I'm in grad school and watch all my lectures using the approach described in the blog post. After following this accelerated viewing approach for almost 2 years I can tell you that in reality the content doesn't get consumed "much more quickly".
First, for most lecturers it isn't possible to listen to recordings more than 1.5 times faster. At this rate, there is a roughly 33% reduction in the time it takes to view a lecture. It is noticeably faster, but not really significant unless you are viewing more than 1 lecture at a time, for example a lecture is typically 50 minutes, so one saves about 17 minutes per lecture. In my experience the savings fail to realize for the reasons I'll cover further.
Second, the rate at which the lecture can be accelerated varies significantly depending on the professor who is presenting the material. Some professors speak quickly and once the playback speed is increased to 1.5, their speech becomes blurred and unintelligible. Often the native English speakers are the ones who can't be sped up much. Luckily, at my school most of the professors aren't native speakers so a rate of 1.5 is adequate. For non-native speakers, the useful playback rate varies pretty widely. One of my professors who is from Turkey speaks slowly and his lectures can be played and understood at a speed up of 2.
Third, the playback speed does affect comprehension. Even at a normal playback rate I find that some topics covered in the lecture require a rewind and a replay. Accelerated playback requires closer attention and more frequent replays. Of course how often one needs to replay the lecture depends on the content and varies from person to person.
"Often the native English speakers are the ones who can't be sped up much. Luckily, at my school most of the professors aren't native speakers so a rate of 1.5 is adequate."
Luckily?... Think about it a few seconds. You don't get any more benefit from listening to a slower speaker at a faster playback speed over a faster speaker at an equivalently less fast playback speed. If you see what I mean...
Strictly speaking, not true. If the lectures were all standardized at, say, 50 minutes each, then a professor that could be sped up more would take less time to get through each lecture than one who could not be sped up much at all. It's kind of like how you can read a page with lots of blank space much faster.
Of course, it all gets muddled in the question of how much info per lecture\minute is actually best for learning, but if your goal is to get through them quickly then a speedable professor is best.
>You don't get any more benefit from listening to a slower speaker at a faster playback speed over a faster speaker at an equivalently less fast playback speed
That assumes that you are comparing the same course being taught by a faster speaker vs. a slower speaker. In reality, you don't have a choice. So if you are stuck with a slower speaker, you might as well speed up their lecture.
I don't think you understood what I mean. I mean that if you have someone who speaks at a rate of x and that gets unintelligible at a playback speed of y and another that speaks at a rate of 2x and gets unintelligible at a playback speed of y/2 then all other things being equal you have no advantage in listening to one instead of the other if you playback at the maximum intelligible speed.
Right. But a lot of material is just too boring to pay attention to at all. Especially for new or occasional speakers (for example, lots of tech talks.) So this is a nice thing to have.
i still want to find a way to do this with ffmpeg (for the video) and sox (for the audio) plus maybe slide changeover detection, slide text recognition, face detection (to remove motion from speaker) etc etc etc. Feels like fixed camera video is not too hard to process...
First, for most lecturers it isn't possible to listen to recordings more than 1.5 times faster. At this rate, there is a roughly 33% reduction in the time it takes to view a lecture. It is noticeably faster, but not really significant unless you are viewing more than 1 lecture at a time, for example a lecture is typically 50 minutes, so one saves about 17 minutes per lecture. In my experience the savings fail to realize for the reasons I'll cover further.
Second, the rate at which the lecture can be accelerated varies significantly depending on the professor who is presenting the material. Some professors speak quickly and once the playback speed is increased to 1.5, their speech becomes blurred and unintelligible. Often the native English speakers are the ones who can't be sped up much. Luckily, at my school most of the professors aren't native speakers so a rate of 1.5 is adequate. For non-native speakers, the useful playback rate varies pretty widely. One of my professors who is from Turkey speaks slowly and his lectures can be played and understood at a speed up of 2.
Third, the playback speed does affect comprehension. Even at a normal playback rate I find that some topics covered in the lecture require a rewind and a replay. Accelerated playback requires closer attention and more frequent replays. Of course how often one needs to replay the lecture depends on the content and varies from person to person.