Hi Team,
I would like to report a significant audio regression introduced in webOS 26 on the LG G6 (2026 model).
This appears to be the same issue that was widely discussed and eventually resolved in webOS 25, as documented in this thread. While webOS 25 restored proper multi-channel eARC passthrough for HLS streams, webOS 26 once again forces all multi-channel audio tracks to output as PCM 2.0 (stereo) when delivered through an HLS container.
When using media applications such as Jellyfin that rely on the native webOS media pipeline through HTTP Live Streaming (HLS), multi-channel audio streams (for example, Dolby Digital Plus 5.1 or Dolby Atmos via DD+) are not handled correctly. The TV downmixes the audio to standard 2-channel stereo before transmitting it over HDMI eARC.
For apps such as Netflix and Apple TV, audio works as expected, with Dolby Atmos and proper multi-channel audio being passed through eARC to my soundbar.
TV Model: LG OLED G6 (2026)
OS Version: webOS 26 / 11.1.0-2707
Audio Setup: Soundbar connected via HDMI eARC (Digital Sound Output set to Pass Through)
Software Version: 43.10.69 (Latest Version) - No further updates available
Dolby Vision: 28.12
Expected Result
The TV should pass the encoded multi-channel bitstream (DD+ 5.1 / Atmos) directly to the external audio device over eARC, consistent with the behavior observed in the later webOS 25 releases and with other apps such as Netflix and Apple TV.
Actual Result
The connected sound system receives a PCM 2.0 (stereo) signal instead of the original multi-channel bitstream.
Note: The “Dolby Vision & Dolby Atmos (via Audio Device)” banner still appears on the TV. However, the connected sound system only receives PCM 2.0 (stereo) rather than the original multi-channel audio stream.
Since the same content and workflow function correctly on webOS 25, it appears that a regression may have been introduced in the HLS media pipeline in webOS 26, specifically in the way adaptive audio streams interact with the eARC passthrough layer.
Could the webOS team please investigate this audio regression affecting 2026 models and provide a fix?
@narae0.kim @buletti