast, sure.
If it's complex stuff that they normally report at a quick pace, sure.
But
If it's an NPR host taking their time to discuss something, 3x is fine. .. ok, that's just hilarious.
If it's slow jazz with a smooth voice, 3x is ok.
If it's a meditation podcast
Re: (Score:3)
I tried Ben Shapiro on 3x and my speakers just made their weird buzzing noise.
Re: (Score:3)
I think you must have done it wrong, that's what he sounds like normally.
Re: (Score:3)
Re: (Score:3)
I watched a Vi Hart video at 3x and almost went plaid.
Re:Depends on the podcast source (Score:5, Funny)
Comment removed (Score:4, Interesting)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
speech synth output at speeds that I couldn't follow at all. Not sure just how fast it was, but I think it was well over 3x.
-jcr
If it's synthesized speech, then what's the "x" you're basing it off of, exactly?
Were they getting speech output in 1/3rd the time it took the text that generated it to be drawn on screen?
Re: (Score:2)
I presume that in context, "3x" means thrice the median speech rate for the language.
Re: (Score:2)
I've worked with blind engineers a couple of times in my career, and they all listened to speech synth output at speeds that I couldn't follow at all. Not sure just how fast it was, but I think it was well over 3x.
-jcr
Sounds like you have some mentats working there. Have you verified their conditioning for loyalty?
Re: (Score:2)
There are a few different types of test-to-speech technology, and not all are equally intelligible when sped-up.
The most modern, most natural sounding, form of TTS uses sequence-based neural nets (e.g. Google's WaveNet) trained to predict individual audio samples. A naive implementation is very computationally demanding since good quality speech requires a sample rate of around 16KHz (i.e. neural net needs to make 16,000 predictions per second to generate audio in real-time).
Prior to sample-by-sample neura
"News Read Real Slow" (Score:1)
News read real slow [podbay.fm] The Dave Ross Show June 13, 2008, podcast (oh the irony)
Also Dave Ross, from 2013, News read real slow: The Bob Filner edition [mynorthwest.com]
Which is why I don't listen to podcasts (Score:3)
Re: (Score:1)
Could you slow down a little? I couldn't understand that.
Re: (Score:3)
Agreed. There are podcasts that are one-on-one conversations that work really well. Eric Weinstein's The Portal is like that, and I wouldn't want to listen to it at anything other than normal speed. However, there are a lot of podcasts that are really just audio from some schmuck's YouTube channel, and these are painful to listen to at any speed, much less watch as they hem and haw around a point.
Rather than shit out an hour a day of you saying "like, you know, uh, so..", sit down, collect your thoughts, an
Re: (Score:2)
Podcasts and their video cousins are terribly inefficient communication media. Besides the natural inefficiency of spoken language, many of them seem to have decided that "a conversation" is better than an organized presentation of information, so you have to deal with crap like discussing the weather as well.
Re: (Score:2)
I have a hard time reading when I'm driving or mowing the lawn or building a fence. Listening to a podcast though is easily done at the same time as doing those things.
Re: (Score:2)
It's not about the efficient transmission of information. Where'd you get that assumption?
Most podcasts are "friend substitutes". Lots of people out there feel sad if they're not in a group, chattering away. Podcasts let them feel that way. It's not as good but it's good enough. It's about *feelings*. People's feelings are super important to them. Curling up with a book and reading in silence for hours seems like death to them. They want to be in the middle of a babbling group, chattering away like monkeys
Yeah language sucks... (Score:2)
"Our brains do not work like computers. We can recognize words very quickly, but to integrate them into a sentence, a sentence into a paragraph, and a paragraph into a larger narrative takes time"
That is because we are all stuck with "culturally significant" languages with all the baggage of the past that linguists love or saddle with concerns about being easy to learn. Esperanto is a little better as a universal spoken language but a new language designed by engineers and neurologists for fast, universal,
Re: (Score:2)
Clearly you are not a poet.
" a new language designed by engineers and neurologists for fast, universal, efficient, and unambiguous information exchange would be much much better."
"Much much better" for WHAT though?
Plus any language constructed like that automatically invites comparisons to newspeak. (https://en.wikipedia.org/wiki/Newspeak)
Whom do you trust to dictate the framework in which people will think, to decide what ideas shall have words ? Because it'll be whoever has the most money to throw around
Re: (Score:2)
"Much much better" for WHAT though?
Let me repeat what I quoted... in the language we use now context is fairly critical.
"Our brains do not work like computers. We can recognize words very quickly, but to integrate them into a sentence, a sentence into a paragraph, and a paragraph into a larger narrative takes time"
For parsing information, quickly, efficiently, and without ambiguity, GLOBALLY. In the modern world connected 24/7 to almost everyone else in the world by near lightspeed links the value here woul
Re: (Score:2)
"For starters nobody else will understand you"
Don't be daft. New words and new meanings enter our language continually with no central authority doling them out and it's certainly not the case that nobody else understands them.
I don't have to invent a way to enforce RFCs. No such enforcement exists and yet your computer is parsing the text I'm typing with mine just fine.
Lousy example; slashdot has some issues with just that: âoeWeâ(TM)ve been complaining about it for years!â
You say that as if it is different than now. Except it is worse because we redefine language and in so doing redefine the words of others from the past written under different definitions.
There is no central authority that dictates what words are allowed, or what single idea they are allowed to convey. What you are proposing is drastically different in that not only do you want centralized cont
Re: (Score:2)
"Don't be daft. New words and new meanings enter our language continually with no central authority doling them out and it's certainly not the case that nobody else understands them."
Don't be daft I'm specifically proposing something DIFFERENT than how language works today intended to address broken behavior, you can hardly cite existing broken behavior as a flaw in an idea intended to resolve that very problem.
"The problem RFCs are those that have large moneyed and motivated interests"
Language has moneyed
Re: (Score:2)
"Don't be daft I'm specifically proposing something DIFFERENT than how language works"
You originally proposed a different way of designing a language - that's doable. But it's a much bigger stretch to propose that one can enforce a change in how people will use language.
"you can hardly cite existing broken behavior as a flaw in an idea intended to resolve that very problem."
The existing broken behavior; that of re-purposing works and ideas and for phrases organically for artistic, poetic, ironic, lyrical, sarcastic, vulgar, and other purposes is 'human nature'. You aren't fixing that by designing a new language. I think you'd need to redesign humans first.
How do you have Orwellian control of published voluntary standards?
The minute they become regulated mandat
Re: (Score:2)
Spoken language isn't for information transfer. That's what written language and pictures are for. Spoken language is for emotions, social bonding, and such.
Ask an engineer or a hard scientist to explain something and the first thing they do is reach for a pen or chalk or something else to draw pictures (or math) with. My sister made me a nice hand sewn leather cover for the disposable notebooks I carry around for just that purpose.
Re:Yeah language sucks... (Score:4, Insightful)
"Spoken language is for emotions, social bonding, and such."
Spoken language is for communicating emotions and everything you communicate is information not just the things that make your eyes glaze.
"Ask an engineer or a hard scientist to explain something and the first thing they do is reach for a pen or chalk or something else to draw pictures (or math) with. "
You aren't contradicting me but making my point. Language is a tool for communication, that same as writing (which is actually a way of recording spoken language), diagrams, and in some cases math. Currently it is a poor tool because it began with common grunts being repeated in the same context and just had further additions bolted on from there.
You imagine when I say engineered that only engineering information would be transferred with it. That isn't the idea at all. The idea isn't that you wouldn't be able to express feelings, concerns, or socially bond... the idea is that the language you use to express your ideas be a more efficient and precise tool for communicating feelings, concerns, social ideas, OR engineering data... to anyone in the world because we'd no longer have the old tower of babel issue. The language would be less dependent on context so less prone to someone missing the context and require less listening before understanding a context. This would also make it easier for computers to process, understand, and communicate with.
Bridges, trains, and traffic management systems are engineered as well, that makes them more efficient than random narrow old european roads that grew up in hodgepodge fashion much like language (though not nearly as badly over as long a period). Being more efficient makes them better for efficiently moving things around and managing more cars it ALSO makes them more efficient for going and spending time with friends and loved ones.
I listen at 1.5x all the time (Score:2)
Doesn't even seem off to me, except when someone calls in with a particular accent or voice. I usually only notice when music plays at the wrong tempo.
Every time someone hops in the vehicle with me they can't handle it - it completely blows their minds and we have to slow it to 1x if they actually want to listen, and it's been to date at least five people I've had to slow it down for. My wife thinks I'm just wired for it, but much more than 1.5x is a little more than I can handle.
Roxannne... (Score:2)
Effects of accelerated listening on everyday life (Score:2)
I occasionally listen to speeds up to x1.4 (*rarely* x2, and mainly for review), but I feel like it started to have a damaging effect on me. It felt like it was making it harder to focus on real conversations happening at "normal" speeds. I already trouble with my focus drifting if I'm not careful, so I largely went back to normal speeds for fear of having a negative impact on my interactions with others. (Though I do still use accelerated speeds for some sources. x1.2 is pretty normal for me for sources th
Re: (Score:1)
I'm not sure about conversations, but I do find it difficult to watch a TV show or movie with friends now if it's not sped up. I generally watch youTube or listen to podcasts at 2x-2.5x, but I'll have to start paying attention to see if that's been impacting my interpersonal interactions. Thanks for making me think about how my habits impact the people around me.
Re: (Score:1)
Re: (Score:1)
That was another effect though... I felt like I was generally becoming more impatient. And it seemed like any slim benefits I was getting from absorbing more content (or just absorbing it faster) was outweighed by the more frequent (and frequently unwarranted) feelings of frustration and impatience.
Again, not that I never listen at faster speeds now. I just moderate it more. I'd hate to get to a point where I couldn't just go see a movie in theaters because the pacing would frustrate me.
Re: (Score:1)
Re: (Score:1)
Ultimately up to you, but there's something to be said for viewing less but enjoying what you do view more fully, or even just allowing yourself time to experience and react to something. It can be like taking time to savor and really enjoy a meal instead of scarfing it down so fast that you can barely taste it.
(Though as someone who frequently is watching TV while doing something else, I really don't have that much room to talk. And the value of taking it relatively slowly may depend a lot on what it is yo
I doubt 3x is some kind of hard limit. (Score:4, Interesting)
It might be like many other training scenarios: you progress it takes exponentially more time to achieve the same linear improvement.
Suppose you decided to start lifting weights. Before training a typical adult man can usually deadlift around 70 kg, or 155 pounds. Following a simple regular training program you can easily double that in a year to 140 kg. But following that same regimen might take you *two* years to up the next 70kg to 210. And that same regimen will never get you to 280 kg -- you'll age faster than you progress.
But 280kg/617 pounds might well be within your genetic potential, you just can't get there using brute force. You need the aid of *science*. Drugs, of course, are a huge help.
I suspect that if you kept at it, you could increase your comprehension rate beyond what is achievable in a few days of brute force effort. But there will be some point where progress by brute force will become so slow it might as well not be happening at all. That doesn't mean a more scientifically designed program wouldn't continue to make progress. It's also possible that some drugs could exhibit a nootropic effect. I'd consider stimulants a good candidate, since there is literature to support their use for improving processing speed in other kinds of tasks.
we should use machine learning here (Score:2)
I find it hard to listen past time 1.5. Mostly because depending on the type of content delivery could already be fast. Also speeding up the sound distort the pitch of the voice which is important to interpret it.
I wonder how hard it could be to recalibrate the podcast to a faster speech pattern rather than just speeding up the sound. For instance, obama was famous for having long pauses in his speeches. Just reprocessing the sound to remove the "mmm" and shortening the pauses would already give you a 2x o
Re: (Score:2)
Machine learning is probably overkill here. There are devices in any studio that can speed up sound in the way you describe - just play the speech back at X times the speed and then shift the pitch down by log2 X octaves. It doesn't deal with spoken glitches (uhms and uhhs), but a good announcer doesn't do that anyway or they do it infrequently enough that it can be handled in post.
If you do want to apply ML to this, just put the speech through a speech-to-text translator and play the text back through a ma
Useful for school (Score:2)
Comprehension (Score:5, Insightful)
The problem isn't that of "how fast can you listen" (or read), but how fast before you really are not comprehending any of it. Part of comprehension is assembling what you are hearing (or reading) into concepts you can really understand, analyze, and remember. I suspect that most who speed things up a lot really have no idea how much they might be impairing their resulting comprehension of the material.
I know this will vary from person to person, what else one is trying to do at the same time, how fast the speaker talks (and with how much clarity and precision), but a little time to reflect on what you are hearing, while you are hearing it, is important.... especially if it is a long presentation or contains many different and/or difficult concepts.
I would love to see actual studies that give comprehension and retention tests after various "consumption rates" to see just how much it is affected (on average).
Re: (Score:2)
I'd bet you retain much less. I'm not aware of any studies for audio but if Comic Sans improves retention then I'd guess the same might be true for audio.
https://www.princeton.edu/news... [princeton.edu]
I actually like podcasts and taped lectures because you can pause and contemplate what was just said if it seems important or hard.
I think it really depends on the podcast. Talkative podcast with lots of descriptions of the where they are, what the weather is like and how the interview subject is sipping his coffee... oh I
Sped up Youtube (Score:2)
Depends on the content and familiarity (Score:2)
There's even various browser extensions available for speeding up any HTML5 video content using keyboard shortcuts. I find that lectures, educational and critical thinking content -- whether podcast or video -- gets difficult for me to consistently parse beyond 1.5x, even when paying sole attention to it.
However, inane gameplay commentary I have no trouble taking in at 2.25x while multitasking, and eventually 2.75x once I get used to the person's voice.
Broken Brain?!?!? (Score:4, Informative)
It is exactly like a computer (Score:1)
Hasson points out that single word comprehension is really only one dimension of comprehension. Our brains do not work like computers. We can recognize words very quickly, but to integrate them into a sentence, a sentence into a paragraph, and a paragraph into a larger narrative takes time.
It is exactly how computers work. For example imagine you want to watch video at high frame rate. Normally you play at 30 FPS, but you have the bandwidth for 120 FPS. At 60 FPS, your decoder isn't fast enough, but you can turn off some features, so the quality is not as good. At 90 FPS, even at the minimum setting, you have tearing, at 120 FPS, you start dropping frames.
The brain works the same way, the "speech recognition" part is fast, but at max speed, the processing part works in degraded mode and skips
blind programmers (Score:2)
try this with blind programmers that are accustomed to screen readers that read out their code at seemingly inhuman speeds.
example: https://www.vincit.fi/en/softw... [vincit.fi]
Re: (Score:1)
I had to listen to that audio clip three times before I understood it.
It is surprising how bad text to speech is, or at least how different it is from normal human speech.
Distribution Chain (Score:2)
3x Is slow (Score:1)
I don't understand what this guy is talking about (and I hate this whole 'broke the x' meme that is happening.)
I listen to 90% of the audiobooks and podcasts that I have at 4x these days. Older audio books and bad audio recordings I have to slow down. (AKA, when the podcast pulls in live audio from a press briefing or something). Also, when I'm not using my headphones it gets harder as well.
I regularly do all of this while doing so many other things ( driving, chores, biking, running )
I've taken it past 4x
Re: (Score:1)
Also, other things that are hard to listen to at 4x are when the presenter has an accent. or other noise is in the background. So for me listening to a british speaker I have to pull it back to 3x or sometime slower.
Also, audiobooks like the ones that these guys make: https://www.graphicaudio.net/ [graphicaudio.net] is terrible for me but more immersive for other people.
Always this way (Score:2)
Measuring brain injury (Score:2)
What about content? (Score:2)
Simple questions, does a stronger appreciation for the subject matter making it easier to comprehend?
If you already understand much of what's being said and you speed it up then surely your brain will able to ignore some parts you already know and get to the meat of the subject. Once you've "trained your brain" to understand speech at twice the speed, pick a subject you have only a passing interest in but have never really taken much interest try to learn it and then check the results of your comprehension
Attention span is (Score:1)
Yes, speed-reading improves basic comprehension that x==y and y==z but it takes a moment of slow to integrate that x==z. Those speed-reading courses omit the necessity of pausing at the end of a paragraph.
For a brief time there was speed-reading software, which put each word in the one spot, thus removing the need to move the eyeball. Even text-only files didn't fit into such software, well. Plus, the software never had the necessary Pause and Rewind functions.
After 20 minutes ...
We've known for some time that mental ale
Listening at 3x... after several years of training (Score:2)
I listen to average podcasts in my native tongue at 3x, and even 3.2x for those of lesser interest.
Using 2.5x speed for the hard ones (like scientific stuff).
But I have been speeding up my listening for years now.
I started at 1.2x, and when I felt comfortable, I stepped up by 0.1x increment, until I felt comfortable again.
Bumping from 2x to 3x like related would have been impossible to me at the time I were to 2x.
Re: (Score:1)
yes. That's similar to my experience. The author seems to have expected to get there after 20 minutes. I wonder what other things we give up on after a cursory effort and assume that it 'broke our brain' or is otherwise impossible.
SubjectIsSubject (Score:2)
In my day... (Score:1)
Sucky increments (Score:2)
Why are the only options 1.5x, then 2x, then 3x? Those jumps are way too big to adjust to. All the apps I use have increments in the tenths. Its jarring, because your app sucks, not because you broke your brain.
1.5x is about the max for me (Score:2)
Most technical people already speak much faster than "normals," so for technical podcasts I usually listen at 1.5x. It's very odd to hear the same person at 1x speed after a long time listening to the faster version. But anything over 1.5x results in me losing the conversation. I have a 30 minute commute each way and being able to squeeze as much learning in that time leaves me time to have a normal life at home. I hate having to use podcasts as a primary learning source though...employers won't pay for tra
Depends on the content ... (Score:2)
I can listen to 'lecture' type content at close to 3x (2.7x) comfortably for many speakers - largely because many lecturers speak at a ponderous rate and are extremely redundant.
However for entertainment (film and TV DVDs, BluRays) - I can comfortably watch at around 1.7x-2x depending on content - probably due to the need to process emotional cues, etc.
Re: (Score:1)