Start by setting each microphone, mixer, game or music source at its own controls, then balance it in OBS. Equalization is an optional way to address a specific tonal problem, not a preset that makes every voice or track sound right; check the result by listening to a normal recording as well as watching the meters.
This order matters because OBS cannot undo a noisy room, a distorted input or a badly balanced application mix. Use its faders and filters to refine a sound that is already usable, and make one change at a time so you can tell what helped.
1. Adjust the original audio source first
Open the OBS Audio Mixer and identify every active source: microphone, desktop audio, media, browser, game or other input. Say a few representative sentences, play the usual music or application sound, and note which meter moves. If a source is missing or doubled, resolve that before processing; two captures of the same desktop audio can sound like an echo or comb filtering.
Begin upstream. Adjust a microphone’s gain or a physical mixer’s input control, then check the operating system’s input level and the application’s own volume. OBS’s Audio Mixer Guide puts it plainly: “Always start at the device in question.” Device controls vary, so consult the device documentation if you are unsure which control changes input gain rather than headphone volume.
Listen at the source if it provides a headphone or monitor output. If the signal is already crackling or clipping there, lowering the OBS fader only makes the damaged sound quieter. If the source is very faint and you raise it substantially later, room noise and hiss may become more noticeable too. Correct the capture level and microphone position first, then use the OBS fader for the mix.
In OBS, check that the intended device is selected and is not also being captured through another source. Speak at the distance and volume you expect to use. A microphone tested with a quiet voice at close range may behave differently during a long devotional reading, commentary or lesson. Make your first adjustment while producing the actual kind of audio the channel will carry.
For a long recording or loop, source consistency matters as much as the opening level. If you are preparing a repeated spoken programme, the workflow in How to loop podcast audio and video in OBS for YouTube may help you think through the media source separately from live microphone capture. Keep the original file clean; filters in OBS are not a substitute for correcting a recording that is distorted before it reaches the mixer.
2. Balance sources with the OBS mixer
Once each source is healthy on its own, set the relationship between them. A listener should be able to understand speech without music or game audio masking it, while the supporting material remains audible when speech stops. There is no single fader position that works for every microphone, song, room or audience, so balance the combination by listening rather than copying somebody else’s numbers.
Start with the main content. Bring the microphone or primary programme to a steady, comfortable level, then bring in music, alerts or desktop sound beneath it. Listen to the transitions: does the music become intrusive when no one speaks, or disappear entirely as soon as the voice returns? Make small changes and pause between them. A sudden large move can obscure which source was the problem.
Keep sources separate where practical. A microphone, desktop audio and media source on separate mixer channels are easier to adjust and diagnose than one combined input. If your capture device or application already mixes several sounds together, OBS cannot independently balance the components inside that mix. Adjust them at the application or device first, or change the routing so you can control them individually.
Check mono and stereo routing. A stereo source with sound only on its left channel may be heard only on the left in the recording. Use OBS’s audio properties and monitoring to check the channel behaviour, and downmix when appropriate for the intended output. Do not assume a meter moving on one side means the sound is centred for viewers.
This is also a good moment to make sure the content is appropriate for the channel’s format. A continuous playlist or lecture stream has different changes in loudness from a live commentary session. For a playlist-based setup, the guide to streaming a YouTube playlist with FFmpeg and H.264/AAC settings covers a different delivery route; whichever route you use, judge the audio from the resulting output rather than from the source file alone.
3. Read the meters, then listen for clipping and imbalance
OBS meters are useful for showing whether a source is active and whether its level is approaching an unsafe peak. The Audio Mixer Guide describes green, yellow and red zones as broad visual guidance: speech may reach the lower end of red, while gameplay or other content may sit in yellow and background music or alert effects in green. Treat those descriptions as a starting reference, not a universal target or a measurement of how loud a viewer will perceive the sound.
Watch the meter while reproducing a normal passage, including the loudest expected syllable, musical hit or alert. A level that repeatedly reaches the top can indicate clipping risk. A signal that barely moves may be too quiet, but the meter alone does not tell you whether a voice sounds clear, whether music masks it, or how the final mix will feel on another device. OBS explicitly cautions that meter appearance does not always match perceived loudness.
Clipping can happen before OBS, inside a plugin or at the output. Listen for harsh crackle or flattened peaks, especially when a speaker laughs, raises their voice or hits a loud consonant. If the sound is distorted at the source, fix the source gain. If distortion begins only after processing, bypass the recent filter and compare. Lowering a later fader cannot restore detail that has already clipped.
Use headphones when checking subtle noise or left-right imbalance, but also consider the ordinary listening conditions of your audience. Small phone speakers can make low-level speech difficult to follow; a headphone mix that sounds pleasing is not automatically clear on a phone. Avoid compensating blindly with more gain or bass. Make a test capture and listen back at a sensible playback volume.
4. Choose a filter for a specific problem
OBS provides built-in audio filters including Gain, Expander, Compressor, Limiter, Noise Gate, Noise Suppression and Invert Polarity, as well as a VST 2.x Plugin filter. Each should answer a defined problem. Open the source’s Filters control, make a recording before and after a change, and bypass the filter if you cannot hear a useful improvement. OBS’s Filters Guide describes the available routes and controls.
A compressor reduces loud portions above a threshold, which can help tame a voice whose level varies. It can also be used on desktop audio to duck music or game sound when a microphone speaks. OBS’s documented compressor defaults are interface starting values, not recommended voice settings: ratio 10:1, threshold -18 dB, attack 6 ms, release 60 ms and output gain 0 dB. Read the OBS Compressor Filter guide for the control definitions and then listen while changing settings gradually. Too much compression can make breaths and room noise prominent or make speech feel unnaturally constant.
For ducking, place a compressor on desktop audio and select the microphone as its sidechain source. OBS gives a starting example of ratio 32:1, threshold -36 dB, attack 100 ms, release 600 ms and output gain 0 dB. These figures are OBS’s example, not a universal mix. Adjust by ear: if music drops too abruptly, barely moves, or stays low after speech ends, the threshold and timing need attention.
A limiter serves a different job. It catches peaks above a ceiling and OBS recommends placing it last in a filter chain. Its documented default threshold is -6 dB and release is 60 ms; those are defaults, not a guarantee that the input is well set. See the OBS Limiter Filter guide. A limiter can help contain a sudden peak, but it does not replace sensible gain staging or a test recording. If it is working constantly, revisit levels upstream.
Noise suppression is intended for mild background or white noise, such as a fan. OBS describes RNNoise as higher quality but more CPU-intensive, while Speex is configurable; stronger suppression can distort a voice. The documented Speex suppression default is -30 dB, but that is not a target for every room. Compare the voice with suppression on and off, especially on quiet endings and breaths. If noise is prominent, improve microphone placement or reduce the noise at its source rather than asking suppression to repair a very noisy room.
| Problem you hear | Filter to consider | What it does not solve |
|---|---|---|
| Voice level varies between quiet and loud passages | Compressor on the voice | Distortion or noise already present at the input |
| Music or game sound competes with speech | Compressor sidechain on desktop audio | A poorly balanced source that needs separate level adjustment |
| Occasional sharp peaks | Limiter at the end of the chain | Consistently excessive input level |
| Mild steady fan or room noise | Noise suppression | A noisy room or severe background sound |
| A specific tonal imbalance | Compatible VST2 equalizer | Poor microphone position or a damaged recording |
5. Equalize only to address a tonal problem
OBS does not list a native parametric equalizer in its filter guide. Its VST 2.x Plugin filter provides a route to add a compatible VST2 equalizer to a source, commonly a microphone or media source. An EQ changes the balance of frequencies; it cannot make a poor recording clean, remove all room echo or repair clipping. First check the microphone position, source gain and room, then decide whether a specific tonal quality still needs adjustment.
Name the problem before reaching for a control. Perhaps a voice sounds unusually boomy, a room resonance is distracting, or a track has a tonal imbalance that is obvious on more than one playback system. Make a small change, compare it with the filter bypassed, and keep it only if the improvement remains clear. A change that sounds attractive on one pair of headphones may make the voice less natural on phone speakers.
Avoid copying an EQ curve without knowing the source it was made for. Voices differ, microphones colour sound differently, and room reflections affect what reaches the mic. Even two speakers using the same model can need different treatment because of distance, placement and speaking style. There is no universal EQ setting that makes every voice, bhajan recording, lofi loop or game mix sound balanced.
The VST route has trade-offs. OBS documents support for VST2 through this filter, but not VST3, VST1.x, MIDI control or input, or shell VST plugins. A plugin can be incompatible, crash, or use substantial CPU. Back up your OBS profile or settings before adding one, check the plugin’s own compatibility information, and test a recording after installation. The OBS VST 2.x Plugin Filter guide describes these limitations. OBS’s examples of plugins used to develop and test support do not mean that every current release will work on every system.
If an EQ plugin adds instability or complexity without fixing a problem you can identify, remove it and keep the simpler chain. Built-in filters are easier to reason about; VST2 EQ offers more tonal control but brings compatibility and CPU considerations. The right route is the one that lets you reproduce a useful result reliably, not the one with the most controls.
6. Record and review a normal session
Use OBS monitoring to hear sources while adjusting, but do not treat monitoring as the final check. Make a recording with the usual microphone, music, desktop audio and transitions. Include a quiet passage, a normal speaking section and the loudest material you expect. Play the captured file back from the beginning and listen for balance, noise, clicks, clipping, abrupt ducking and missing channels.
A recording captures how the sources and processing combine in the output. If speech is clear while the music is present but becomes hard to understand when the game gets louder, reduce or duck the competing source. If the file is clean but too quiet, check the source and fader levels before adding output gain. If one channel is absent, revisit routing rather than trying to fix it with EQ.
Make one change at a time and record again. Keep a note of what you changed and whether the result improved on headphones and a typical speaker. This makes it easier to undo an unhelpful compressor threshold or an EQ change that sounded better only at first. A recording also helps distinguish a real problem from the expectation that a meter should sit at one particular colour.
For a 24/7 channel, consistency across a long session deserves attention: one test at the start may miss a device that changes level or an application that becomes audible later. If your audio is part of a wider always-on workflow, keeping OBS streaming after shutting down your computer is a separate operational question from getting the mix right. If managing a long-running broadcast means you need your computer switched off, StreamNeo can remove that specific burden by running an uploaded video as a YouTube live stream; it does not replace checking the audio in the file you upload.
When your source file or channel is ready, compare the operating options before choosing a workflow.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Should I use an EQ preset for my microphone?
Not without listening to the result on your microphone and in your room. EQ should address a sound you can identify, such as a distracting resonance, rather than apply a curve simply because it is labelled for speech. Compare a recording with the EQ on and off.
Why do OBS meters look safe when my audio sounds too loud?
Meters show signal level, not perceived loudness for every listener or playback device. The balance between voice, music and other sources also affects clarity. Use meters to spot activity and peaks, then listen to a normal recording.
Should I use a compressor or a limiter?
Use a compressor when loudness varies and you want to reduce that variation; it can also duck desktop sound under speech. A limiter is a peak ceiling, generally placed last in the chain. Neither corrects a clipped or badly set source.
Can OBS equalize audio without a plugin?
OBS’s listed route for EQ is its VST 2.x Plugin filter, which requires a compatible VST2 equalizer. The filter does not support VST3, and plugins may be incompatible or resource-intensive. If you do not need a specific tonal correction, leaving EQ out is a sound choice.