I’ve been using Clipgrab (macOS) to download music (mp3s) from YouTube. I’ve seen some mentions that the quality (real bitrate as opposed to what the file states) is bad to begin with on YouTube and that downloaders might make it worse. How bad is the quality of what I’m downloading? Any better downloaders or does the quality suck to begin with?
YouTube audio quality isn’t amazing, but it’s not that bad if you get the right format. Using yt-dlp, you can get opus audio that sounds way better than mp3. Example: yt-dlp -f 251 https://www.youtube.com/watch?v=dQw4w9WgXcQ
Separate audio is rarely available on streaming sites like YouTube, so if you’re ripping a song from it, you essentially are transcoding the files three times:
Encoding by the content creator
Extracting the audio from the the content
Encoding in the format of choice
So no matter how great the quality is at either of these transcoding ends, you are still transcoding in lossy format at least two times
I haven’t seen a YouTube video without a separate audio track. If you use yt-dlp just add -F and the URL to get all available download options/formats. Then use -f and the number of the download option and URL to download the video/audio. Audio on YouTube is available in two formats most of the time: opus and m4a.
yt-dlp can just download the audio. It usually comes down in m4a at quality that I would describe as “very listenable”. So only the first of those three steps are mandatory if you do it that way.
From my use of this, it downloads a video and extracts the audio stream from it. In fact, you can see this as it leaves the video file on disk while it’s extracting the audio, and the audio-extraction process uses a decent amount of cpu (takes a few minutes for long concert sets on my raspberry pi). AFAICT, there’s no way to only download the audio without incurring the unnecessary bandwidth and CPU usage.
Edit: I just checked and the CPU usage seems to be coming from the fact that I am asking yt-dlp to convert the audio format. Now I’m questioning whether it is actually the video that I’ve seen temporarily on disk or just the audio file in the youtube-native format.
It varies based on the age of the video, newer ones do indeed have separate audio downloads. You can force audio only with
yt-dlp -f bestaudio <url>
This will cause the script to only consider audio-only formats, if bandwidth is a concern. However, how it decides which one is “best” is beyond me. For example, I tried one video and got a webm that contains only an audio track:
~ $ yt-dlp -f bestaudio https://www.youtube.com/watch?v=dQw4w9WgXcQ
[youtube] Extracting URL: https://www.youtube.com/watch?v=dQw4w9WgXcQ
[youtube] dQw4w9WgXcQ: Downloading webpage
[youtube] dQw4w9WgXcQ: Downloading ios player API JSON
[youtube] dQw4w9WgXcQ: Downloading android player API JSON
[youtube] dQw4w9WgXcQ: Downloading m3u8 information
[info] dQw4w9WgXcQ: Downloading 1 format(s): 251
[download] Destination: /data/data/com.termux/files/home/storage/movies/ytdl/20091025__Rick_Astley_-_Never_Gonna_Give_You_Up_Official_Music_Video.webm
[download] 100% of 3.28MiB in 00:00:00 at 6.91MiB/s
Yes, I deleted my comment after I checked the command line I was using and testing it out. The CPU usage was because I was using the --audio-format flag with something other than m4a/opus or whatever it is that youtube uses natively. And the file I saw on disk was not the video, it was the audio file in youtube’s native format.
For reference, I’m using $ytdl -f "bestaudio/best" -ciw -o "$audioroot/%(title)s (%(upload_date>%Y-%m-%d)s).%(ext)s" --extract-audio --audio-quality 0 --audio-format mp3 "$vidurl".
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !piracy@lemmy.dbzer0.com
⚓ Dedicated to the discussion of digital piracy, including ethical problems and legal advancements.
YouTube audio quality isn’t amazing, but it’s not that bad if you get the right format. Using yt-dlp, you can get opus audio that sounds way better than mp3. Example:
yt-dlp -f 251 https://www.youtube.com/watch?v=dQw4w9WgXcQHere is an alternative Piped link(s): https://piped.video/watch?v=dQw4w9WgXcQ
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I’m open-source, check me out at GitHub.
Please somebody correct me if I’m wrong:
Separate audio is rarely available on streaming sites like YouTube, so if you’re ripping a song from it, you essentially are transcoding the files three times:
So no matter how great the quality is at either of these transcoding ends, you are still transcoding in lossy format at least two times
I haven’t seen a YouTube video without a separate audio track. If you use yt-dlp just add -F and the URL to get all available download options/formats. Then use -f and the number of the download option and URL to download the video/audio. Audio on YouTube is available in two formats most of the time: opus and m4a.
yt-dlp can just download the audio. It usually comes down in m4a at quality that I would describe as “very listenable”. So only the first of those three steps are mandatory if you do it that way.
yt-dlp -x <url>From my use of this, it downloads a video and extracts the audio stream from it. In fact, you can see this as it leaves the video file on disk while it’s extracting the audio, and the audio-extraction process uses a decent amount of cpu (takes a few minutes for long concert sets on my raspberry pi). AFAICT, there’s no way to only download the audio without incurring the unnecessary bandwidth and CPU usage.
Edit: I just checked and the CPU usage seems to be coming from the fact that I am asking yt-dlp to convert the audio format. Now I’m questioning whether it is actually the video that I’ve seen temporarily on disk or just the audio file in the youtube-native format.
It varies based on the age of the video, newer ones do indeed have separate audio downloads. You can force audio only with
yt-dlp -f bestaudio <url>
This will cause the script to only consider audio-only formats, if bandwidth is a concern. However, how it decides which one is “best” is beyond me. For example, I tried one video and got a webm that contains only an audio track:
Here is an alternative Piped link(s): https://piped.video/watch?v=dQw4w9WgXcQ
https://piped.video/watch?v=dQw4w9WgXcQ
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I’m open-source, check me out at GitHub.
Yes, I deleted my comment after I checked the command line I was using and testing it out. The CPU usage was because I was using the
--audio-formatflag with something other than m4a/opus or whatever it is that youtube uses natively. And the file I saw on disk was not the video, it was the audio file in youtube’s native format.For reference, I’m using
$ytdl -f "bestaudio/best" -ciw -o "$audioroot/%(title)s (%(upload_date>%Y-%m-%d)s).%(ext)s" --extract-audio --audio-quality 0 --audio-format mp3 "$vidurl".But is the source separate on YT’s server is what I’m wondering.
If not, there’s a transcoding stage