Plex adds Boost Dialog and loudness normalization in Plex Media Server 1.43.4
On September 15, 2026, Plex shipped two Plex Pass audio options: Boost Dialog, which lifts the frequencies where voices live, and Normalize Loudness, which evens out volume swings. The loudness analysis that powers them runs on your CPU during maintenance, so measure the cost before you turn it on for a large library.
September 15, 2026. Plex announces two new audio playback options in Plex Media Server 1.43.4: Boost Dialog and Normalize Loudness. September 15, 2026. Both ship immediately, gated behind Plex Pass. September 25, 2026. Self-Host Weekly flags the news, framing it as the end of a problem as old as the home theater: turning the volume up to hear dialogue, then getting your ears blown out three seconds later. Why it matters: these two toggles turn the self-hosted media server into a machine that analyzes audio in the background, and that analysis has a CPU cost worth planning for.
Two features, one complaint
The scenario is universal. You watch a film, a dialogue scene is mixed too low, you raise the volume, and an explosion arrives. Boost Dialog and Normalize Loudness attack the problem from opposite ends.
Boost Dialog isolates the frequencies where human speech lives and brings them forward, so a whispered line stays intelligible without dragging everything else up with it. Normalize Loudness holds the audio at a consistent level and softens the peaks of very loud scenes — no more reaching for the remote every time the score swells.
Both are turned on mid-playback through the on-screen controls, under Playback Settings and then Playback Options. One detail matters: enabling either option forces the audio track to be transcoded. On a modest server, that is added CPU load on every play, not just during indexing.
Loudness analysis, the invisible part
Normalize Loudness does not work on the fly for any track. It requires the track to have been analyzed by the server first, and that analysis does not happen by magic.
Two settings drive it, both on by default: Analyze audio tracks for loudness, set to As a Scheduled Task, and Enable Loudness Analysis in the advanced settings of each library. The analysis runs during the server’s scheduled maintenance window, which means Normalize Loudness only spreads across your library over the following days.
For self-hosters the consequence is concrete: a large library means a lot of analysis, and the server shows higher CPU usage while it works through the backlog. Setting the option to Never disables it if the box is already busy, while As a Scheduled Task and when media is added kicks off analysis the moment new content lands.
A track that has been processed is identifiable in two places: Media info shows true, and the stream XML carries the attribute canNormalizeLoudness="1". You can check it through the local API:
# An analyzed track exposes the canNormalizeLoudness flag in the stream XML
curl -s "http://127.0.0.1:32400/library/metadata/$RATING_KEY" \
-H "X-Plex-Token: $PLEX_TOKEN" | grep -o 'canNormalizeLoudness="[0-9]"'The Plex Pass toll, at the center of the debate
Neither feature is free. Boost Dialog and Normalize Loudness require an active Plex Pass on your account, or membership in a Plex Home whose admin has one. The loudness analysis itself only runs on servers whose admin holds a Plex Pass.
That is a deliberate commercial choice, and it lands in a specific context. For months, part of the self-hosting community — the so-called Plex Exodus — has been migrating to Jellyfin or Emby, alternatives that deliver comparable features without a subscription. Placing the audio upgrade behind Plex Pass is a bet: keep paying users by fixing a real usability flaw, while leaving the free tier unchanged.
The bet has logic behind it. Inaudible dialogue is one of the most-cited gripes against home media servers, and no competitor has yet shipped a fix this legible. But every paid feature also strengthens the argument of those who prefer free software without a toll.
Client-side prerequisites
On the player side, compatibility is uneven. Plex Media Server 1.43.4 is the minimum on the server, but a recent client is also required: Plex for Mobile from 2025.28.0, Android TV and Apple TV from 2025.28.0 (Apple TV and Android TV require the new-experience apps), Fire TV from 2025.8.0, and Roku from 9.0.31. Plex Web and Smart TVs arrive “soon,” per the announcement.
That gap has an operational implication: in a household with several devices, the option can be present on the phone and absent on the living-room TV until the client updates. Checking client versions is part of the rollout, not just the server.
Loudness normalization has a standard behind it
Behind Normalize Loudness sits a real measurement, not a simple volume knob. The EBU R128 standard — and its ITU-R BS.1770 counterpart — defines a unit, LUFS (Loudness Units Full Scale), which rates the average perceived loudness of a program rather than its instantaneous peak. That is the whole difference between “normalizing” and “compressing”: a peak limiter crushes dynamics, while LUFS-based normalization aligns the overall level without destroying the swings the mixer intended.
For Plex, the point is to precompute that value per track. Loudness analysis reads the audio stream, measures its integrated LUFS level, and stores the result as metadata. At playback the server only has to apply a constant corrective gain — hence the transcoding: adjusting gain without re-encoding would be a plain volume offset, which Plex already does elsewhere, but the vendor chose the transcode path to guarantee the result across every client.
The practical consequence is twofold. First, the analysis cost is predictable: one pass per track, scheduled, not a per-play load. Second, the transcoding cost is recurring: every play with the option enabled spends CPU re-encoding the audio. On a server that already transcodes video, the addition is marginal; on one that direct-plays everything, it is a change of regime worth measuring before you roll it out widely.
The storage footprint of loudness metadata
One question keeps coming up on the Plex forums: does loudness analysis eat disk space? The answer is reassuring. The analysis does not re-encode or duplicate the audio; it computes a loudness value and stores it as a small piece of metadata alongside the library entry. For a library of thousands of items, the total footprint is measured in megabytes, not gigabytes — negligible next to the media itself.
The real cost is compute, not storage, and it is front-loaded on the initial sweep of a large library. That is why the scheduled-task window exists, and why Plex lets you defer analysis to when new media is added. Operators watching a busy NAS should treat loudness analysis the way they treat thumbnail generation or intro detection: a background job whose budget belongs in the maintenance window, not in the playback path.
Verdict
If you already have Plex Pass and a server that idles overnight, enable loudness analysis as a scheduled task and let Normalize Loudness spread across the library: the comfort gain is real, and the CPU cost is confined to the maintenance window. If your server is a small NAS that is already loaded, set analysis to Never or to As a Scheduled Task and when media is added only, and enable Boost Dialog on demand — audio transcoding adds up on every play. And if you are weighing Plex against Jellyfin or Emby, this kind of toll is exactly the variable to put on the scale: audio comfort has a price at Plex, and your alternatives deliver it differently.