A reliable way to reduce audio peaks in a 24/7 waterfall stream is to test the actual programme, keep its output below digital clipping, and put an OBS limiter last in the audio filter chain. The limiter sets a ceiling; it does not establish a universal loudness target or guarantee that every part of a loop will sound clean.
Use the same video and audio you plan to broadcast when you test. Then check the live feed as well as the OBS meters: a clean-looking meter cannot tell you whether a loop seam, sudden splash, or distorted source sounds right to a listener.
Why waterfall audio can peak
Waterfall recordings often sound even at first listen, but their level can change from moment to moment. A close splash, a gust moving across a microphone, a change in the recorded scene, or a level jump where the file loops can create a brief peak above the surrounding wash. A recording can also contain distortion already, before it reaches OBS.
The key distinction is between a source that is consistently loud and one with occasional peaks. Raising the overall level to make a quiet recording more audible can push its loudest moments towards clipping. A limiter can restrain the output at its threshold, but it cannot restore detail that was clipped in the original recording or make a poor loop transition disappear.
OBS describes digital clipping in relation to the 0 dBFS ceiling: streamed or recorded audio should remain below it to avoid clipping distortion. Its mixer bar represents peak level, so watch for the meter reaching the top rather than judging only by the average appearance of the bar. See OBS’s audio mixer guidance alongside the official OBS mixer technical details.
Before adjusting filters, check the file itself. Listen to the loudest passage and to the transition between the last and first moments of the loop. If distortion is present in the source, find a cleaner recording or edit the source; applying a limiter downstream may cap the level but will not undo that damage.
Test the actual stream feed
A test is useful only if it resembles the stream you intend to run. YouTube recommends that tests include audio and video movement similar to the planned broadcast, and advises checking stream health and messages. Its encoder settings guidance is a useful reference for setup, but technical ingest settings are not a loudness instruction.
Load the actual waterfall file or playlist, use the intended audio routing and filters, and send it through the encoder to a test broadcast or other appropriate private test setup. Listen to the resulting feed, not just the file playing locally. The encoder, OBS filters, routing, and YouTube ingest all sit between the source and what a viewer hears; testing only the source skips those steps.
During the test, listen for at least one loud passage and a full loop transition. Watch the OBS meter for abrupt peaks, but also listen for crackle, harshness, unexpected silence, or a level shift. If the waterfall is intended to be gentle background sound, listen at an ordinary playback level as well as on headphones; the aim is to catch unpleasant changes, not to make a subjective loudness judgement from one device.
A useful test note can be simple: which part sounded wrong, what the meter showed, and what changed after the adjustment. Change one control at a time, then repeat the same passage. This makes it easier to distinguish the effect of source gain from a filter setting or a playback problem.
For a longer prerecorded programme, the file and playlist matter as much as encoder settings. The advice on running recorded piano theory classes as a 24/7 stream also applies to planning continuous playback: establish what repeats, where transitions occur, and what a viewer will hear after the first pass.
Put the OBS limiter last
In OBS, open the audio source’s filters and add a Limiter to the signal that will go to the stream. Place it at the end of that source’s filter chain. OBS’s Limiter Filter documentation explicitly says the limiter should be the last filter in the chain. That position allows earlier processing to happen first, while the limiter acts as the final ceiling in that chain.
A source chain might contain gain adjustment, an optional compressor, and then the limiter. Do not add filters merely to fill out a chain. If the recording is steady and already at a sensible level, the limiter may be the only dynamic filter needed. If the level varies enough to benefit from earlier control, a compressor can reduce the size of stronger peaks before the final limiter.
Check that the limiter is applied to the audio that actually reaches the stream. If the waterfall audio comes from a media source, apply and inspect that source’s filters; if it is routed through another mixer source, confirm which signal is being sent. Avoid assuming a filter on one source protects a different output path. After changing the chain, test again from the stream feed.
A limiter is not a substitute for sane input gain. If the source arrives extremely hot, the limiter may be working constantly and the result may sound flattened or unnatural. If the source has already clipped before OBS processes it, the limiter cannot repair the distortion. Reduce excessive source or input gain first, then use the limiter to control remaining output peaks.
Understand the documented defaults
OBS documents the Limiter filter with a threshold of -6.00 dB and a release of 60 ms as its defaults. Treat these as documented starting settings in the software, not as a YouTube requirement, a recommended loudness for every waterfall recording, or a target that every ambience stream should use. The threshold is the point above which the limiter acts; the release setting describes how quickly it stops reducing the signal after it falls back below that point.
A threshold is a ceiling control, not a measurement of perceived loudness. A waterfall loop with a modest dynamic range and a recording with sharp, isolated splashes may behave differently under the same setting. The documented default can help you begin a test, but listen to the result and examine the meter before deciding whether the output suits your material.
Do not confuse the limiter threshold with the 0 dBFS clipping boundary. One is a filter control value; the other is the digital ceiling OBS identifies as the point to stay below to avoid clipping. Nor should you interpret a threshold value as a target for the average level. The sources do not prescribe a universal integrated loudness or true-peak target for this kind of stream.
YouTube’s encoder settings documentation lists technical audio options for RTMP/RTMPS, including AAC or MP3, and recommended advanced values of 44.1 kHz for stereo and 128 kbps for stereo. Those are encoding settings, not instructions for how loud the ambience should be. Keep the distinctions clear: choose compatible ingest settings, control peaks in the audio chain, and decide by testing and listening whether the resulting programme is comfortable.
Choose a continuity setup that you can monitor
Audio control and stream continuity are related operational concerns, but they are different problems. OBS can process the audio in a local broadcast workflow; it does not by itself decide whether your computer, connection, and programme will keep running through the night. For a 24/7 prerecorded stream, choose a continuity approach that fits your ability to keep equipment running and respond when something stops.
A local encoder gives you direct access to OBS and its filters, but your computer and connection need to remain available. A cloud-based service intended for continuous prerecorded video can remove the need to leave your own computer running, but you still need to test the submitted file, understand how you will monitor the broadcast, and check the service’s current capabilities and cost before choosing it. YouTube’s encoder setup guidance discusses connecting an encoder and stream key; treat the key as a credential, not as something to share casually.
| Consideration | Local OBS encoder | Cloud-based prerecorded stream |
|---|---|---|
| Your computer | Must stay available for the stream | Does not need to stay on for the broadcast to run |
| Audio controls | Direct access to OBS filters and routing | Confirm which controls are available before relying on a particular adjustment |
| Monitoring | Watch OBS and the YouTube live feed | Arrange a way to check the YouTube feed and service status |
| Main trade-off | More direct control, with local equipment and connection to maintain | Less dependence on your computer, with service availability and cost to assess |
The right choice depends on what you can operate reliably, not on a promise that one method never fails. If using a local workflow, rehearse a restart and reconnection rather than assuming the encoder will run unattended indefinitely. If using a cloud workflow, test the actual uploaded loop and find out how you will know if the feed stops or its audio changes.
For a local audio station feeding YouTube, the AzuraCast and FFmpeg setup guide covers a different route to continuous audio. If your chosen encoder loop instead uses FFmpeg, keep the reconnect behaviour in a continuous stream in mind when planning continuity; reconnect logic does not replace checking the actual audio.
Monitor audio and video while live
A successful test is a starting point, not a reason to stop checking. YouTube advises operators to monitor audio and video quality continuously while streaming. Keep a practical way to inspect the live feed available, particularly after launch, after any change to the programme or settings, and when the stream has been running long enough to reach another loop seam.
Listen to the viewer-facing stream where possible. The OBS preview and mixer show useful information, but the public feed is the result that matters. Check for a sudden change in level, distortion, silence, a stalled picture, or a mismatch between the sound and image. If the live sound becomes harsh, compare it with the source and the meter before changing the limiter threshold; the cause might be a source change, routing, or an earlier filter.
Keep a short record of the changes you make. Note the source, filter order, gain adjustments, and the passage you used to test. If the stream begins to sound different later, that record gives you a known working setup to compare with. It also helps avoid making several changes at once while the broadcast is live.
Do not assume continuous delivery means a complete archive will be available. YouTube’s encoder setup documentation says streams under 12 hours are automatically archived. An uninterrupted 24-hour stream is outside that condition, so plan separately if you need a complete recording; check YouTube’s current guidance rather than relying on an automatic full-day archive.
Check for distortion before going live
Before starting the long broadcast, review both the source and the encoded feed. Confirm that the waterfall file is the intended version, the audio is not clipped before it reaches OBS, and the meter stays below the 0 dBFS ceiling. Play the loudest passage and the loop boundary through the complete signal chain. A limiter may keep the output from crossing its threshold, but it does not make every source sound clean.
If you hear distortion, isolate the point where it appears. Listen to the source file alone, then the OBS output, then the live test feed. If the source is already distorted, replace or repair it at the source rather than adding more processing. If the meter reaches the top after gain is added, reduce that gain and test again. If the limiter is visibly or audibly affecting much of the programme, revisit input levels and the source rather than treating heavy limiting as the intended sound.
A compressor is relevant only when the signal varies in a way that benefits from earlier control. OBS describes a compressor as reducing the likelihood of large peaks and recommends placing it near the beginning of the filter chain; the limiter remains last as the maximum-output control. See the OBS Compressor Filter documentation for the filter’s role. A steady ambience recording may not need a compressor at all.
Finally, keep the programme rights and loop design in the same preflight as audio. Confirm you are entitled to use the footage and recording, check that the repeat is intentional, and inspect transitions for gaps or abrupt level changes. Make a short test after the final file or filter change rather than relying on an earlier test of a different version.
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
Does the OBS limiter prevent all audio peaking?
It caps output according to its threshold when placed correctly in the signal chain, but it does not guarantee peak-free or distortion-free audio. The source may already be clipped, another routing path may bypass the filter, or the result may sound poor even when the meter stays under the ceiling. Test and monitor the actual feed.
Should I use the documented -6.00 dB threshold for every waterfall stream?
No. OBS documents -6.00 dB as the limiter’s default threshold, not as a universal loudness target or YouTube requirement. Begin with the documented setting if useful, then judge the real programme by its meter behaviour and listening test.
Where does the limiter go if I also use a compressor?
The compressor, if needed, belongs earlier in the chain to manage variable peaks; put the limiter last. You do not need both automatically: a steady ambience recording may need no compressor, while the limiter provides a final ceiling.
Can I rely on YouTube to archive an uninterrupted 24-hour stream?
Do not assume that it will be archived in full. YouTube documents automatic archiving for streams under 12 hours, so plan another recording or archive approach if you need the complete 24-hour programme and check the current official guidance.