|
|
|
07-12-2019, 08:58 AM
|
#1
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Original vs SoundCloud vs YouTube
See this post as to why I performed this test:
https://forum.cockos.com/showpost.ph...5&postcount=28
Using sample audio from some random track I recorded last year @ 48/24 WAV, I exported as MP4/48k from reaper, uploaded to YouTube - downloaded from YouTube from my YT control panel (this allows me to download the YT processed copy directly). I took that and placed it back on the timeline in Reaper below the original 48k WAV, sample aligned, normalized. I then randomly switch between the YT version and the original WAV multiple times and rendered that out at 48/24:
http://wallsonic.com/public/posts/YTAT.wav
Conclusion: It doesn't null but listening tells me it isn't a night and day difference that I personally would have much that concern with.
I did the exact same test with SoundCloud:
http://wallsonic.com/public/posts/SCAT.wav
Conclusion: I can hear the difference from the SoundCloud test better best I can tell, and can explain what that is later, though subtle, worse than YT in my opinion. There is also some other crazy weirdness going on with SoundCloud that I can explain later but this is why I prefer YT over SC and personally don't think YT is as bad as people tend to claim.
Last edited by karbomusic; 07-12-2019 at 09:55 AM.
|
|
|
07-12-2019, 11:16 AM
|
#2
|
Human being with feelings
Join Date: Jan 2018
Posts: 1,651
|
Interesting, I will have to try this test somehow blind, I feel like i can hear a small difference but maybe only because you said so, haha. Like you said, it's not night/day difference.
I think this is important to understand, I often get material from clients through SC, sometimes this is the only source and it will be helpful to have more info to share with them.
Looking forward to the explanations
|
|
|
07-12-2019, 11:29 AM
|
#3
|
Human being with feelings
Join Date: May 2009
Location: Polandia
Posts: 3,578
|
You guys are golden eared, I randomly skip all over the samples and hear the same sounding stuff, would never guess it's comped off different quality parts.
|
|
|
07-12-2019, 11:29 AM
|
#4
|
Human being with feelings
Join Date: Jan 2006
Posts: 1,595
|
Interesting test.
In my own uploads I've been pretty impressed by Youtube's audio quality.
|
|
|
07-12-2019, 12:19 PM
|
#5
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Just to make sure I didn't confuse anyone, the first sample YTAT.wav is both the original WAV and the YT compressed version, but I muted one or the other in random places as it plays so that one may/should/shouldn't/can't hear where it changes from one to the other - I find this a much better test than many. Same thing for the SoundCloud example - which I can hear but still it's random flips between SC and original WAV. Here's a screen shot to show what I mean but I changed the split points and number of splits as to not give away the positions:
So in that regard it is blind enough for what we usually care about. IOW, if YT was destroying audio as net lore often claims, everyone here would be saying "oh wow, did you here where it went from original to YT @ x,y and z seconds in? That's sort of the point of the comparison.
Last edited by karbomusic; 07-12-2019 at 12:25 PM.
|
|
|
07-12-2019, 12:37 PM
|
#6
|
Human being with feelings
Join Date: Jan 2006
Posts: 1,595
|
I think you did a great job in balancing the levels of the segments of the original file and the YT compressed "implants". Implants, heh. Quality-wise, YT hold up here.
I said that, actually, before seeing the waveforms.
|
|
|
07-12-2019, 12:39 PM
|
#7
|
Human being with feelings
Join Date: Jan 2006
Posts: 1,595
|
And thanks for showing the waveforms, karbo. I'm going to listen again while watching.
|
|
|
07-12-2019, 12:51 PM
|
#8
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Originally Posted by Timothy Lawler
And thanks for showing the waveforms, karbo. I'm going to listen again while watching.
|
I changed the transition positions in the screenshot so as not to give it away but once a few people listen, I'll see if I have copy of the original timestamps where they switch.
|
|
|
07-12-2019, 02:46 PM
|
#9
|
Human being with feelings
Join Date: Dec 2012
Location: Germany
Posts: 3,014
|
thanks for this test, karbo. It backs my impression that people are often bashing youtube audio for no reason. However, I'm sure that the final quality of an uploaded video can vary and will surely depend on the quality/resolution of the source audio. Also, when using "Youtube Video and Audio downloader" Opera extension (raylothian) I noted severe artefacts in several aac audio-only files (even high kpbs ones) while the corresponding opus audio-only file @ 160 kbps sounded pristine.
btw, what audio properties did your MP4 file have when you uploaded it? Whenever possible, you should choose to render videos containing wave as audio to avoid an unnecessary second lossy compression pass.
.
|
|
|
07-12-2019, 02:50 PM
|
#10
|
Human being with feelings
Join Date: Aug 2014
Posts: 11,044
|
YouTube audio is much better than it used to be. It also used to be that audio quality would take a big nosedive with running a video at lower resolution to save bandwidth, but I'm not sure if that is still the case.
By this point the warnings about it are getting toward old wives' tale status.
On the other hand, something has always been iffy about Souncloud's encoding.
|
|
|
07-12-2019, 02:52 PM
|
#11
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Originally Posted by SonicAxiom
However, I'm sure that the final quality of an uploaded video can vary and will surely depend on the quality/resolution of the source audio.
|
Yes, of course, it can't turn a shit render into gold.
Quote:
Also, when using "Youtube Video and Audio downloader" Opera extension (raylothian) I noted severe artefacts in several aac audio-only files (even high kpbs ones) while the corresponding opus audio-only file @ 160 kbps sounded pristine.
|
I agree which is why I got it directly from my upload section - however my comparison concern is always what people say they hear when just playing the vid on YT - that said, YT has NOT always been this good. I've been on YT since right after it began around 2007 - back then you had to jump through hoops just to get it in stereo much less acceptable quality.
Now to be fair, I wonder how many times people have low bandwidth and get reduced quality due to the connection where YT is only trying to get an uninterrupted stream to the user, that's valid to the ears but not really a YT upload issue.
Quote:
btw, what audio properties did your MP4 file have when you uploaded it? Whenever possible, you should choose to render videos containing wave as audio to avoid an unnecessary second lossy compression pass.
|
Well, since I wasn't really intending to post it at first, I just created at 1920x1080 PNG, dropped it into reaper, dropped a 48k/24bit WAV into reaper, then rendered as 1920x1080 MP4/MOV @48k @95% quality and dragged that directly to the YT upload.
|
|
|
07-12-2019, 02:57 PM
|
#12
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Originally Posted by Judders
YouTube audio is much better than it used to be. It also used to be that audio quality would take a big nosedive with running a video at lower resolution to save bandwidth, but I'm not sure if that is still the case.
|
Right and I don't see that as an upload/recompress issue, it's a bandwith delivery issue and should be a separate subject IMHO - but at the same time we should clarify like we are doing now so that people don't conflate the two. I actually think you guys can hear the SC compression if you close your eyes and listen. I could be biased but I noticed it by accident as in "wtf was that" when I wasn't looking, when I looked it was in fact the transition from WAV to SC.
Quote:
On the other hand, something has always been iffy about Souncloud's encoding.
|
There is some weird shit going on. When null testing against the original, the diff gets really loud in the high freq when that high hat comes in, then suddenly darker when it leaves. And in some spots it just seems to change out of nowhere.
The YT diff actually sounds worse from a swooshy standpoint at first until you realize it's still ~6dB lower than the SC diff across the board and it doesn't do that weird compression thing to the high end when the high end is present. I've always though SC sounded a little weird but assumed it was me, but today I created an account just for this test and not so sure I was crazy after all.
|
|
|
07-12-2019, 03:00 PM
|
#13
|
Human being with feelings
Join Date: Dec 2012
Location: Germany
Posts: 3,014
|
Quote:
Originally Posted by karbomusic
Well, since I wasn't really intending to post it at first, I just created at 1920x1080 PNG, dropped it into reaper, dropped a 48k/24bit WAV into reaper, then rendered as 1920x1080 MP4/MOV @48k @95% quality and dragged that directly to the YT upload.
|
ok, but what were the audio rendering parameters of that MOV/MP4 (format, bitrate)? By default, Reaper selects AAC @ 128 kbps when you choose QT/MOV/MP4 video.
.
|
|
|
07-12-2019, 03:02 PM
|
#14
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Oops sorry... 16 bit 128kbps.
|
|
|
07-12-2019, 03:06 PM
|
#15
|
Human being with feelings
Join Date: May 2016
Posts: 2,065
|
YouTube can sound pretty decent if EVERYBODY involved with the mastering and video preparation does it right.
SoundCloud is hamstrung by their 64kbps OPUS stream (unless they changed it recently) so even if you provide a 24-bit WAV, with enough peak-headroom to survive the encoder, it's still going to be pretty garbled. Sample rate changes during the encode and playback are a concern too.
Here's a fun thread about optimizing audio for YouTube:
https://www.gearslutz.com/board/mast...ic-videos.html
__________________
REAPER, just script it bro.
|
|
|
07-12-2019, 03:07 PM
|
#16
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
The GS thread looks interesting - I tend to use this since it could change and is straight from the source:
https://support.google.com/youtube/answer/1722171?hl=en
That said, part of the point here is what people actually notice and hear - I have sort of grown to hate those "let me find some tool that shows you a graph that says it isn't perfect @ -120dBFS so you should stress out for absolutely no reason" examples.
You do raise an additional good point that YT gets blamed for bad mastering/uploads.
|
|
|
07-12-2019, 03:14 PM
|
#17
|
Human being with feelings
Join Date: Aug 2014
Posts: 11,044
|
Quote:
Originally Posted by karbomusic
There is some weird shit going on. When null testing against the original, the diff gets really loud in the high freq when that high hat comes in, then suddenly darker when it leaves. And in some spots it just seems to change out of nowhere.
|
Yeah, the SC distortion is not level dependent. Years ago I noticed a quiet passage with some high register instrument playing would distort no matter what I did. I messaged them and they just said "yeah, it's our encoding, we're working on it".
|
|
|
07-12-2019, 03:18 PM
|
#18
|
Human being with feelings
Join Date: Dec 2012
Location: Germany
Posts: 3,014
|
Quote:
Originally Posted by karbomusic
Oops sorry... 16 bit 128kbps.
|
ok, PCM Wave, 16 bit is the correct choice to A) keep the original audio quality intact in the rendered file and B) make the unavoidable youtube audio compression pass less damaging.
Any specific reason why you do not opt for h.264 as the video codec? This gives far better file size vs. video quality ratio and is much more optimized for streaming than MJPEG. Essentially, it's exactly what youtube uses to re-compress videos during upload. I'm using these settings and quality remains flawless:
.
|
|
|
07-12-2019, 03:44 PM
|
#19
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Any specific reason why you do not opt for h.264 as the video codec?
|
Simple... because my version of Reaper, VLC or FFMPEG doesn't have that option at all so I should probably figure out which is out of date.
That said, yes of course, all my actual YT work comes from davinci where I always use PCM 48k/16 bit but this was a quick test using reaper and didn't even notice, but seems like that further strengthens what we are all saying.
|
|
|
07-12-2019, 03:51 PM
|
#20
|
Human being with feelings
Join Date: Feb 2013
Location: Northeast Michigan
Posts: 3,460
|
Just curious, Karbo... what is wallsonic? I can't find any info on it????
|
|
|
07-12-2019, 04:00 PM
|
#21
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Originally Posted by Jeffsounds
Just curious, Karbo... what is wallsonic? I can't find any info on it????
|
I own 30 or 40 domain names cuz I'm a geek like that. Comes in handy since I have a honking server in the other room and that makes it easy to point DNS to my web server for various personal web sites. I only use it for posts now when I need to host a WAV file since other services are going to compress to something else. I had to limit its public use because too many connections coming into my house from all over the world lol. IOW, those two WAVs in the link are sitting on a computer about 20 feet from me.
|
|
|
07-12-2019, 05:34 PM
|
#23
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Originally Posted by SonicAxiom
|
Thanks!
|
|
|
07-12-2019, 09:41 PM
|
#24
|
Human being with feelings
Join Date: Feb 2013
Location: Northeast Michigan
Posts: 3,460
|
Quote:
Originally Posted by karbomusic
I own 30 or 40 domain names cuz I'm a geek like that. Comes in handy since I have a honking server in the other room and that makes it easy to point DNS to my web server for various personal web sites. I only use it for posts now when I need to host a WAV file since other services are going to compress to something else. I had to limit its public use because too many connections coming into my house from all over the world lol. IOW, those two WAVs in the link are sitting on a computer about 20 feet from me.
|
Very cool! Thanks for the info. I thought it was another Dropbox type thing, which it is, but only for you. Does the player that pops up do anything to the music or just play it straight?
|
|
|
07-12-2019, 11:55 PM
|
#25
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
That's invoked by the browser when it sees .WAV.
|
|
|
07-13-2019, 04:03 AM
|
#26
|
Human being with feelings
Join Date: Jun 2011
Location: Belgium
Posts: 5,246
|
Can't hear any difference atm, in a short listening test. I'll try again later, when my tinnitus has died down...
__________________
In a time of deceit telling the truth is a revolutionary act.
George Orwell
|
|
|
03-27-2022, 10:34 PM
|
#27
|
Human being with feelings
Join Date: Jan 2016
Location: Switzerland
Posts: 130
|
Hi all, I recently tried to do this comparison and found Soundcloud really messing up the high end. I made a video here:
https://youtu.be/3CPvebjpfJ4
I don't really know why or have a solution, but just some analysis. Likely it is poor compression settings they are using.
|
|
|
03-27-2022, 11:41 PM
|
#28
|
Human being with feelings
Join Date: Jun 2015
Posts: 323
|
That’s expected.
Soundcloud compression is at 128kbps, idk if they switched to aac instead of mp3, but audio quality is the worst among the platforms i know and use ( i don’t use soundcloud this days because of this).
Youtube compression should be AAC or FLAC i believe, but quality is pretty good if you upload an high quality file.
When you export a video you can choose the audio format, and FLAC should be the recomended for YT.
Sometimes there’s a very strange cut at around 16khz and i believe it depends on a compressed upload format being compressed again by YT.
Also resampling is a factor, youtube audio should be @44.1khz as per specifications.
By the way, null test are supposed to fail against a wav because of compression and/or samplerate conversion (unless you use flac or wav @44.1khz during the encoding of the video).
Here you can find some specs:
https://support.google.com/youtube/answer/6039860?hl=en
Upload format is very crucial for YT audio quality
|
|
|
03-28-2022, 05:16 AM
|
#29
|
Human being with feelings
Join Date: Aug 2011
Location: All Hallows End
Posts: 2,132
|
Quote:
Originally Posted by Judders
YouTube audio is much better than it used to be.
|
It used to be really awful
|
|
|
Thread Tools |
|
Display Modes |
Linear Mode
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT -7. The time now is 06:49 PM.
|