General questions about OGG Vorbis

[TheDreadPirate]

Ah, yes.  I remember this discussion about OGG gain metadata.  Not in 10.8 AFAIK.

If I have understood everything correctly, when I use the official apps (Linux desktop and Android), a conversion of the OGG -> AAC format takes place. Recoding always means a loss of quality. Or does it?

Jellyfin will not transcode OGG/Vorbis, or any audio codec, unless configured to do so.  The official clients and the two 3rd party clients I've used (Finamp and Symfonium) have a setting for "Internet bit rate".  The desktop clients also have a max audio bit rate setting, IIRC.  If the source bit rate is higher than this configured max bit rate it will transcode.

The official clients will request a transcode to AAC, Finamp also requests AAC, and Symfonium requests OGG/Vorbis.

If you set this max audio/Internet bit rate to auto, it should never transcode unless there is a compatibility issue with the client.

Transcoding from lossy to lossy will always result in quality loss.  Probably not noticeable for reasonable bit rates.

I like OGG-Vorbis because it is an open format and is also qualitatively better than MP3 in some respects. Will Jellyfin fully support this at some point?

See above.  The client requests what codec is used for the transcode.  AFAIK, there is no way to change what audio codec the client requests for the official clients or 3rd party clients.

Another question in this context is the support of dynamic bit rates. When creating OGG files, I select -m 256 -M 512 as the lower and upper limit. Does Jellyfin process at all when streaming to the client?

OGG isn't the only codec with dynamic bit rates.  MP3 can do VBR and FLAC is VBR by design.  AFAIK, if a Jellyfin client supports an audio codec, it supports all features of it.

[pixel24]

In the current version (10.9.9), I enabled the "LUFS scan." Finamp noticeably adjusts the volume based on this, which often results in the start of the song being much louder before being immediately corrected. This can sound strange.

Wouldn't it be better or possible for the appropriate tag with the correction value (-/+ dB) to be evaluated by the server, so that the volume is already sent correctly? This would make LUFS unnecessary, and the client wouldn't need to make any adjustments. Or am I misunderstanding something?

[TheDreadPirate]

The LUFS value is just a number listed in the database entry for a song/album. The server makes no modification to the song when it is direct played. Not sure it does anything with the LUFS value when it transcodes either.

This would be a question to for the Finamp devs. They have their own Discord channel.

https://discord.gg/HgFZh6by

I've always encountered this issue, btw. I only notice it when I skip from a "quiet" song to a "loud" song (one with a big negative dB adjustment). I haven't noticed it when songs naturally end and the next starts.