Audio tips for newcomers

There is a lot of bad advise on the internet, so I thought I'd use some of my video and audio production background to offer something useful. I'm not trying to call anyone out, but a lot of myths have been circulating for so long that people truly believe them and don't even know they are giving bad advice. So if you are new to this audio production thing, I'm here to tell you that a) you're probably overthinking it and b) throwing money at the problem will not solve it.

How do I get rid of noise/echo?

Let's start at the beginning: your physical space. I see posts almost every day asking how to get rid of noise. Should I use a gate? RTX Voice? Everyone wants to throw more processing at the problem, but the answer is simpler than that: just get rid of the noise. Garbage in, garbage out, as the saying goes. Noise reduction processing should be the last resort, not first. No matter how much you spend on a mic, it's still going to pick up the noise and the reflections in your room. If you don't want to hear your computer fans, buy quieter fans and/or put your computer on the floor instead of your desk (mini rant, people just need to stop putting computers on desks, it's terrible for noise control). If you don't want to hear your keyboard clicking, get quieter switches. If you don't want your voice to echo, get acoustic treatment. Acoustic treatment is an entire other subject, but you do not need to permanently mount it and you don't need to spend a lot. You can just hang moving blankets on mic stands and it will work well enough. The more you try to fix these problems with processing, the worse your voice will sound. Yeah, AI processing is pretty good these days, but there's still no way to remove the noise without altering your voice.

Side note: noise suppression is also usually the reason why your voice cuts out when you're yelling or making weird noises. Another reason not to use it.

Should I get a dynamic mic?

A lot of people will tell you to buy a dynamic mic because it's better at background noise rejection. This is one of the oldest and most pervasive myths out there and I really can't even blame people for thinking this. Heck, even some mic brands have taken advantage of it for marketing purposes. But a dynamic mic is just a type of microphone technology, there's no particular magic about it. Many dynamic mics happen to do well because they are nearly all designed to be used close to a loud source and for live music where off-axis rejection is really important. None of that has anything to do with them being dynamic. Case in point, shotgun mics (you know, those long mics on boom poles) have even better off-axis rejection, but those are condenser mics. There are also condenser mics made for live vocals on stage that will share the same off-axis rejection, pop filter, and shock rejection as a similar dynamic mic. What is true is that a large diaphragm, studio condenser mics are generally not designed with good off-axis rejection in mind because they are expected to be used in treated studio spaces. They are meant to be used farther away from your mouth, on a stand with a shockmount, and with a pop filter. So yeah, many dynamic mics will do a better job, but it's not helpful to just recommend any dynamic mic as many of them are no better than any studio condenser. It's much more important to look at the specific polar pattern of the mic you are interested in to see how directional that specific mic is. It can also show you specifically where the mic will pick up the least sound so you can put your keyboard there, for example.

Basic OBS audio setup

Before you add any filters, we just need to set your gain properly. Speak (or yell) into the mic as loudly as you are ever likely to speak while streaming. If you are clipping, which is to say if you hit the very top of the meter in OBS and the whole meter turns red, your gain is set too loud. Turn it down until you stop clipping. If you are much quieter than that, turn it up until you start clipping, then back off until you stop. Lots of people have opinions about what actual dB value you "should" be hitting, but it's all completely irrelevant. You just don't want to be clipping. Don't overthink it.

If you are using a normal audio interface, you may notice that your computer groups the inputs. For example, if you have two inputs, Windows will report them as inputs 1+2, with input 1 being the left channel and input 2 being the right channel. In OBS, you will need to go into the advanced audio properties and tell it to downmix to mono. You will also want to set the balance slider to the left for input 1 and to the right for input 2. After doing that, you will probably see that you actually cannot clip no matter how much you turn up the gain. This is because when you downmix two channels to a single mono channel, OBS reduces the level, essentially dividing by the number of channels. This makes sense if it's actually a stereo signal, but in this case, it isn't, so you will need to add a gain filter and set it to +6 dB to account for the difference. You only need to do this if your OS groups your mono inputs into stereo pairs! Some interfaces (like the Wave XLR, I believe) treat the input as if all of this was already done, which is to say that OBS will see it as a stereo signal with your voice on both channels.

Filters

Originally, I wrote an entire novel here, and maybe I will make a separate post for that, but honestly, the truth is that you just need to set up EQ and compression to your own preferences. My recommendation is to record a sample of you talking (just record it in OBS), then set that up as a VLC source with monitoring enabled, then add filters to that so you can hear what your changes sound like. Once you have a sound you like, copy those over to your mic track.

As a general guideline, compression is there to control your dynamic range, the difference between the loudest and quietest sounds. If you want people to hear you but you don't want to blow out their ears when you yell, that's what compression is for. Faster attack and release times, higher ratios, and lower threshold settings will all yield more compression; slower times, lower ratios, and higher threshold settings will sound more natural. Find a balance you like. Do understand that faster release times will also tend to bring up echo in the room, so if you want to control that echo, try setting it slower.

Once you're happy with the sound, you can start increasing the makeup gain to get your voice up to a level you are happy with. I find that hitting between -15 and -5 dB on average on the meters is a pretty decent place to be. You won't be blowing people's ears out, but you also won't be the quietest person on the platform. Basically, you're not going to be turning people away because you were too quiet or too loud. At this point, there's a good chance that you're clipping again, so add another filter, this time the limiter. Set the threshold to -1 dB and... well, that's it, really. Mess with the release time if you want, but the default is pretty good for this purpose. This will just stop you from ever peaking above -1 dB which allows you to set your levels to a good level without having to worry about clipping.

NOW, after all that, is the time to think about noise gates and noise suppression. Compression will inevitably make any noise louder, so even if your PC is nice and quiet, you may still get some sneaking through. Noise control should generally go before the compressor, and the less noise suppression/gating you use, the better it will sound. There are a lot of tutorials out there, just remember that less is more here. Fix your problems at the source, then use gates and suppression.

After all that, you should be all set, but if something wasn't clear, let me know! Happy to answer questions.