I know YouTube compresses the audio in videos, but I’ve noticed recently that the levels in my videos are all over the place. It never used to be this bad I’m sure.
For example, if I have a melodic intro then the audio starts off at quite a high level, but then as soon as the kick drum comes in the audio level drops by half and I have to turn the device volume up to compensate. So clearly some seriously heavy compression going on.
I’ve been recording and uploading videos direct from my iPad, and did read that I should be transferring to the computer and uploading the file via the web browser. I’m not sure why that would make a different though.
I wonder whether there are any best practices when uploading music videos to YouTube? What db should I be aiming for? Maybe I should not make the videos so loud in the first place so the compression effects is less noticeable, but then my videos would be lower volume than other people.
Anyone have any thoughts/advice?
Also, on a slightly related note. Has anyone tried recording directly from the Digitakt/Syntakt/Digitone to iPad via USB-C? I find the audio level to be quite low. Is there a way to increase it on the device? Obviously if recording into a DAW you can adjust gain (with utility plug-in in Ableton for example) but when recording video on the iPad this isn’t possible.
I’ve not uploaded anything for a few years now, however, i ran some experiments to see if i could get around the compression youtube uses. Bassicly, i used a constant noise generator above our hearing range at constant amplitude level higher than the audio track and this got me the best results.
Interestingly, this technique can often fool some platforms auto copyright bot, and by adding another noise gen at the lower end it confuses it even more.
There must be something weird going on in your mix. Youtube compresses your signal quite a bit but it usually doesn’t pump up the signal like that. Maybe your levels are way too high in the first place, idk.
My audiolevels are maxing at -0,5 db for youtube and i only compress the signal 2-4 db. My tracks have a lower volume than others though. But i don’t care as long as the sound is good.
If you search for “loudneds targets for streaming platforms”, you’ll find Youtube recommends -1.0 dBTP and -13 to -15 LUFS (integrated/long term) and not more than -9 LUFS on peaks (short term) . If you get your mix to sit within that range, youtube shouldn’t mess too much with your audio.
YouTube recommends exporting your video either as the DVD-compliant MPEG-2 format or as an MPEG-4 using the H.264 video codec.
For MPEG-2, set audio codec to either MPEG Layer II or Dolby AC-3 with an audio bit rate of 128kbps.
For MPEG-4 AAC audio codec with an audio bit rate of 128kbps or higher is recommended.
Sticking to YT’s recommendations should avoid too much unwanted processing resulting in negative side effects, I guess…
Youlean Loudness meter has a youtube metering option, and as @Schnork says, its -14 LUFS. with a little range around that.
I think there is a free version that you play your track over, and it will tell you where its peaking and where it needs gain, with an overall average.
I use Izotope Ozone, which has a mastering assistant for streaming, but it really destroys a mix if you are heavy into it… I am not… I cant master for shit, and noone is looking for my music, so its a good tool for me to use.
youtube have probably changed their compression technique since, so i have no idea how this will fare nowadays, let me know how it goes if anyone trys!
…all “compressed” online codecs might vary ur audioresolution but not ur dynamicbehaviours…
if u experience such things, ur overall loudness, all recommended luv levels and final dynamics are totally off the charts…
try some limiting, keep an eye on luv recommendations or dare to upload audio content that’s peaking around -6 db…if u have no final treatments on ur material, that will sound much quieter in overall comparison, but at least u captured all dynamics without such bad surprises…
but whatsoever, compressed codecs are a totally different thing than compressed signals…
one is all about data stream and load efficiency…the other for taming audio dynamics…
“they” take care of all final data flow …but first u must take care of ur content…
Just stick to the loudness target. -13 to -15 LUFS integrated, not more than -9 LUFS short term. For the short term meassurements loop the busiest/loudest section of your track. For meassuring integrated let the mix play through.
Youlean Loudness Meter will work fine for this (IIRC you can toggle between integrated and short term LUFS in the free version). Insert a limiter in front of the loudness meter, set limiter to -1dBTP (dB True Peak) and add gain watching your LUFS meter. If you don’t have a true peak limiter, set the limiter to a little less than -1dB and if possible enable oversampling (with high oversampling, you won’t have much of a problem with intersample peaks.
There shouldn’t be any obvious difference between your mix and the youtube stream then.
The issue I think for me is that I want to just record jams/tracks straight into my iPad and then upload to YouTube directly, so up until now I’ve just set levels by ear. I sometimes check the audio levels on an app first before switching to the camera just to make sure nothing is clipping, but if I’m honest I’ve never heard of LFUS before.
So I think I may have found a solution to make things easy for me. I bought a Presonus Revelator io44 audio interface which works with the iPad. It has built in DSP including a limiter and uses a Universal Control app to set up the effect chains. It doesn’t look like you can add a limiter to the master, but I can add a limiter on the Line In channel strip set to -1.00db. Do you think this is a good solution?
I don’t think you need to record with the limiter on your audio interface. It’s probably fine as a safety limiter, but if you have a DAW or audio editor (I guess you do?), then you can just continue to set levels by ear and then record your jam as you did before. Then throw the audio into a DAW or audio editor and try to get it to the loudness target.
-13 to -15 LUFS integrated is fairly moderate, with any half decent mix it should be possible to get there just by applying gain and maybe a bit of compression or limiting (1-2 db gain reduction).
Maybe do a few test sessions to see how your LUFS are before you record the next jam, could be all you need is a bit of gain boost or maybe even lowering the levels.
It’s just a different way of meassuring audio levels.
You don’t have to treat the numbers any different than the RMS or peak level values you’re used to, it’s just a different system.
Think as the LUFS integrated as a meassurement over longer time, like we use RMS to look at how audio behaves over longer times in contrast to peak levels which only tell you about whats happening momentarily (LUFS momentary/short term).
LUFS takes into account how humans hear so when trying to work with standard levels, LUFS is better than RMS, because it’s more akin to how humans perceive audio.
And streaming platforms need some form of standardization, nobody would want the next video to blow there ears out.
Instead of looking at your peak meter if it’s in the range of -3dbFS to -1dbFS (or whatever you’d shoot for) and you RMS values, just try to match the LUFS.
If you haven’t payed much attention to dBFS and RMS before, like I said, you don’t have to dive into the maths behind LUFS, just throw the Youlean Loudness Meter or another plugin that shows LUFS on the track and dial in the gain until the numbers are what you want. If you can’t get there, try compression or limiting or go back to the levels on your gear, but the Youtube loudness targets are really moderate levels, so it’s probably much more easy than you think.
