It's nice that this was also brought up. At least for me, even though I don't listen to audiobooks, it's too easy to become too "performance-oriented" when reading something. Meaning that ticking the box "achievement unlocked, read another book" becomes more important than what you really get out of the book. Did you now retain as much information as possible? Did you savor the work of art or just gulp it down? Ironically, if you don't use tools like these carefully, listening to books may end up being even a bigger waste of time.

This leads to: ffmpeg -i happy_book.mp3 -af loudnorm -af silenceremove=1:0:-50dB:detection=peak -af rubberband=tempo=4 happy_book_out.mp4

You could roll all those options into one filter command, but leaving like this for newer-users to digest. I also tend to re-encode using libfdk (though you need to be sure that your ffmpeg is compiled with this enabled) for a better mp4 (or .m4a or .m4b), so : ffmpeg -i happy_book.mp3 -af loudnorm -af silenceremove=1:0:-50dB:detection=peak -af rubberband=tempo=4 -c:a libfdk_aac happy_book_out.m4a

Sometimes I listen to one chapter in a book and I need a whole day to digest it: I get into a state where I don't want to listen to _any_ music, etc.

I am happy I’m past the quantity stage and into actually enjoying things part of my life.

After that I have been experimenting a bit with speed settings and what feels comfortable really depends on the narrator. It isn't just how fast or slow they narrate, but also the cadence and the material they are reading. Some material requires some time for processing and thinking while listening. I end up adjusting the speed for every book I listen to and it is rare that I go any faster than 1.4.

Fastbook was a bit uncomfortable to listen to. It totally kills the rhythm of how someone speaks. However it does a surprisingly good job of not making it sound too bad. It is unpleasant to listen to, but it isn't immediately obvious why. So I'd call that perhaps a partial win?

I've tried to gradually up the narration speed, but even if I can understand what people are saying at 2x, it doesn't leave much room for thinking about what they are saying before they are on to the next idea. There is no time for processing. I noticed that I'd pause books to get some time to think - which is clumsy and inefficient.

For video courses I often vary the speed between 1x and 2x speed. Some instructors have a terrible teaching style where their narration is peppered with irrelevant and annoying asides. For instance when they take time to explain something that ought to be obvious and which isn't directly related to the material. Then again for video courses, I make much more use of pausing, speed adjustment, I go back to play things again, stop to think etc.

I don't think this obsession with very high speed is useful. What is useful is to adapt the speed to where it feels comfortable and where you can absorb what is being said.

My wife thinks I'm insane anytime she hears the audio. It's interesting that there seems to be some training component to ramping up speed.

I would like the silence between words removed. But the silences betweens higher level structural blocks are necessary. You need a pause between chapters to signal the change of topic. Or depending on the books structure, sometimes between paragraphs too, when they indicate a switch of topic/perspective/thread.

One of my annoyances with some narrators (directors?) is that the gap between sentences and chapters is equal length, which often makes a confusing jump of context. I like it when each new chapter is announced as it comes.

My wife and I have both been family plan members of Audible for about 10 years and when you include their free audio book Plus Catalog, we have probably well over 1000 shared books.

I sometimes speed up listing by up to 1.5 times faster for books that I like well enough to finish, but not 100% liking. For some books I really like, I may end up listening to them two or more times - so it is not all about saving time. Yesterday, I was 3 hours into a 12 hour book and I was not loving it but wanted to enjoy the ending so I skipped forward to just the last two hours.

Even though I am mostly retired, I find there is not time enough for much of what I would like to do. My personal best strategy for getting time back is learning how to enjoy just the first few episodes of streaming video series, and then stopping unless I really love the series.

One last thing on the subject of using time effectively: I find that the recall from reading some books I really loved is far from perfect. My wife and I subscribe to the Blinkist service, and I find a really good use of time is to occasionally read (or listen) to a 15 minute summary of a previously read book and then spend another 5 minutes thinking about the book.

After a couple of years I realized :

- most modern books just don't worth the time spent reading them : a simple summary would do and often doesn't bring anything new anyway

- the rare books worth reading require time to be processed (for essays) or to get immersed (for novels).

So such apps solve a false dichotomy, in my opinion. But it can be useful as an "improved podcasts" maker. For the audiobooks I was making, I used this app which is great and highly customizable (speed, pauses time, export in multiple formats, cut into chunks, ...) : balabolka[1]

[1] http://www.cross-plus-a.com/balabolka.htm

   - transcribe

   - automunge that transcription

       - remove "ah", "um" etc

       - remove repeated phrases [0]

       - optionally summarize [1]

   - stitch the surviving segments back into continuous audio

   - match words-per-minute between the speakers [2]

   - do all the "smartspeed" stuff that Fastbook and Overcast.app do

[1] No shade, because I understand it's a chat and not scripted, but Accidental Tech Podcast could benefit from summarization. I love that they're talking stuff out. But sometimes I just want the conclusions.

[2] Like when Lex Fridman interviews someone who talks faster than him.

I'm not planning to stop but I have noticed that after a few years of doing this, my speech has gotten faster. This is not good because I feel sometimes I don't let people process what I say and sometimes I make them feel rushed. I suppose fast listening is subconsciously training me to speak faster

Another interesting observation is that I often don't know what the theme musics of my favorite YouTube channels sound like because they sound very different at a faster speed. When I accidentally hear them at normal speed I think to myself that they actually sound a lot nicer than what I'm used to.

So, either I could listen to audio, using this tool, but keep the speed 1x, OR use >1x speed but not use this tool (i.e. keep the gaps).

Cannot do both.

I can still listen at a reasonable 1.5x (or faster if you wish).

They have transcriptions you can use to read along or use to create notes.

For popular podcasts you can review a list of snippets people saved.

It's easy to share your snippets with friends, note & storage apps.

I would love to see a version of this for books and audio books. I do know that you can buy both the written & audio books & then use one of many review or summary websites to compare & see what other people have taken from the book. But it would nice to see this all in one app. For podcasts, Snipd has set the bar for the time being.

I feel like the pursue for speed/efficiency kills the time our brain needs to actually absorb and digest what it just listened. If after a week you don't remember anything, what was the point?

But hey, if it works for you it works.

However, I believe there is a way to do this with a single ffmpeg command. I remember watching lecture videos with higher speeds wondering about the huge amount of silence that made it still pretty annoying. I didn’t find anything simple to trim silence on audio and video combined and finally only listened to the audio with trimmed silence using ffmpeg.

Edit: It's literally just an audio filter in ffmpeg, see https://stackoverflow.com/a/29411973

Current stats: * Listened for 27 days 9 hours, * Manual Skipping: Saved 17 hours, * Increased speed: Saved 31 days 21 hours, * Auto trim silence: Saved 6 days 14 hours

Additionally, I struggle with hosts who have a habit of ending every sentence with an upward inflection, which I find quite annoying.

Some texts, and usual human speech, are low-density, high-redundancy. If you can easily understand the ideas they carry, you can listen at a really high speed and still don't miss anything, and later be able to adequately recall the ideas.

Some texts are either higher-density, or talk about concepts novel enough for the listener, and it takes some time to unpack the ideas and commit them to memory. Such texts cannot be read too fast, and sometimes require going back a sentence, or a paragraph, and re-reading.

An ideal (from the efficiency standpoint) audio book device would have a speed dial to control the level of adaptive compression, like described by the linked page. It would also have controls to jump back a sentence and replay it slower.

But such device is unlikely, because I suppose that the main audience of audio books is commuting drivers, who need to keep their hands on the wheel, and their eyes on the road, so they can't properly read.

And if course I suppose that the speedup and compression are barely applicable, or need a different approach, to books read with some artistic expression (that us, most fiction worth reading).

You're expressing a common skepticism of 2x+ speedup but it misunderstands why people do it. It actually improves the presentation to 200+ wpm so the brain is more receptive to learning. The slower ~100 wpm (i.e. 1x speed) acted as a barrier to learning. I tried to explain the 2x+ advantage previously and how blind people had utilized this technique years before the general public did:

https://news.ycombinator.com/item?id=29661550

https://news.ycombinator.com/item?id=29660750

https://news.ycombinator.com/item?id=29665690

The motivations for speedup mostly applies to non-fiction audiobooks. For fiction, you may deliberately leave it at 1x because you want the deliberate pacing of the voice acting in addition to the info conveyed by the bare text.

Audio playback speed adjustment is just a tool. If you're a musician trying to learn a complicated guitar lick or drum fill, slow the music down to 0.5x or even 0.25x if that helps unlock it. If you're listening to Lord of the Rings, then playing it at 1x seems very reasonable. If you're listening to a speaker discussing inflation for an hour but keep tuning out because he talks too slow, speed him up to 4x if you have to so the information is presented at the higher speed your that brain prefers receiving it.

EDIT to replies to clarify "The slower ~100 wpm (i.e. 1x speed) acted as a barrier to learning."

This statement in isolation looks like an absolute science claim that 1x causes "zero learning" but of course that wasn't what I was saying. The context is "barrier to _optimal_ learning" for that particular reader. If 1x is too slow and makes some readers not stay mentally engaged, or abandon audiobooks, or leave podcasts/lectures in the queue and never listen to them, that's the "barrier to learning" that I'm trying to convey.

Likewise, if one needs to slow it down to 1/2 speed to comprehend difficult-to-parse text, that also meant 1x was too fast and also a barrier to optimal learning.