OBS can control peaks and make individual lesson sources easier to balance, but its built-in compressor does not automatically match the loudness of separate clips. For a steadier study stream, compare each lesson source by ear, adjust its level, use compression only where the source needs it, and put a limiter last to catch excessive peaks.
The reliable check is a recording, not a row of meter readings. Test representative lessons and transitions, then listen back at a normal playback level before you leave the channel running.
What “normalised across lessons” really means
When people ask how to normalise lesson audio in OBS, they usually mean that a spoken explanation in one recording should not sound much quieter or louder than the next. That is a comparison between separate sources or clips. It is different from reducing the difference between quiet and loud moments within one lesson.
A compressor works on the signal passing through it: when the level exceeds its threshold, it reduces that part of the signal. This can control a lecturer who sometimes speaks sharply or a recording with occasional loud passages. It does not listen to lesson A, calculate its average loudness, and make lesson B match it. OBS documents compression as dynamic control, not an automatic cross-clip loudness matching function.
That distinction matters in a looped study channel. A lesson may have been recorded on a different microphone, at another distance, or with different room noise. Applying the same compressor settings to every source can leave one clip quiet and another prominent. The settings may behave similarly, but the audio going into them is not necessarily similar.
It helps to separate three jobs. Per-source gain or a mixer fader brings a consistently quiet or loud lesson into a sensible range. A compressor reduces excessive variation within a source. A limiter catches peaks that pass through the earlier processing. Neither the compressor nor the limiter raises quiet speech in the way a gain adjustment does.
If you are building a channel around recorded teaching material, the planning choices are also covered in this guide to streaming a recorded algebra course from a Mac. The playback platform and source setup differ from one channel to another, but the distinction between preparing the recordings and managing their levels remains useful.
Compare each lesson source before adding filters
Start by identifying what OBS is actually receiving. Lessons might be separate media sources in a scene, browser audio, desktop audio, or one already-mixed signal. Look at the Audio Mixer while playing each item and confirm that the signal appears on the channel you expect. If several lessons enter through one desktop-audio channel, OBS may not expose a separate fader for each lesson; you may need to adjust the playback application, separate the sources in your scene, or prepare the files before they are mixed.
Choose a representative speech passage from each lesson. Avoid comparing an introduction with silence against another lesson's loudest sentence. Use similar material where possible: a teacher speaking at an ordinary pace, with a comparable amount of background sound. Play the passages one after another through OBS and listen through the same headphones or speakers, at an unchanged listening volume.
Adjust one point at a time. First check the playback source or application level. Then check the operating-system level if it controls that source, and finally use the OBS source gain or mixer fader. This order makes it easier to see which control changed the result. OBS's Audio Mixer Guide recommends beginning at the device and checking the signal both there and as it reaches OBS.
Aim for similar perceived teaching level, not identical meter peaks. A low-pitched voice and a bright voice can register differently while seeming similarly clear to a listener. Likewise, a short consonant or a hand clap can create a high peak without making the lesson sound loud overall. Listen to the words, not just the meter.
| Control or check | What it can do | What it cannot do |
|---|---|---|
| Source gain or fader | Raise or lower one source relative to others | Reduce the source's internal loudness variation by itself |
| Compressor | Turn down signal above a threshold to control dynamics | Automatically make separate lesson clips equally loud |
| Limiter | Catch peaks above a chosen ceiling | Raise quiet sections or balance lessons |
| Recording and playback | Reveal the audible result, including transitions | Replace careful source routing or level adjustment |
OBS describes 0 dBFS as the digital maximum; a signal pushed beyond it can clip and distort. Its technical mixer notes discuss meter alignment and headroom, but those figures are guidance for interpreting OBS's meter, not a universal YouTube loudness target. Do not chase a particular peak reading as proof that two lessons are equally loud.
If your channel has a fixed sequence, test each source and each hand-off in that sequence. For a broader workflow on keeping a playlist in order, see how to schedule a YouTube livestream playlist to repeat. A change of lesson is precisely where a small source-level mismatch is easiest for a viewer to notice.
Use compression for variation, not matching
Add a Compressor filter to a source only if listening reveals an issue that compression can address: for example, the teacher is sometimes hard to hear and sometimes unexpectedly loud, or a few emphatic phrases jump out of an otherwise even recording. If the whole lesson is consistently quieter than the others, first adjust gain. Compression is not a substitute for that comparison.
OBS's compressor controls include threshold, ratio, attack, release, and output gain. The threshold is where compression begins; lowering it means more of the signal is treated. The ratio controls how strongly levels above the threshold are reduced. Attack and release affect how quickly the compressor responds and lets go. A fast response may control a sudden peak but can change the character of speech; a slower response may leave some transients intact. Listen while adjusting rather than assuming a preset fits every recording.
The OBS documentation displays defaults of -18 dB threshold, 10:1 ratio, 6 ms attack, 60 ms release, and 0 dB output gain. These are the filter's displayed defaults, not settings prescribed for every voice or lesson. Begin with a modest change from the source's untreated sound, then replay both a quiet passage and a loud one. If quieter words become less intelligible or the sound starts to pump, back off and reassess.
A compressor's output gain raises the signal after compression. That can make the result sound louder, but it also uses peak headroom and may send more level towards the final limiter. Avoid stacking strong compression and substantial output gain without checking the whole chain. You may simply move the problem from the lesson's dynamics to the output's peaks.
In a filter chain, OBS says the compressor usually belongs near the beginning. You can put a gain adjustment before or after it depending on what you are trying to achieve, but test the resulting level rather than treating order as a shortcut to matching. The OBS Compressor Filter documentation explains the controls and their purpose.
Noise suppression and a noise gate solve different problems. They can help with background noise or unwanted sound when configured appropriately, but they do not make two lessons share a perceived loudness. A gate that closes on a quiet teacher's words can make the lesson less usable, so treat noise processing as a separate decision.
Put a limiter last to catch peaks
A limiter is a final peak guard, not a normaliser for quiet material. Place it at the end of the relevant source's filter chain, after the processing that might raise or reshape the level. OBS specifically recommends that the limiter be last. If a loud syllable or another transient would otherwise exceed the chosen ceiling, the limiter can constrain it; it will not bring a quiet explanation up to the level of another lesson.
OBS lists a -6 dB threshold and 60 ms release as displayed defaults. They are examples from the filter, not mandatory targets for a YouTube study stream. A lower threshold caps the signal at a lower level, but if it engages constantly, it may flatten the source rather than merely catch occasional peaks. Watch and listen to find out whether it is doing occasional protection or continuously changing the sound.
Keep the chain understandable: set the source level, use compression where its internal variation calls for it, and use the limiter last to catch remaining peaks. Then play the loudest likely passage. If the limiter seems to be working hard throughout the passage, return to the earlier gain and compressor choices instead of relying on the limiter to repair them.
Watch the mixer while you test
The OBS mixer is useful for confirming that the expected source is active and that the signal is not reaching the digital ceiling. It is not a full measurement of how loud a whole lesson will seem. OBS's meter is peak-oriented, and the technical documentation discusses a VU indicator that integrates RMS over 300 ms. That indicator can give a different impression from the peak display, but it still is not the same as a proper integrated loudness measurement.
Use the meter as a warning and comparison aid, not a loudness verdict. If one lesson regularly approaches clipping while another sits much lower, investigate the source levels. If peaks look similar but one voice sounds much thinner or quieter, trust your listening comparison and adjust the relevant source. A peak is a brief maximum; perceived loudness depends on more than that instant.
During testing, keep an eye on the channel that carries the final mix as well as the individual sources. This catches situations where each source seems acceptable alone but overlapping audio or a transition creates an excessive output. If you are also troubleshooting a channel that must stay running on a particular connection, the practical considerations in streaming a YouTube live loop with JioFiber are separate from audio balancing, but they reinforce the value of testing the complete path you intend to use.
Record a representative session and listen back
A live meter cannot tell you how a viewer will experience an entire sequence. Record a representative test in OBS, including ordinary speech, a louder passage, a quieter passage, and at least one transition between lessons. If music, an opening slate, or other audio is part of the channel, include those elements too. The test should resemble the material you expect to loop, rather than a convenient fragment chosen because it sounds easy.
Listen to the recording from start to finish at a normal listening level. Do not keep changing your headphones or device volume between lessons; that hides mismatches. Ask practical questions: can you understand the quietest speech without straining? Does the next lesson start abruptly louder? Are peaks controlled without making the speaker sound unnaturally constrained? Does background noise become distracting when a lesson is quieter?
If you hear a mismatch, identify which source or transition causes it and change one control. Record the affected section again and compare it with the earlier result. Keep notes on the source gain and filters for each lesson, especially if you revisit the project later. A small, documented adjustment is easier to maintain than a collection of copied filters whose purpose is unclear.
Do this before the channel goes live, and repeat it after replacing a file, changing a source route, or altering filters. OBS's Audio Mixer Guide recommends recording a normal session and listening back before going live. If you plan to let a study stream run overnight, this check matters because you may not be present to correct a loud hand-off when it happens.
Do not rely on YouTube to fix every mismatch
YouTube describes playback enhancements that can include Stable volume, which balances quieter and louder parts for playback. Its help page also says enhancements are not applied to every video. That is not a reason to leave obvious source differences in your programme mix, nor a guarantee that each viewer will hear the same processing.
A viewer's device, listening environment, and playback choices also affect what they hear. A phone speaker in a noisy room is not a useful substitute for checking whether your own recording has an abrupt jump. Make the stream reasonably consistent at its source, while accepting that playback cannot be controlled completely once a viewer chooses a device and listening level.
YouTube's video and audio quality enhancements help page is the right place to check its current description of playback features. Do not treat any enhancement as a universal repair or as an official loudness target. If your workflow has no person present to restart a stream after a disruption, a cloud-run broadcast can remove the need to keep your own computer on; StreamNeo is useful for that specific operating problem, while the audio still needs to be balanced and checked before the file is used.
For a channel that depends on a personal computer staying on, the overnight operating burden is another decision apart from audio filters. A guide to keeping a broadcaster running overnight on a Windows PC can help you think through that trade-off. Whichever method you use to keep the stream running, prepare and test the sound before leaving it unattended.
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
Will an OBS compressor normalise each lesson automatically?
No. The compressor reduces signal above its threshold within the source it processes; it does not compare the average loudness of separate lesson clips and match them. Balance each source with gain or a fader, then use compression only where a lesson's own loudness varies too much.
How do I make lesson audio the same volume in OBS?
Play comparable speech passages from each lesson through OBS and listen at an unchanged playback level. Adjust each source's gain or fader until the teaching level is reasonably consistent, then check the loudest and quietest passages in a recording. Similar meter peaks alone do not prove that the lessons will sound equally loud.
Does YouTube automatically balance the volume of every video?
YouTube describes Stable volume as a playback enhancement, but says enhancements are not applied to every video. Do not rely on it to correct your stream's source mix for every viewer. Balance and test the programme before it reaches YouTube.
Should I put a limiter before or after the compressor?
OBS recommends placing the limiter last in the filter chain, so it can catch peaks after earlier processing. A limiter prevents excessive peaks; it does not raise quiet sections or match one lesson to another. Check the complete chain by recording and listening back.