Hello all. I usually have been able to figure issues out, but this one has me stumped. I am running Windows 11 with the latest OBS. We used to have everything set perfect, but one day, the audio started cutting out. I am using a fairly cheap line in to USB audio capture device for audio from our mixer. The levels on the meter in obs look normal, but in YouTube, it is almost too quiet to hear. Any ideas?
Quiet Audio
- Thread starter fish2222
- Start date
What's "normal" to you on the meter? There are two conflicting mindsets that you have to keep straight, and live broadcast uses both simultaneously. So you *really* have to keep *that* straight!
---
Live work doesn't know what's coming, so you have to leave some headroom. That's where the -18dBFS rule of thumb comes from, plus an assumption about how "peaky" things usually are.
Mastered recordings know exactly what's coming, and they were typically distributed on lousy physical media compared to what the studio had to work with. Vinyl, cassette tape, etc. So they're turned all the way up as high as they'll go, to overcome the shortcomings of the physical media. The idea is that the consumer will keep their volume controls low, and thus reduce the noise from the medium itself.
Live broadcast has both sets of problems with few benefits. Live source, so you don't know what's coming, and a lousy distribution medium, so you want it loud to overcome that.
---
So for live broadcast, the raw signals use the "live" mindset, and you actually keep that mindset until you get close to the end. By the time you're *at* the end of the processing chain, you're solidly in the "mastered" camp, and the transition uses at least two stages of compression to keep it inaudible: a gentle compressor that's almost always active, and a "safety net" limiter.
If you don't compress, then you'll clip at a lower level than what it sounds like other broadcasts are getting away with effortlessly.
So what's a "normal" broadcast level to you? If it doesn't completely fill the meter, it's probably too low. And what processing chain do you have to keep it there and not go over?
---
YouTube is constantly changing how it works. It may or may not adjust the audio level for you to fit in their standard range, using a mathematical version of how we perceive volume, not the maximum sample level. If you're asking about a change, it could be that they used to bump you up and now they pass you through as-is.
---
Live work doesn't know what's coming, so you have to leave some headroom. That's where the -18dBFS rule of thumb comes from, plus an assumption about how "peaky" things usually are.
Mastered recordings know exactly what's coming, and they were typically distributed on lousy physical media compared to what the studio had to work with. Vinyl, cassette tape, etc. So they're turned all the way up as high as they'll go, to overcome the shortcomings of the physical media. The idea is that the consumer will keep their volume controls low, and thus reduce the noise from the medium itself.
Live broadcast has both sets of problems with few benefits. Live source, so you don't know what's coming, and a lousy distribution medium, so you want it loud to overcome that.
---
So for live broadcast, the raw signals use the "live" mindset, and you actually keep that mindset until you get close to the end. By the time you're *at* the end of the processing chain, you're solidly in the "mastered" camp, and the transition uses at least two stages of compression to keep it inaudible: a gentle compressor that's almost always active, and a "safety net" limiter.
If you don't compress, then you'll clip at a lower level than what it sounds like other broadcasts are getting away with effortlessly.
So what's a "normal" broadcast level to you? If it doesn't completely fill the meter, it's probably too low. And what processing chain do you have to keep it there and not go over?
---
YouTube is constantly changing how it works. It may or may not adjust the audio level for you to fit in their standard range, using a mathematical version of how we perceive volume, not the maximum sample level. If you're asking about a change, it could be that they used to bump you up and now they pass you through as-is.
Well, we only have one audio input, and on the meter in obs we try to keep it in the yellow. When the band plays, we used to have to drag the slider to around -7, and when someone is just talking we slide it back to 0. Now, we had to add the gain filter and turn it up to about 14 to get the same level at 0 that we used to have at -7. But even then, the live YouTube feed is really quiet and harder to hear. Hopefully this makes sense.
Here's what I have at church, screenshotted this morning during rehearsal (and killed the quality so the forum would accept the file size):
And what that broadcast actually was:
- YouTube
Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.
www.youtube.com
- YouTube
Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.
www.youtube.com
- The highest recent peak, as the small bar above the meter.
- The present peak, as what looks like the meter itself.
- The present average, as a small black bar a little bit lower down. -20dBFS as shown here.
- Left goes to the stream and recording, in stereo.
- Middle is mono to the headphones, to make sure it still works that way too. A lot of people have either a single speaker, or their "stereo" speakers are too close to each other to actually produce appreciable stereo.
- Right is stereo to the headphones, and what I normally listen to. It's identical to the broadcast.
Anyway, our external mixer is a Behringer X32, that has a TON of compression at several different stages, so the meter is pretty much pinned to the top no matter what's happening, except when something is clearly supposed to be quiet.
This is the individual channel compressor. Every input has its own one of these with its own settings. This one happens to be for the guy in the middle of the screenshot above, who normally leads the singing.
After that, the individual inputs go through the Front-of-House faders and then my post-fade aux faders to feed the Broadcast bus. The Broadcast bus master just has an anti-clipping limiter, which is another copy of the channel compressor above, but with the threshold and ratio all the way up, and a soft knee. If it was solely for anti-clipping, it'd be a hard knee just to stay out of the way as much as possible, but I also wanted it to still be "invisible" even then, so I had it start working a bit before that and ease into the limit.
The Broadcast bus then feeds a Matrix, along with the Speaking bus and the Room bus, so that we can have the Broadcast mix minus the room mics for the hall and hearing assistance, and also share the Speaking processing across several different mics and then send that to several different places by itself.
The Matrix master has a vintage LA-2A emulation, that works HARD!:
It's probably responsible for most of the overall volume leveling.
After that, on the same Matrix master, is another channel compressor, that is set as a limiter again, and designed along with the output fader and routing to put OBS's meter exactly where it needs to be:
Normally, there'd be about 22dB of makeup gain for something like this, with the threshold at -23dBFS, leaving 1dB of headroom, but between the output fader being all the way up (+10dB) for the ability to fade it down and put it back exactly where it was, and the difference between the board's pro line level and the USB card's consumer line level (~11dB), it ends up pretty close already.
10dB + 11dB = 21dB of effective gain following this limiter already, which with the cap at -23dBFS here, results in about -2dBFS peak in OBS, which is not quite what the screenshot up top has, but pretty close. We *were* kinda quiet this week; we normally have more musicians than that.
That's just the final limiter and level set. The volume leveling, to keep the meter there regardless, happens earlier, mostly with the LA-2A emulation.
OBS then passes that straight through, with no processing whatsoever, except for a noise gate to catch a small amount of ground loop noise on the analog wire between the X32 and the USB sound card. Once I get an isolator in there, that noise gate can come out too.
Thank you. We have a Presonus mixer with only one line out, and that is actually split between the laptop with obs, and a desktop with audacity to record the sermons so we can do podcasts. The strange thing is that audacity levels are still the same, so it has to be a setting in obs that has changed. I'm almost thinking a Windows update might have caused it, so maybe reloading Windows might fix it.