Audio basics for streamers part 3: levels, filters, and effects

Last time we wrapped up our look at the hardware components; in this post, we’re going to wrap up by looking at what you can do in OBS (and probably other software) to enhance that signal with some finishing touches. First things first, let’s talk about levels, which means that we actually need to step back into the hardware realm for a bit.

Input levels

Remember when I said that preamps are used to boost a mic level signal to line level? Well, there’s a reason why there’s a knob to control how much you boost it. If you just crank it all the way, you’ll be overloading the preamp and distorting the signal. Similarly, if you don’t boost it enough, then it’s going to be hard to hear and your signal-to-noise ratio (the difference between your input signal and the noise floor of the preamp) will not be very good. What’s the sweet spot? Well, it’s pretty simple to find, really. Turn up the gain on your preamp while speaking the loudest you might possibly speak/yell/whatever. Many interfaces have a clipping indicator on them; typically, red indicates that your signal is too loud. Once you hit the red, turn the gain back down until you stop hitting the red. Easy peasey, right?

The only catch is that you might find that your normal speaking voice is significantly quieter than your loudest possible sound and that, actually, you almost never get that loud in the first place. In that case, it might actually be better to turn the gain up a bit so that your normal speaking is more audible. Yes, if you happen to get super loud you might distort a bit, but in the grand scheme of things, that’s not that big of a deal. It’s better to have a good signal level with some clipping than no clipping and low signal level.

Filters and effects

Here’s where the fun really begins. OBS comes with several useful plugins, but I would also suggest downloading the excellent ReaPlugs from Cockos: https://www.reaper.fm/reaplugs/. These come with their recording software Reaper, but are also available for free for use with other software. You don’t need all of them; when you install them, I recommend ReaComp, ReaXcomp, ReaEQ, ReaFIR, and ReaGate. I’ll talk about what each of these do in a minute, but for now, it’s enough to know that the plugins serve a couple of purposes: reducing dynamic range, expanding dynamic range, noise reduction, and equalization. In OBS, you can find the filters by clicking the gear/cog icon next to the audio source you want to apply filters to in the mixer and choosing “Filters” from the menu. From there, use the + button to add a filter. ReaPlugs are added by choosing the VST option from that menu.

Noise reduction/suppression

Noise reduction should always be the first thing in your signal chain. OBS comes with both an expander and a noise gate, though the gate is technically just an expander with a high ratio. ReaGate does the same thing, but offers a few more options. Essentially, you specify a signal level that you want to be the minimum level and anything below that will be reduced (with an expander) or cut completely (with a gate). This also means that if you don’t talk above that threshold, you won’t be heard, either. For that reason, it’s best to cut noise at the source (turn off your A/C, close the windows, etc), but the noise gate is still a useful tool to get rid of the last bit of noise.

With an expander, you can make the cutoff a little more subtle since the audio is only reduced instead of cut. The ratio determines how much the sound is reduced. With ReaGate, you can also blend in some of the dry sound (that is, unaffected sound) and use the lowpass and highpass filters to control what frequencies actually trigger the gate.

With expanders and gates, I find that the default settings for the gate in OBS work quite well. The only thing you really need to adjust is the threshold as the cutoff point is going to depend on your gear and your voice.

OBS also comes with a noise suppression plugin, but I do not recommend using it. It’s designed to monitor the incoming audio for noise and filter out just that noise. In theory, this sounds much better than a noise gate; in practice, it also has a big impact on the sound of your voice and usually causes a noticeable degradation of quality. This is where ReaFIR comes in. It’s a pretty complex plugin that can do a lot of different things, but for our purposes, it’s an extremely effective and transparent noise suppressor. Change the mode to “Subtract,” then click the “Automatically build noise profile” checkbox. You want to be completely silent at this point so that it is only detecting the noise you want to filter out. You’ll see a red line come appear and, at some point, settle in. Once it stops moving (probably after a few seconds), uncheck the box. That’s it! This isn’t foolproof and is really only effective for things like A/C noise, fan noise, or other rumbles and white noises, but I find that I can use just this without any gating and get very clean audio.

Reducing dynamic range (compression)

Compression is usually the next plugin I will add (though you could put EQ first, the difference is subjective). In OBS, there are two plugins that accomplish this: “Compressor” and “Limiter.” ReaPlugs provides several types with the ReaComp, ReaXcomp, and ReaFIR plugins, though we will not be using ReaFIR for that. Compressors serve the opposite purpose of an expander: they reduce the volume of anything over a certain threshold you set. The amount that the signal is reduced is controlled by the ratio, just like with the expander. Typically, a 3:1 ratio is pretty good for speaking. The general idea here is to figure out the volume of your quietest speaking voice and set the threshold just above that. This way, your louder speaking will be reduced in volume to better match your quieter speaking volume. As you might imagine, this reduces the overall volume of the signal, so you can raise it back up using the output gain slider.

Unfortunately, there are no meters within the plugin, so you have to look at the meters on the mixer to determine your approximate levels. This is where ReaComp and ReaXcomp come in. You can easily see your volume levels and set the threshold right on the meter. You may want to experiment to find the exact settings that work for you, but I find a relative quick attack (2-4 ms) and release (10-20 ms) work pretty well. You may also want a bit softer knee for a more subtle effect. I’ve found that a 3 dB knee is pretty good. Tweak the threshold until your levels are pretty consistent, then adjust your output levels to compensate for the decrease in volume. You can try checking the “Auto make-up” box, but, if that doesn’t work, simply increase the “Wet” slider until you get a good output level without clipping. This article provides some good additional reading if you want a little more detail on the settings, but keep in mind that all of these recommendations are just starting points.

If ReaXcomp confuses you, don’t worry about it too much. It is a multiband compressor, meaning that it can compress different ranges of frequencies differently. It’s a more advanced version of compression that I don’t recommend messing with for now, but you might want to play around with it later and might even end up liking it.

Limiters, much like gates, are just extreme compressors. Any compressor with a ratio of 10:1 or more is considered to be a limiter. They are useful for setting an absolute maximum signal level to prevent clipping. You might, for example, put a limiter as the last thing in your signal chain with a threshold of -1 dB just to make sure you can’t clip the output.

Equalization

This is probably the filter most familiar to you, at least in theory. It allows you to raise or lower the bass, mids, and treble of your signal. It’s extremely useful, but, unfortunately, not included with OBS. Instead, you will need to use ReaEQ. This is going to look a little bit more complicated than what’s on your radio, but here’s the gist of it. Each of the tabs is a point that you can adjust. You can have as many or as few as you want. By default, the first one is a low shelf and the last one is a high shelf. The other bands in between will be set to band. A shelf filter basically takes everything below or above a frequency point and increases or lowers it, hence the shelf moniker. A band filter, meanwhile, adjusts a range of frequencies up or down. The range is determined by the bandwidth or Q. You can make it control a very wide range of frequencies or a very small range. In general, wider ranges will sound more natural and narrow ranges will sound more artificial. However, narrow ranges can be useful for filtering out very specific problem frequencies.

In general, you don’t want to make huge changes here. You can boost a bit around the 100 Hz range to increase the low end of your voice or boost with a high shelf around 4.5 kHz and up add more presence/crispness. You might also reduce frequencies around the 200 Hz range or a little above to reduce muddiness. The exact settings are going to depend on your voice, your mic, and your recording environment. For me, a little cut around 3 kHz (to make me sound less nasal) and a high shelf boost around 4.5 kHz works quite well, but you’ll need to find the settings that work for your mic and voice.

Output levels

Now that your voice is sounding good, it’s time to look at the meters and make sure you aren’t blowing out your viewer’s eardrums. If you look at the mixer, you’ll see that 0 dB is at the right (or top in vertical layout). From there, there’s a red zone, a yellow zone, and a green zone. The colors are a little misleading, honestly, but the general idea is to stay out of the red zone so that your overall level doesn’t distort. Maybe one day they will add an overall output meter to give us a better idea what the output really looks like. Until then, here’s how I set it up.

Leave the fader for your voice track at the maximum position. I also recommend going to the advanced audio properties and clicking the “Downmix to mono” for that track, especially if you are using an audio interface, just to make sure your voice is coming out of both speakers. That’s not strictly necessary, but if you see two meters for the track and only one is moving, you’ll definitely want to downmix. While you are adjusting your filters, you want the final signal level to live in the middle to upper range of the yellow portion of the meter. It’s okay if it hits the red on occasion, but you never want it to hit the 0 dB line. You’ll now have a good maximum volume for your voice track. Since that should be the loudest thing in your stream anyway, you can just leave that slider alone.

From there, adjust your other volumes down until you find a sweet spot where they are still audible but not so loud that they drown out your voice. The OBS wiki mentions that you’ll want these non-voice tracks to be somewhere in the green range of the meter, but you’ll probably need to do a few test recordings to find the sweet spot.

One last note on the mixer. In part 2, I said we’d talk about why you need to think of your inputs as working in pairs and only put mics into every other input. The reason is that OBS does not support ASIO, instead relying on standard Windows audio for inputs and outputs. ASIO allows every input to be a unique input, but Windows, for whatever reason, reads them in stereo pairs. This means that, for a 4-input interface, OBS will see it as two stereo inputs where input 1 is left, input 2 right, input 3 left, and input 4 right. If you want to put different effects on the inputs, this effectively halves the number of inputs you can use. I haven’t tried it, but there is a plugin that adds ASIO support: https://github.com/pkviet/obs-asio. From what I can tell, it looks like this will allow each input to be used separately, so if that’s something you need, give that a try.

Final final thoughts

And that’s really all there is too it! I know this was a lot of information and I probably still missed some things, but I hope this clears up most of the mystery. I didn’t go super in-depth with compressors and gates, but honestly, the default OBS settings (aside from the threshold) are pretty good, so you don’t really need to mess with them too much. If you want to understand them better, there are a lot of good explanations on the web that should help with that. But if you have any questions, let me know!

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.