People Who Speed-Listen to Podcasts
buzzfeed.com
buzzfeed.com
Uh... that's not how modern speedup works. Here's a brief rundown:
1. break the audio up into very small windows 2. FFT audio in a window from time domain to frequency domain 3. FFT the audio back from frequency to time domain, with an even smaller window (or larger, if you want to slow playback down).
The frequencies resulting are the same, but shorter in duration. Obviously there's some sampling bias that can creep in, but these algorithms are generally pitch-stable.
#include <math.h>
typedef struct {
double r;
double i;
} complex;
void fourier(complex * x, complex * y, int np)
{
int n, k;
for (k = 0; k < np; ++k)
{
y[k].r = y[k].i = 0;
for (n = 0; n < np; ++n)
{
double c = cos(2*M_PI*k*n/np);
double s = sin(2*M_PI*k*n/np);
y[k].r += x[n].r * c + x[n].i * s;
y[k].i += -x[n].r * s + x[n].i * c;
}
}
}
On my iMac (3.4 GHz Intel Core i5), compiled with "cc -Ofast", here are timings:
size time/FT FTs/sec
---- --------- -------
128 0.24 ms 4167
256 0.93 ms 1074
512 3.8 ms 263
1024 15.5 ms 65
2048 65.6 ms 15
4096 243 ms 4
[1] https://en.wikipedia.org/w/index.php?title=Discrete_Fourier_...
it really feels good to pump interesting thoughts in at the speed limit of perception and recognition :)
No kidding. Overcast has a "Smartspeed" feature that cuts out these pauses. The counter on my app says it's saved me 195 hours to date.
"Smart Speed has saved you an extre __ hours beyond speed adjustments alone."
(Just went through this exercise to find it last week)
Also, holy shit, this has saved me 255 hours, not counting speedups. That's crazy.
And with that experience, what are your favorite podcasts?
I may get downvoted as just an old man yelling at clouds, but I wish that people would actually take more time to enjoy silence in its own right.
My own kids seem to like doing everything with a pair of headphones on, or a video playing on their devices constantly, but some of my own moments of pure bliss have been sitting in my garden, sans technology, with a book or just petting my cat and stopping to breathe, listen to the sounds of nature and think about ideas.
I am in my 20s now but I do agree with you. I enjoy just being with silence around me, especially when reading.
As a consequence, I really appreciate the quiet moments and nature walks etc.
It's pretty hard to enjoy silence in two second chunks.
Someone might use smartspeed to save half an hour, and then use that time to enjoy nature uninterrupted.
I'm not going to downvote you, but you basically are yelling at clouds. The two kinds of 'silence' are completely unrelated.
Even two seconds of silent time is useful to ponder and reflect. When I am not in 'flow' state while working, I will often just look away from the screen to pat my cat or pick up my guitar for a few seconds to noodle unplugged. I find that helps me to step down from frenetic brain activity and let my subconscious stew on things.
To equate it to a musical analogy - sometimes it is the space between the notes. As much as I respect the technical ability of guitarists that can cram 3000 notes per minute in a long shred solo, I cannot listen to that for more than 30 seconds before my brain is screaming for some relief. I would much rather listen to a blues guitarist or saxophonist play a few notes with enough pauses between phrases to fully appreciate them and wait for the next one with anticipation.
Some silence just is not enriching. I would rather spend an hour in a forest than waiting at the DMV. I would rather enjoy a cup of coffee than wait 10 minutes for a colleague to arrive on time to a meeting.
There are some things that are worth experiencing in their fullest, and there are some things that are only worth experiencing by skimming through.
Like some books are worth deep study, reading, rereading, analyzing, digging, and getting every ounce of value from them. Others give all they have in a rush.
You think you saved 195 hours, but in reality you saved far less hours (I'm not saying you didn't save any hours).
Silences serve to aid your brain from getting tired, allowing you to catch up and process the information, or to get in diffused mode of thinking.
If silences were completely useless they wouldn't ever be used. You'd download an audiobook from the official vendor, and it wouldn't contain any silences.
I find that when I’m watching/listening something as long as there are no pauses, I don’t need the 2x speed. Particularly because I find speech often difficult to follow at anything more than 1.3x while at 1.3x the pauses are still way too long.
Smartspeed feature has been a godsend.
This is also a something that NPR has been doing on shows for years - perhaps not 'all' but they certainly do a lot of cleanup.
They also have some training material online, which may be of use to many people doing editing: http://training.npr.org/audio/the-ear-training-guide-for-aud...
The transcript of Pulling Back the Curtain may also be of interest, I think it may have been rebroadcast at some point as well so audio may get turn up.
Edit: http://www.wnyc.org/story/129697-pulling-back-the-curtain/
I certainly don't notice it easily but I have saved several hundred hours according to it.
Listen to folks using screen readers reading off menus and navigation and such. Those users adapt to stuff going much much faster than 2x!
[0] https://webaim.org/techniques/screenreader/ [1] https://www.vincit.fi/en/blog/software-development-450-words...
But then I would shift back to 4,000 wpm and suddenly it seemed so easy to keep pace. And the longer you stayed at 10,000 wpm the easier 4,000 became.
I feel our brains are totally up the challenge.
I'm not sure I'd retain much from a technical conversation at 3x -- for me the drop seems to be somewhere between 2x and 3x.
It's one thing to listen to a very consistent voice repeating very much familiar material. Another to have novel or complex material thrown at you.
My listening speeds range from 1x to 2x, for the most part. If I'm focused on material, 1.5x is almost always listenable. If I'm highly focused, or the speaker is slow and boring, 2x+. If the material is difficult or the speaker exceedingly good, intelligent, and the material well-prepared, 1x - 1.25x works quite well.
At some point you'll start to hear what birds and other fast talking animals are saying. I'm not joking. I mean you'll still probably wouldn't understand them.
Most lectures at 1x border on unwatchable except for the very best presenters (and then you wonder how much you're learning as opposed to being WOWed.
But I mostly listen to podcasts for entertainment reasons (i.e. people I wanna listen to about topics that interest me) so my desire to speed through is not so high.
For lectures, I tend to speedwatch with stops, replays and skips. In some cases, I am 100% convinced that like you, I'd have never been able to stay tuned. The whole thing has a tendency towards escalation, and the result comes with its own troubles.
In other words, allow creative thoughts about the things the lecturer is saying to form – wander around in idea-space a bit – and think about them, while simultaneously keeping up with the thread of the speech. You'll gain a lot more from your time in the lecture hall, and it'll also be a lot more fun. A lecture shouldn't just be (and doesn't have to be) dry information stuffing.
With video tutorials I have the same problem. On normal speed I will work ahead, take a wrong turn because I didn't know where the lecture would go or drift off in to something slightly creative and than notice that I missed a section where finally something new was explained. So I have to skip back and find where I started missing something (Something really really annoying with audio material; No thumbnails, no search, no skimming backwards, mostly no identification possible from a single frame or word (the easiest way to find a part again, to which I listened before, is to remember where on my journey I was when I listened to that part and where that was in relation to where I heard what is currently playing)).
Books are an intrinsically better format for this since the amount of time required to 'process' the new information will vary a lot from person to person, and you get complete control with books. I can't see any benefit to the lecture format aside from mass consumption—aside from question-asking, which I do think is vital, but it is typically under-used. Books + lots of questions to an expert seems ideal.
[1] http://nautil.us/issue/52/the-hive/modern-media-is-a-dos-att...
Another analogy might be trying to ride a bike too slowly. It's very difficult not to wobble extremely and fall off. It takes more effort to ride slowly. My mind can't wander with slow speech because it takes great effort and attention to maintain ideas in active memory long enough to make it through sentences. I am better able to have branching thoughts while keeping up with the thread of speech when the speech is faster.
I frequently drive with audiobooks on and it is usually not possible for me to follow along without a speed boost of at least about 25% or so.
Reviewing data structures? 2x
Ad retargeting in marketing course? 1.5x what wait 1x
And then its back to 2x, I concentrate at listening better
I too went through a phase of getting addicted to podcasts and subscribing to almost every one that people recommended to me. But then I realised that I simply could not listen to them all within my allotted 'listening time' - which is usually during a 1hr walk in the mornings.
So now I simply don't bother trying to keep up with listening to consecutive episodes, no matter how much I enjoy the podcast. I just go to my list of waiting episodes as I walk out my front gate, and hit 'play'. If it finishes during my walk, I start another random one from the list. If I return home before it finishes, I hit 'stop' and rarely go back to it unless it was particularly informative.
I just treat my podcast list as a virtual sock drawer that I can reach in and grab anything out of without looking. I no longer live with the FOMO anxiety that I might miss a brilliant episode somewhere. I have too many other important things to do in my day.
I should probably try bumping it up another .1x myself, now that I think about this.
I don't have a problem with listening quickly, it's the very long list of podcasts that suggests misguided focus.
When I was involved in online gifted parenting groups, it was common to see smart kids "obsessed" with something that worked well for them. In my family, the dysfunctional behavior was resolved by helping my son find ways to genuinely get his needs met so he wasn't intellectually starving anymore. I also helped him find outlets foir his interests.
The example I always gave him was that there was nothing wrong with liking debate, but go join a debate club instead if picking arguments with your mom every day. He never joined a debate club, but he did learn to find socially acceptable outlets and stopped making me apoplectic.
I will add that gifted kids who were "obsessed" with a particular thing were often twice exceptional. They were gifted and also had some kind of handicap. Their obsession was something that adequately met their intellectual needs after years of starving because it somehow was both sufficiently information dense and delivered the information in a way that worked well for them. Someone who speed listens to podcasts constantly sounds to me like someone seeking a density of information in a format that works for them.
So yeah, I get it. But it seems to encourage the wrong sort of content - long, windy, with no effort made into presentation or editing.
Compare a16z with Masters of Scale with Reid Hoffman (gosh even that name sucks). a16z has no ads, no sound effects, no blabber, but absolutely excellent content. Masters of Scale tho move slowly, keep repeating same things and feels like it's targeting children, way more edited.
Edit: maybe it's just the format that makes it interesting. Google Cloud Podcast, Good Day Sir, Software Engineering Daily - all have conversational format, almost a debate.
An episode of 99% Invisible though... I want to slow down and enjoy like a fine wine.
p.s. I do like 99% Invisible, my most recent find.
Maybe I'm just not as good in processing such social interactions at a faster pace than real-time. But I love "sharing a laugh" as if it's happening right now. That feeling completely goes away when the audio is sped up.
There's no way that with speed listening or reading everything sticks equally as listening or reading a 1.0x speed multiplier. It may feel like you got more done, but in reality you did not. You're essentially fooling yourself, feeling like you accomplished more than you did.
I do the same, but with movies/series. I "watch" them on a second screen without being all the time focused to it. Except if its a really good movie/serie. But if it isn't a good one, why watch it in the first place?
My own intuition tells me that everyone probably has a maximum (and maybe minimum) speed that they're comfortable with, and there's no reason to assume that a podcast host's preferred speaking rate is the same as an audience member's preferred listening rate.
Speeding up your primary focus is not at all comparable to consuming something as a secondary focus. I agree that "watching" something in the background is terrible for comprehension and retention.
I'm not sure if we're discussing the same thing here. I am talking about learning.
Yes, there's also a lot of people who believe speed reading allows them to read quicker. Sure, it might seem that way, but does the information stick? Do you think anecdotal evidence to test if information was learned counts as evidence?
Dr. Barbara Oakley and Dr. Terrence Sejnowski didn't mention your proposed learning method in their Learning How to Learn course [1]. They also debunked speed reading in that course. If your proposed learning method works well, they should incorporate it in their course. However, it seems to go against the principles of chunking and overlearning.
Of course it all depends on the host. If you find a podcast where the host is already reading at 250 wpm, you'll probably have a hard time catching everything at 2x speed.
I've been thinking a bit more about it and I don't remember audio-speedreading being mentioned in the Learning How To Learn course (which I followed, and I hold both teachers in high regard). I think I am biased from the text-speedreading shenanigans.
> Of course it all depends on the host.
Point taken.
My theory is it also depends on the reader. As I grow older, I find it is more difficult to follow things. Sure, my vocabulary still slightly increases but I also think things through more, am more experienced, and eventually I'll end up with Alzheimer's. Hence my suggestion it is catered to the elder. Certainly not for HFAs. (I used to talk very quick as well as child. It was annoying as fuck to everyone, not in the least my parents.)
I think reading and listening are not as different as they seem.
I listen to a lot of podcasts, and if I had a 500-hour backlog in one, I would conclude that I care about other things more, and either unsubscribe and delete it, or stop downloading episodes by default. I used to auto-download Fresh Air, for example, but when it started backing up, I dialed things back, and now listen to 25-50% of episodes. I also used to listen to The Talk Show, but decided there were other things I would rather do.
Personally, I think we're over-indexing on knowledge acquisition while underemphasizing taking action.
Other commenters' points that it is also a sort of focus enhancer also ring true to me--I have arrived at the office with nearly 100% retention of what a specific recording was about, especially since I tend to listen to technical podcasts like TWIML[1] (which focuses on stuff I do at work), and I get nothing but value from being able to listen to two podcasts during my one-hour commutes.
Softer stuff like Merlin Mann, Gruber, etc. I will either navigate with the chapter guide or making liberal use of the skip buttons. I seldom (if ever) listen to any kind of podcast at normal speed, except for some language courses.
Even then, I tend to slow down only for languages like Mandarin or Hebrew (which I'm less acquainted with). Spanish, French, etc. (which I know from childhood) tend to be drawn out enough that I can even speed up a little :)
[1]: https://twimlai.com/
Having hotkeys ">" and "<" also help control the speed when needed.
javascript:document.getElementsByTagName(%22video%22)%5B0%5D.playbackRate=2.5
Interestingly some browser start to clip audio at certain speeds. Also, heavy accents become difficult to listen at higher speeds.
p.s. comma and dot (got me confused with ><) on youtube skips video frame by frame.
It is clear that the playback software shifts the pitch downward so it matches the 1x pitch, but there's still some "artifacts" that show up when you go beyond 1.5x.
Is there a precise word for this kind of distortion? besides "choppy"?
Also, I wonder how far you could get in time compression by merely removing spaces between words without actually compressing the words?
That is interesting, but a good speaker uses spacing to better convey their message. In order for it to work, it would have to somehow take into account rhythm. Perhaps just cutting longer spaces instead of spaces between words. Maybe even starting with the creator, or users tagging unnecessary bits such as coughing, or reaching for a drink.
And here's the thing I realised earlier - most interesting podcasts have usually two speakers, discussing something. Not much room for rehearsal and/or retakes I guess.
Some audio time-stretching libraries do exactly that, reducing the amount of samples in silent periods compared to samples of non-silent audio.
It's a really difficult problem and actually all the on-the-fly things don't do it really well (at some speed vocals blur in to consonants and consonants get clipped. good software that has no simple FIFO algo can do it much better (conserving consonants etc.).
This is Overcast's USP.
Anyhow, I wanted to share an anecdote, I am listening to 'Note to Self' and host is sharing her feelings and concerns and fears... And I am thinking, this lady is so nervous and jumpy and her friends too. So next time when I was opening an app, I am thinking, do I really want to listen to this nervous lady telling me about how to handle social media... then it dawns me that she just sounds nervous because I've been listening on 1.4x and everyone will sound nervous if they talk about their feeling on speed.
So there...
Unless I'm doing something mostly physical or mindless I can't listen to a podcast and work at all.
a more aggressive version of this is actually skipping 15-20 minute chunks of audio at a time and trying to pick up contextual clues about the preceding chunks to verify if they are worth listening. the greedy approach often pays off here if you are just looking for one or two specific portions in a giant pile of noise. you might end up doing 10 skips and listening to 8-15 minutes of 90 minutes of content.
information addiction leads to a sensation of impending doom. the knowledge you would like to access and understand is spreading like a dark cloud and your global-time is counting down.
my solution was to read the ~15 core books and retain their surface sensation. the rest is now processing and categorizing "new" information, pattern matching categories and bias detection to remove false positives.
The real key to all this though is when someone finally creates this for online articles or books - sure, I could download a version, format change in calibre and then have it read to me via Voice Dream, but I need a better voice than the technical voices nowadays. I would definitely pay for this.
Lastly, this also occurs in other media for me - pretty much any movie before 1990 can be watched at 1.5X on VLC. Similarly, every YouTube clip is set to 2X usually as well. It may sound exhausting, but you just get used to it after a while.
I wonder how many people select tech videos based on length? Surely YouTube / Vimeo etc. has studied this? I wish there was a search option that let me filter content based on length.
Give me a well scripted well produced drama, comedy stand-up routine or documentary and I want to listen to it at 1x.
Everything else is information sparse ad-libbing off of semi-improvised material re-hashed from books and articles then strung together with conversational back scratching. Or the one dimensional drone of someone reading from a book/script.
Is slow-core a thing yet? Audially ram-packed soundscapes lovingly crafted over months of careful layering of complementary edits that rewards the patient and attentive listener with hours of meat-speed re-listening. Sounds a bit hipster tho.
But stuff from Twit.tv is simply unbearable to me.
Yet, I cannot listen to podcasts on anything more than 1x speed. Must be a processing issue somewhere in my brain that causes this disconnect...
This is basically a long winded way of saying that I think audio book makers expect the average person to bump the speed up a little.
[I listen to audio books on Audible at around 2x speed.]
The voice actor, Steven Pacey, is so amazing I've considered listening to other books he's narrated solely due to his performance. I haven't quite gone that far yet, though :) However, I do recommend this book/series -- if you like fantasy.
I make sure to keep the list to a minimum though. Sounds like these guys are listening to way too much.
Is there something open source that works great?
Unfortunately I can only stand one voice - "Alex" from OSX, at about 1.3x ... 1.5x speed. Other voices I find irritating. I'm following the new WaveNet TTS voices with great expectations, but they don't seem to come near my laptop.
WaveNet: https://deepmind.com/blog/wavenet-generative-model-raw-audio...
People who focus enough will read a book at the right speed their brain can best process the information. But that speed is not the same for everyone, some people will go through a book twice as fast as others while extracting the same amont of learnings, and we won’t tell them they read at the wrong speed.
If it’s true for books, why would it be different for audio contents ?
There’s really no special rule that makes 1x playback a golden speed, especialy considering some podcaster speak faster in person than when they record for instance. Or the otherway round, podcats like ATP have the silences auto-trimmed to some extent making it denser than when they record.
To each their speed.
no distractions, no ads, no external links == no million tabs open, ->> you cannot easily scroll and tldr stuff <<-
I can enjoyably listen at 2x for some people, but others I cannot go beyond 1.5x and I've been listening to audio at 1.5x or 2x for a long time.
It's much less about their pace too. It's overall line quality and how clear they enunciate words.
For my own recordings I try hard to make them feel good at 2x (even tho I naturally speak fast and manually edit out pauses). About 10% of the people who take my developer courses listen at 1.5x or 2x and have enjoyed the clear audio experience.
(maybe I could learn to do it with podcasts playing more quickly but I don't think I listen to enough podcasts to do that)
Any good android and windows player recommendations? I can recommend:
windows: windows media player (surprisingly good, once you found the playback speed menu)
android: antennapod (sadly you can only cycle through speeds one way, which is really annoying). But it has the benefit of supporting chapters, subscriptions, auto-downloading and removing files etc.
I just never even realized this was a thing, and that that 1.0X next to the playback buttons actually did something...
1. Speed-listening certainly increases the number of words I hear, but does it actually increase the amount of ideas I absorb?
2. To what degree does speed-listening impact my ability to engage fully in normal-speed human conversations?
(Also, do some podcast producers speed up their recordings and edit out silences before publishing? For example, I suspect a16z does this with their podcasts.)
Often when I'm listening back, those pauses are just too long, especially when you're the only one talking.
In regards to speed-listening, I just don't think it's a good idea. If you're rushing through anything (or at least feeling rushed), more often than not, you're not truly enjoying it.
I don't think I've run into anything where I have been distracted by 1.1x, besides theme songs to shows I am familiar with which obviously don't sound like what you expect. It's basically unnoticeable.
I hadn't heard of the silence skipping feature but just checked and my player of choice (Podcast Addict) has it. Just turned it on so I'll see how that goes.
When I first started listening to podcasts in general and played around with the multiplier even 2x sounded like completely incomprehensible gibberish to me. The strangeness of super speed becomes apparent again when it starts autoplaying when someone else can hear. Their confused look is kinda funny though haha.
It has never, ever happened the other way.
It takes a bit of effort but isn't so hard to get used to. Love it.
On the topic of 2.3x video watching, I've observed that "question periods" after lectures are very difficult to watch at 2x speed. It seems the pace of "lecture speech" is deliberately slower for most prepared speeches, but when it comes to live question-answer periods rates jumps up. Has anyone else noticed this?
Sadly many things are video/audio nowadays
Interesting job that provides enough down time to listen to podcasts at faster than normal speed 8 hours a day.
I'm sure there are other jobs that are like that.
> Take 34-year-old Jason Strickland, who works for a land surveyor in Winston-Salem, North Carolina. He listens to around eight hours of podcasts at work every day ...
You can think of it like compression in computer science; not every file is equally compressable.
Even if it's a purely educational podcast or audiobook, any form of pacing or diction is lost at 2x, which, ironically, makes the story boring as well imo.
Sawbones is an edu-comedy podcast about medical history, and I tried listening to it faster and just didn't enjoy it as much. I really enjoy the comedic timing and interplay between the hosts, and it's already information rich enough that I feel fine listening to it at normal speed is fine.
One of the things I really like about WtNV is the almost menacing, deliberate pacing of the narrator. His speech is so perfectly even, regardless of the content, that it's almost hypnotic. I tried that at 1.25x and it felt off, but again it's a personal preference thing.
It's not about enjoying life at normal speed, it's about cutting down on the BS.
That's quite a generalization! I have a number of 90 minute podcasts in my feed right now.
I would love to find data on listening comprehension based on audio speed and activity, whether it be lying down, driving, doing chores, walking, or exercising. My intuition is that there are many people that best comprehend the material at around 1.5x-2.0x and comprehension drops precipitously at both ends. My guess is that, as with many mainstream broadcast choices, the canonical broadcaster voice is purposefully slow to cater the widest possible audience, including ESL listeners.
There's so many examples. Bryan Cantrill also speaks quickly.