I'm watching almost anything between 1.21 and 1.77 (these are the default speed-up steps in mpv - 1.1^x), depending on speaker. I just cannot keep attention to most things when played at 1x. When I miss something, I seek back a few seconds (mpv seeking is not limited to keyframes) and maybe slow down. If it's an important lecture, I watch the most critical parts later once more (maybe a week or a month later). This works better for me than watching at the original speed once. As a bonus, you get video that does not tear your eyes, as most things are still produced at only 25 or 30 FPS nowadays.
I cannot keep at 2x with almost any speakers though, neither in English nor in my native language.
What really matters is the audio resampling algorithm. I consider the default in browsers and YouTube really bad (apart from that, YT offers only 1.5 and 2x, which is too coarse). I actually prefer no "smart" resampling - just play the samples faster (and I guess low-pass them to avoid aliasing?) -- audio-pitch-correction=no in mpv is what I use. I think some adaptive algorithm that compresses spaces between syllables more than the "payload" might be great, but I have not seen any yet (suggestions?). I also have a tiny bit of hearing problems, and maybe this shifts the frequencies to a region where my "brain receiver" works better?
I also downmix all audio to mono (af=lavfi=[pan=1c|c0=0.5c0+0.5c1]), as I have problems (kind of nausea, or trying to follow the sound with my head) when listening to stereo with headphones. (I have tried to use remap-sink PulseAudio module, but it added terrible latency)
Because of this, watching anything in the browser is a terrible pain. I really wish browsers allowed you to specify an external player and pipe the stream to it. (I use youtube-dl, but of course it does not work for everything)
I cannot keep at 2x with almost any speakers though, neither in English nor in my native language.
What really matters is the audio resampling algorithm. I consider the default in browsers and YouTube really bad (apart from that, YT offers only 1.5 and 2x, which is too coarse). I actually prefer no "smart" resampling - just play the samples faster (and I guess low-pass them to avoid aliasing?) -- audio-pitch-correction=no in mpv is what I use. I think some adaptive algorithm that compresses spaces between syllables more than the "payload" might be great, but I have not seen any yet (suggestions?). I also have a tiny bit of hearing problems, and maybe this shifts the frequencies to a region where my "brain receiver" works better?
I also downmix all audio to mono (af=lavfi=[pan=1c|c0=0.5c0+0.5c1]), as I have problems (kind of nausea, or trying to follow the sound with my head) when listening to stereo with headphones. (I have tried to use remap-sink PulseAudio module, but it added terrible latency)
Because of this, watching anything in the browser is a terrible pain. I really wish browsers allowed you to specify an external player and pipe the stream to it. (I use youtube-dl, but of course it does not work for everything)