If you are watching a playlist, try YouTube’s Stable volume setting; it may reduce changes between quiet and loud parts, but it is not available on every video. If you are building a continuous YouTube Live programme from your own clips, measure and adjust those clips before they reach the stream, then listen across every join.
These are different jobs. Stable volume is a viewer-side playback feature, while loudness normalisation is a preparation or live-mixing step under the creator’s control. Neither one guarantees identical sound across every video or playback setup.
Decide who is controlling the audio
First establish whether you are trying to improve what you hear from someone else’s playlist or what your channel sends to YouTube. The controls and limits are different, and treating them as interchangeable leads to wasted effort.
As a viewer, you can check the player’s Settings menu for Stable volume. YouTube describes it as a playback feature that is on by default for supported videos and reduces variation between quiet and loud sections. The setting may not appear for a particular video, and the official guidance does not promise identical behaviour in active live playback, a replay, and an ordinary upload. See YouTube’s explanation of Stable volume for the current description.
As a creator, you control the source clips, or at least the mix entering your encoder. If a bhajan video is quiet and a spoken announcement immediately after it is much louder, a viewer-side player setting cannot be your production plan. Prepare the audio consistently before streaming, or use an appropriate processing chain while mixing live.
A source-preparation workflow also has a clear boundary: it can only affect files you process. If you add a clip later, replace a track, or insert an unmeasured sponsor announcement, that material can still sound out of place. Keep a record of which versions were checked.
For a channel assembled from prerecorded pieces, it is useful to think of the programme as a sequence rather than a folder of unrelated exports. Planning clip order and transitions matters alongside audio; the guide to switching podcast episodes automatically on a YouTube Live stream covers the scheduling side of that sequence.
Try Stable volume in the YouTube player
If you are listening rather than producing the stream, begin with the simplest reversible test. Open a playlist item, open the player’s Settings menu and see whether Stable volume is offered. Listen to the same passages with the setting on and off if the option is available; that helps you judge whether it suits the material and your listening setup.
Stable volume is intended to reduce variation during playback. It does not edit the uploaded file, change the source used by the channel, or give a creator a normalised master for later use. If a track still sounds much quieter than the next item, you can adjust your device or player volume for that listening session, but that changes your local experience rather than the programme sent by the creator.
For example, someone listening to a long playlist of devotional recordings may encounter a quiet older recording followed by a modern, densely mastered track. Stable volume is worth checking, but the listener should not assume that it will make those two recordings identical. Recording dynamics, tonal balance and the feature’s availability all affect the result.
The same distinction applies when the playlist contains live-stream replays. A replay is a video in the player, but YouTube’s help material does not establish that Stable volume behaves identically for every replay and active live stream. Treat the player control as an available aid to test, not a promise about all sessions.
Understand where the player feature may not help
A control that is present for one video may be absent for another. YouTube’s help page explicitly notes that Stable volume is not available for all videos. That alone prevents it from being a dependable playlist-wide fix: one unsupported item can interrupt an otherwise helpful listening experience.
It also does not provide the creator with a way to repair the outgoing programme. If viewers report that a particular section is too loud, turning on a playback setting may help some listeners but does not correct the mix for everyone. Review the source section and the stream chain instead.
There is an evidence limit around context, too. YouTube describes Stable volume generally for videos, but the reviewed official material does not settle whether its behaviour is the same for an active live broadcast, a replay of that broadcast, and a regular upload. Avoid basing a channel workflow on an assumption that it handles all three in the same way.
If the symptom is silence rather than uneven loudness, investigate that separately. A useful companion is the checklist on fixing silence in a 24/7 YouTube radio livestream. Silence, missing audio and a track that is merely quieter are different faults; loudness matching will not repair a broken audio route.
Measure perceived loudness across source clips
When you control the files, do not compare only their highest sample peaks. A brief drum hit can reach the same peak as a sustained voice or devotional chorus while the overall programme sounds much quieter. Peak level describes a maximum; perceived programme loudness needs a loudness measurement across time.
Use the same meter and measurement method for each complete clip. Compare integrated loudness to understand the overall level over the programme, true peak to see how high reconstructed peaks may go, and loudness range when the difference between quieter and louder passages matters. The EBU R 128 recommendation describes programme loudness together with Loudness Range and Maximum True Peak Level rather than treating one peak reading as a full match.
A practical first pass is to measure each clip, write down its results, and sort out the outliers. Include intros, spoken links, endings and any long quiet sections that will actually be part of the live sequence. If you measure only the loudest chorus or a short sample, you may miss the level shift that listeners encounter when the next full clip begins.
Do not treat a particular LUFS value as a universal YouTube requirement. EBU R 128 version 5.0, published in 2023, recommends average programme loudness of −23 LUFS for its broadcast context. EBU guidance also gives a ±1 LU tolerance for live programmes. Those are EBU recommendations, not evidence of a required target for every YouTube channel, genre or audience. The EBU’s loudness guidance provides the standard context; choose a target for your production deliberately rather than labelling the broadcast recommendation as a YouTube rule.
For a small channel, consistency in method is often more useful than chasing a number picked from an unrelated broadcast workflow. A study channel, a local news loop and a music-led ambience stream have different material and listening expectations. Set a sensible working target for your own programme, apply the same measurement process across its clips, and check the result by ear.
Keep a simple clip log with the filename, the measured integrated loudness, true peak, any adjustment made and the date or version identifier you used. This avoids reprocessing a file by mistake and helps when a revised export is substituted. It also makes it easier to find one unusually quiet or loud section without guessing from memory.
Use FFmpeg loudnorm for preparation
FFmpeg’s loudnorm filter is a documented option for EBU R 128 loudness normalisation. Its documentation describes dynamic and linear modes, and supports single-pass processing for live streams and files as well as double-pass processing for files. Read the FFmpeg filter documentation before choosing parameters, because available options and their meaning belong to the tool’s documentation rather than to a generic recipe.
For a library of prerecorded clips, process files before you place them in the rotation. A two-pass workflow can first analyse a file and then use the measured values during processing. This gives the filter more information about that file than a simple adjustment based on a momentary peak. Keep the original untouched, export a processed copy, and compare the copy against the source before adding it to the playlist.
Single-pass processing is useful where a file or stream must be handled without a prior analysis pass, but do not assume it yields the same result as a file-specific two-pass process. The choice is a workflow trade-off: two passes suit material available in advance, while one pass can fit a live or on-the-fly chain. The filter itself does not decide what target is musically or editorially appropriate for your channel.
A file workflow might be: inventory the clips, measure them consistently, choose a target based on your programme, process copies with loudnorm, and measure or audition the exports. If you use a graphical editor or another meter, the same principles apply. FFmpeg is not mandatory; it is one documented way to perform the processing.
Avoid blindly copying a command line from an unrelated source. Input codec, channel layout, sample rate, target choices and filter options can affect whether a command is suitable for your material. Consult FFmpeg’s current documentation and test a short copy first. Keep a known-good export and compare it on the same headphones or speakers as the unprocessed version.
Normalising files before they enter the sequence also helps when a programme combines formats: a recorded talk, an animated sponsor slide with narration, and several music clips. Each element can be checked at the source rather than relying on a late correction after the full broadcast has started. If you are arranging prerecorded event material, the article on showing sponsor slides between sessions on YouTube Live is relevant to sequence design, though audio still needs its own review.
Control peaks and listen across joins
An integrated loudness match is not the same as an identical listening experience. One clip may be compressed and dense, another may have a wide dynamic range, and a third may place speech over music. Matching their long-term loudness can still leave the voice less intelligible or make a sudden chorus entrance feel startling.
Check true peak as well as integrated loudness. True-peak control helps avoid peaks rising too high after processing and conversion, but a peak ceiling on its own does not make quiet material sound as loud as dense material. Use the measurement as a guardrail, then listen to the rendered file rather than assuming a meter has settled every question.
Audition the actual transitions in sequence. Listen to the end of one clip, the join or gap, and the start of the next at the intended playback level. A short silence may be intentional for a station ident or prayer; the same gap may sound like an interruption in a continuous ambience stream. The edit should reflect the programme, not a general rule that all gaps must disappear.
Make a test sequence with the combinations most likely to expose a problem: speech into music, a quiet recording into a dense one, and a clip with a strong opening transient. Listen through the same speakers or headphones your production uses, and also check an ordinary device if practical. A low-frequency-heavy mix can feel different on small speakers, and speech that seems clear on headphones may be masked elsewhere.
If you change gain or compression to solve a join, recheck the neighbouring clip and the true peak afterwards. Processing one item in isolation can simply move the mismatch to the next boundary. Keep the listening notes specific: “announcement is hard to hear over the bed” is actionable; “audio feels wrong” is not.
For a 24/7 station, decide how you will check replacements as well as the original library. A new track, re-export or revised spoken message should go through the same meter and audition routine. The Sikh kirtan radio station setup guide covers the broader continuous-channel context; the audio level of each source still needs to be managed separately.
Separate audio preparation from delivery checks
Once the source programme sounds consistent, check that the stream itself is configured correctly. YouTube’s Live API documentation covers stream configuration and health details such as audio codec, sample rate, channel count and bitrate-related issues. Those checks help establish that audio is being delivered in a valid way; they do not normalise the loudness of one clip against another.
Keep these checks in separate stages. First prepare and audition the content. Then confirm that your encoder or streaming workflow is sending the expected audio configuration and that YouTube is receiving it. If the signal drops out, clips, or arrive on the wrong channel, investigate delivery. If every clip is present but one is much louder, return to the source mix and normalisation work.
This separation is useful when troubleshooting reports from viewers. Ask which video or time in the programme they heard, whether the issue is loudness, distortion or missing sound, and whether it occurs on the live stream or replay. Those details help distinguish a source mismatch from a transport fault or a playback-specific experience without claiming that one tool can solve all three.
Make the workflow repeatable
For a small channel, a repeatable checklist is more reliable than relying on memory after a late-night change. Before a clip enters the rotation, confirm that it is the intended export, measure it using the established method, apply the chosen processing, check peaks, and audition its start and finish against its neighbours. Record any exception, such as a deliberately quiet meditation passage.
Keep the target and measurement method consistent within a programme, but revisit them if the channel changes character. A loop of soft ambience may call for a different listening balance from a news programme with spoken links. The available sources do not define one target that suits every YouTube Live playlist, so use your editorial judgement and listeners’ feedback rather than presenting an external broadcast standard as a platform mandate.
If you need to correct material in real time, process the outgoing chain and monitor it during setup. That can be appropriate for a live mixer or a programme whose sources are not available as files in advance. It also means the result depends on the live chain and its settings, so verify that the processing remains active and does not introduce pumping, distortion or unwanted changes to dynamics.
For prerecorded clips, preparing files beforehand makes faults easier to isolate: the same processed file should sound the same each time it enters the sequence, subject to the playback chain. Whether you use FFmpeg or another tool, preserve originals and label processed versions clearly. This is particularly helpful when several people maintain the channel or when a rotation includes older recordings whose original levels vary.
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
How do I make all the videos in my YouTube playlist play at the same volume?
If you are watching the playlist, check whether Stable volume is available in the player for each video. If you own the source clips and are assembling a live programme, measure and adjust them before streaming, then listen across transitions. Neither method guarantees that every video will sound identical.
Is Stable volume available on every YouTube video?
No. YouTube says Stable volume is not available for all videos, so you should not rely on it as a playlist-wide solution. Its help material also does not establish identical behaviour across active live streams, replays and ordinary uploads.
Should I use −23 LUFS for a YouTube Live playlist?
−23 LUFS is the EBU R 128 broadcast programme-loudness recommendation, not a universal YouTube target. The suitable working level depends on your programme and audience; measure consistently, control peaks and audition the result rather than treating that reference as a platform requirement.
Does loudnorm make different clips sound exactly the same?
No. FFmpeg’s filter can help bring measured loudness towards a common target, but matching integrated loudness does not make dynamics, tone, speech clarity or transitions identical. Listen to the processed clips in sequence and make any needed editorial adjustments.