Low Gain in YouTube?

[giste]

Hello community!
I've been using OBS for professional streaming for many years. I'm quite happy with it, great tool! (I made my donation thou haha)
Nevertheless, I do find an issue while streaming to YouTube from OBS. I have the feeling that the audio is quite low in the streaming compared to the audio levels I monitor in the program.
As far as I know, -10db is a good level for your sound, even less (-15db) but if I stream with this level, the audio in the streaming video on YouTube will be way low. Why is this? It is like it loses some gain on its way from Obs to YouTube, what do you think? I now have to stream with my audio levels almost in 0db but I think that is not a good solution :(

I don't find the problem while recording, just streaming. In fact, I have just listened to my video that has been both recorded and streamed, and I listen to it differently.

Let me know your thoughts. Thanks!

 

[AaronD]

As far as I know, -10db is a good level for your sound, even less (-15db)

Is that peak? Or average?

Peak needs to be full-scale, or nearly so. 0dBFS or -1dBFS. Compress and limit to keep it there no matter what, without ever going over. Bad compression settings are clearly audible, but good ones can do that without ever being noticed.

Once you have a compressor setting that sounds good, and puts the peak at full-scale, you'll probably find that the average is about where YouTube wants it as well.

That's the difference between "live" and "broadcast" mentalities. Your "good level" is actually a bit high for "live" - it can still allow an unexpected peak to clip - and definitely low for "broadcast". The compressor translates between the two, usually at the last moment as it goes to the encoder.

 

[giste]

Hello Aaron! Thanks for your reply.
I will start using compressors to keep my signal at a higher level. It is curious thou, because when I do video editing, the average is -10 or even -15. But I will try as you say, keeping it almost -0 by default.

I don't understand your last paragraph (live vs. broadcast). Could you please explain it further?

best

 

[AaronD]

I don't understand your last paragraph (live vs. broadcast). Could you please explain it further?

Live:

You don't know what's coming, and a lot of things have "rogue peaks". Too-close plosives, extra hard drum hits or guitar plucks, etc. So you leave a fair amount of headroom to allow those peaks to happen without clipping. -18dBFS seems to be a de-facto standard for digital, but nothing says you *have* to be exactly there. Just don't let it clip.

If it's only for a live concert, you might just keep it that way all the way through the system so that the PA also produces those rogue peaks. That's a big part of "feeling live", so a live sound engineer that compresses individual channels for a sort of "automix", needs to be careful to not overdo it and kill that.

---

Broadcast:

You have a terrible distribution medium, like vinyl, cassette tape, radio, TV, internet stream, etc. Fixed amount of noise and other nastiness, regardless of signal level, so the larger (louder) signal you can put on it, the better. The end-listeners have their own volume controls, so the lower you can make them set it, the less they'll notice the deficiencies of the medium. That, plus the "loudness war", puts everything right up to full-scale.

It's relatively easy to do that for a studio session, and slightly harder for a live broadcast because you still don't know what's coming. So a live broadcast tends to have more aggressive compression to catch and squash something that doesn't exist yet.

At any rate, viewers expect the audio to come to them at full-scale, and their volume controls are set for that. Having to turn it up for you and (forget to) turn it down for someone else, is mildly annoying.

 

[giste]

Got you now dude! This is not about music but interviews, conferences and so on, so it's much easier.
Thanks for your help! I will use OBS compressors from now on.

 

[AaronD]

Yeah, that would be easier...in some ways. But harder in others. Fewer channels to mix, and have the result sound good, but we're a bit more sensitive to a single voice being messed with, than we are for musical instruments. So it's more important to have the compressor set *right*, whatever that ends up being for you, so it still sounds "natural" and also well-behaved technically.

I'd start with an instant Attack and slow Release (maybe 150ms or so), Threshold all the way up so it doesn't do anything yet, and a fairly high Ratio. Output or Makeup Gain at 0dB. Look where the meter is (assuming that your preamp gain is already set to never clip), then set the Threshold a little bit lower than that so the compressor is working most of the time. It'll be softer now, but that's what the Output/Makeup Gain is for. Turn that up to put the meter close to full-scale without ever going over.

If you need two stages of compression - low Ratio followed by high Ratio - that's perfectly okay. And you might also want a Limiter as the very last thing, just as a safety net that rarely does anything, but guarantees that you never go over.

My rig uses a DAW plugin, outside of OBS, to give me a really-soft-knee limiter as a sort of "auto Ratio". OBS doesn't do that, but you can approximate it with multiple stages of hard-knee compression, which OBS does have, with progressively higher Ratios, and Thresholds set for the (metered) output level of the previous stage.

 

[giste]

copy that! I will try it.
again, thanks a million mate