Post by j***@astraweb.comPost by J. P. GilliverI've just
noticed you said "to youtube-dl": do you mean yt-dlp _as_ the successor?
yes.
Right. (I have it shortened to just y anyway, to save typing.)
Post by j***@astraweb.comPost by J. P. GilliverI read somewhere that it defaults to the best audio and video available.
I've recently discovered that's not the case - at least, assuming higher
resolution video is "better". Maybe it isn't, if it's just being
upscaled at YouTube! (The 720 - or less if the source material is less -
it serves me is fine for me: I don't have any display with more than 768
pixels vertically, so no point in having files containing more.)
Post by j***@astraweb.comI was not aware that was the case for audio. if you do a ' -F' arg,
you get available files and one
video will be tagged as best and it will be chosen -- that it would
choose the "best" video I was aware
I must do a few -Fs and have a look at which is so tagged. I'd been
_assuming_ it was one of the highest resolutions offered, but maybe not.
(Does it always _have_ that tag?)
Seems odd _if_ YouTube are offering upscaling, when surely that could be
done at the user; you'd think it would just mean they'd have to serve
more data, so bigger load on the servers. I suppose they can perhaps
have a _better_ upscaling algorithm/process than the user might, but
still ...
Reminiscent of flatbed scanners offering huge interpolated resolutions -
_in the scanner_, rather than in the driver software which surely would
be a better place for it: putting it in the scanner only results in
slower scanning (if limited by the interface), _and_ it can't be
upgraded.
Post by j***@astraweb.comof. If similar is done for audio, i never noticed.
(after checking a couple and finding no difference i continued to use
the 'youtube-dl /?' *lengthy* help
file which I redirected to file some years ago.)
Me too! (Both of us are probably thus using not the latest help, but I
don't think it changes much.)
I've never noticed whether there's a "best" tag for audio formats. I
will admit, if I just want audio, I still download the video and then
"extract original audio" using something else (Pazera, but there are
plenty such extractors). I can usually have that done in fewer seconds
than a -F/-f would take. (Most audio tracks I want are 2-5 minutes
long.)
Post by j***@astraweb.comFor an audio file to be used as background music by VLC i use a .bat that incorporates
"-x --audio-format mp3 --audio-quality 1".
I am not a fan of background music (no criticism implied); I didn't even
know VLC _had_ that ability. What sort of thing are you watching - do
you watch a lot of silent films, or your own (silent) cine film scans?
(I've just started converting mine, but the results need a lot of work
[which I may never get round to] before I'd think of adding audio to
them.)
Post by j***@astraweb.comI had always used 128kbps as "CD quality" but i notice i can tell a
difference in that and, say,
320kbps"
jack
Interesting. I recently was rather shocked to discover that my hearing
now rolls off around 8 kHz - though apparently that's normal for people
in their 60s, though on the low side slightly in my case. (I'll be 64
next month.) I'm not aware of the degradation - in particular, I have no
difficulty understanding people. (I never attended loud discos or had
noisy employment, so I don't know the cause of the decline - just
ageing, I guess.) However, even when I had more, I don't think I could
ever tell (for mp3, anyway: I could for mp2, as used on DAB) any
advantage from going above 128, and for a lot of material, 96. When I
examine an audio file, I do look at it on a spectrogram (I use GoldWave,
since I bought it just before Audacity came along; I'd probably use the
latter if starting out now), and - even though I can't hear it - I don't
reduce the bitrate such that any of the content is lost. (I know mp2
bitrate doesn't map linearly to frequency response; I've built up a
table of what corresponds to what, from experience, erring on the
cautious side. [I can share it if anyone's interested.]) I _do_ lop off
the top if there's no (relevant - I'm not talking about surface noise or
tape hiss) content up there, though, which is surprisingly common (apart
from what I assume is video timebase breakthrough, which is there on
90-95% of all clips!). [I even half the _sample_ rate if there's nothing
above 10 kHz.] The biggest reduction, though, is recoding as mono for
mono material: the vast majority (NOT all) of what's on YouTube, at
least when downloaded with yt-dlp default, is encoded as 44100 stereo,
regardless of whether it's mono or not: that recoding alone more or less
halves the filesize. (I think mp3 encoding is supposed to notice if the
two channels are similar/identical and take advantage of that, but it
clearly doesn't.)
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)***@T+H+Sh0!:`)DNAf
They are public servants, so we will threat them rather as Flashman treats
servants. - Stephen Fry on some people's attitudo to the BBC, in Radio Times,
3-9 July 2010