For us novice-to-intermediate-level streamers, nothing is worse than being told during the stream that our audio sucks. Sometimes we end up talking for 15 minutes and no one can hear us over the game audio, or maybe we can’t find that happy medium between cut-scene dialog and our own snappy commentary.
My audio issues started because I have the kind of deep voice that doesn’t lend itself to traveling very far unless I put some oomph into it, which I didn’t want to do on stream lest it seems like I’m shouting at the viewers. So to try and boost my voice when streaming I added a gain filter to my mic input in Streamlabs OBS (SLOBS) hoping to increase the volume, but that was only the beginning of my plunge into the world of streaming audio. By the time I was done messing with settings, I was very pleased with the outcome, and figured that this could be information that other new-to-intermediate streamers might also find useful.
Note that this info pertains specifically to SLOBS, but should have analogues in XSplit and OBS Studio which I am not covering here.
Before We Begin…
There’s an important step to consider when trying to get your broadcast audio correct: listening to yourself. I know, no one likes to listen to themselves talk, but it’s really important that you suck it up, go into the settings of the application (the gear at the top right next to your account name) and select the Advanced option on the right side nav.
Under Audio you’ll see the “Audio Monitoring Device”. This is what you’ll use to listen to yourself. “Default” is probably your desktop speakers, but if you have another output device that you can plug headphones into (or if you have a set of Bluetooth earbuds you can use), select it here. Also, make sure the “Disable Windows audio ducking” box is unchecked.
One more setting before we start. On your input meters there’s a gear icon on the top right which is really useful for advanced audio tricks, but here we want to open it and select Audio Monitoring for all of your inputs. Change each one to Monitor Only or Monitor and Output depending on how you want to conduct your test.
Note: depending on which device you have as “Default”, you might end up with a lot of audio through one output and not realize that any modifications are taking effect. I advise you to use the local recording option of SLOBS to record a local video if you don’t want to use live audio monitoring. It’s slower and more cumbersome, but it’s also truer to what your viewers will see than what you’re likely to get using the audio monitoring option. Alternatively, you can fire up a live stream if you have viewers who are accommodating and someone on the outside willing to help you test your settings.
FYI, live audio monitoring is really disorienting. Make sure you turn this off before your next broadcast, or you’ll sound drunk.
Gain on the mic is important if your voice isn’t very loud. No one should be forced to stream with a mic up their nose, so adding a gain filter to your mic input is a good way to boost the output by boosting the mic sensitivity. Naturally, if you have an advanced mic like the Blue Yeti, there’s a knob for gain that you should play with before you add more layers but if that still doesn’t do it for you, here’s what you need to do:
First, find your audio meters. What we want to do is touch the gear icon beneath the Mic/Aux meter, on the far right.
We add filters by clicking on the + icon on the left. This will give us a lot of choices (not depicted), but Gain is the first one in the list. Adding it will allow you to specify how much of that oomph you want to add on top of your input. Positive values means that the mic will pick up more noise, and negative values mean that you will have to shout to be heard. Save that and test by talking at normal volume, louder, quieter, and move around in your chair to ensure that the mic can hear you wherever you expect your head to wander during the stream (super important for non-headset mics with directional pickups).
Naturally, once the mic is more “sensitive” it might start picking up background noise. Being in the basement, viewers might often hear the sound of my HVAC system kicking in or the dog walking around in the background. To combat this garbage noise, we need to add another filter, the Noise Gate.
A noise gate is what it sounds like: It “shuts the gate” on noise that exceeds a certain decibel level, and opens it again when things quiet down. The Close Threshold is the setting at which your mic will stop transmitting audio, and the Open Threshold is the level at which it’ll resume. You can play with these in real time while you monitor the input, but if you are doing this by recording, you’ll have to let yourself know what you’re doing unless you get all meta and record the screen which displays your SLOBS console…Trippy! Attack, Hold, and Release indicate how quickly the gate shuts, how long it’ll hold before it starts to open, and how quickly it’ll open after that, respectively.
If garbage noise is still getting picked up, like breathing or the wet sound that your eyes make as you scan your surroundings, there’s always the Noise Suppression filter.
The lower this value, the harder the mic will have to work in order to pick up sounds. Because your mouth will be close to the mic — at least, closer than whatever background noise it’s currently picking up — you should play with this on the lower end until no one can hear your eyeballs moving, because that’s just nasty.
Duck, Duck, Duck…Ducking
People should be able to hear you better now that you’re louder and there’s not as much garbage noise in the background, but there’s still one noise you can’t afford to get rid of as a live-streamer, and that’s your game audio. Not all game audio is relevant, but it’s difficult to find the happy medium between in-game audio and the input of your mic. Too much game audio and no one will be able to hear you; too much mic and the flavor of the game will be lost — and if you aren’t using advanced techniques like virtual audio cables to pipe other audio into the stream, like from Discord, then everything runs through the desktop input and collides with your mic.
Ducking is the act of attenuating one audio source when another is activated, like turning down the game volume when you start speaking. Getting the right game-to-you volume ratio is hard because sometimes different games have different volume levels, and sometimes this varies wildly within the game itself. Ducking can help, and we have a filter for that.
This time, we want to add a new audio filter called Compressor, and this time we want to add it to the Desktop Audio input.
Compressor, as the name implies, compresses the audio of the input it’s applied to. The Ratio value indicates how much to compress the audio (here, the desktop audio), and the higher the number the more the audio will attenuate. Because we’ll be using this in tandem with another audio input, the ratio represents “units of desktop audio to units of other audio” where in this case the “other audio” is the mic input which we’ll specify in a bit.
Like the noise gate on the mic, the Threshold, Attack, and Release values determine when the compression kicks in, how quickly it kicks in, and how quickly it’s allowed to return to normal after the mic source stops transmitting. There’s no “golden value” here, so you’ll have to play around with the settings. I like to have the mic be pretty aggressive so that it really stomps on the desktop volume quickly, and then to wait a few seconds before ramping back up, just in case I pause when speaking. The Output Gain should just be left as is.
The important setting here is the Sidechain/Ducking Source which should be set to your mic/aux input. This is what will trigger the desktop compression as the other part of the initial ratio value.
You can visualize the results of this filter by playing a game or YouTube video and watching the desktop meter in SLOBS hit somewhere in the yellow zone (which I learned is the “optimal zone” for good audio levels). When you start talking on the mic, the desktop meter should shrink back towards the lower end of the bar as the mic meter creeps towards the yellow of its own bar. The audio results will be most noticeable when you record a local video, which is strongly advised as a means of testing things out to make sure they are to your liking.
I am not an audio expert, so I managed to cobble this info together through various sources. With a little trial and effort, I was able to cut out a bunch of ancillary noise that I don’t think I even realized was being broadcast on the stream. I believe my voice will be louder now, and more importantly, I can leave my desktop input at full volume to be controlled by the levels from the mic. This was super important as the desktop audio was also the input through which Discord conversations were piped. Having to turn down the desktop input meant that Discord was also affected and barely audible.
Hopefully this will help you if you use SLOBS, and if you use Xsplit or OBSS you’ll be able to take some of this knowledge and apply it to those applications as well.