ISO Files - Subtitle and Audio Language

[deibit]

Hi,

I have around 2.000 ISO images (UHD, BD, DVD) in my Jellyfin library, I use Jellyfin to build and steer the library and to "cast" the ISOs to a Oppo 203 Player.

Thing is, Jellyfin is able to partly detect and extract metadata information from the ISO, but the language of the subtitles and the language of the audio is not shown.

In the Audio dropbox i see the audio streams with the format (e.g. DTS HD-MA 5.1) but not the actual language (e.g. English). Same for the subtitles, the different subtitles are listed in the dropbox for "Subtitles" but all are listed as "Undefined - PGSUB"

I know the info is in the ISO itself, since TMM5 is able to extract it. The NFO generated by TMM5 also contains the information, but apparently Jellyfin is not reading those tags from the NFO anyway.

Apologies if this has been discussed in the past, but the search didn't return any result.

Thanks!

[TheDreadPirate]

Unfortunately, while we document that ISOs should "work", as in you can play them, they are not officially supported for reasons like this.

Can you share the contents of an NFO? It should have a language tag and I believe Jellyfin reads them. What Jellyfin version are you running?

[deibit]

Hi!

I’m not really waiting for ISO playback support in the Jellyfin client. I wouldn’t use that anyway, since I simply cast the ISO to an external hardware player. What would be nice, though, is if the metadata extractor worked with ISOs (the same way TMM5 does).

What I don’t understand is why Jellyfin doesn’t read the NFO stream information at all. I’m using Jellyfin 12 now, but it never worked in previous versions either.

I’ve attached an extract of the NFO <streamdetails> section below (I can paste the full NFO or other examples if needed).

What Jellyfin shows (for example, for the first audio stream) is: “Dolby TrueHD + Dolby Atmos – 7.1”  and it doesn’t show the language.

For every subtitle it just shows: “Undefined – PGSSUB”

So I assume Jellyfin isn’t reading these tags from the NFO at all (it seems to read other tags, like <tag>, <plot>...). It looks like it tries to parse the information directly from the ISO, fails to extract everything, and only displays whatever it was able to detect.

I don’t really mind if Jellyfin can’t parse stream information from ISOs (even though that would be very nice to have)… but at the very least it should be able to read the available information from the NFO, right?

 On a side note, Emby actually does it worse… it shows an empty list for languages and subtitles. It looks like Jellyfin at least tries harder (and then fails later) 😊

<fileinfo>
    <streamdetails>
      <video>
        <codec>hevc</codec>
        <aspect>1.78</aspect>
        <width>3840</width>
        <height>2160</height>
        <resolution>4K</resolution>
        <hdrtype>hdr10</hdrtype>
      </video>
      <audio>
        <codec>truehd_atmos</codec>
        <language>eng</language>
        <channels>8</channels>
      </audio>
      <audio>
        <codec>eac3</codec>
        <language>eng</language>
        <channels>8</channels>
      </audio>
      <audio>
        <codec>ac3</codec>
        <language>eng</language>
        <channels>2</channels>
      </audio>
      <audio>
        <codec>eac3</codec>
        <language>deu</language>
        <channels>8</channels>
      </audio>
      <audio>
        <codec>eac3</codec>
        <language>fra</language>
        <channels>8</channels>
      </audio>
      <subtitle>
        <language>eng</language>
      </subtitle>
      <subtitle>
        <language>deu</language>
      </subtitle>
      <subtitle>
        <language>fra</language>
      </subtitle>
      <subtitle>
        <language>eng</language>
      </subtitle>
      <subtitle>
        <language>deu</language>
      </subtitle>
      <subtitle>
        <language>fra</language>
      </subtitle>
    </streamdetails>

[TheDreadPirate]

Can you share your server logs via pastebin? I don't see any github issues that sounds like what you're seeing. I can't imagine you would be the first person to notice if this is an actual bug.

[deibit]

Will try to paste the server log this evening. I didn’t find anything in the forum but Gemini returns the information down below. I guess the difference is that TMM5 uses Mediainfo to extract metadata while Jellyfin just checks the stream with ffprobe. Ffprobe will never return this data, since this information is in the IFO and not in the stream. The thing I don’t understand is why Jellyfin is not falling back to the information in the NFO… 

Why the NFO Metadata Fallback Fails
NFO Scope Limitations: Jellyfin uses its NFO reader strictly for high-level library metadata—such as the movie title, release year, plot summary, cast, directors, and genres. It explicitly ignores media-level stream tags inside the NFO.
Hardcoded FFprobe Dependency: Jellyfin determines audio, video, and subtitle stream details (codecs, channels, and languages) exclusively by probing the media file itself via FFprobe. It does not allow external XML/NFO files to inject stream-level properties.
The ISO Dead End: When Jellyfin hands the ⁠.iso⁠ file to FFprobe, FFprobe fails to read the language headers from the DVD ⁠.IFO⁠ / Blu-ray ⁠.clpi⁠ mapping files inside the image. Because FFprobe returns ⁠Undefined⁠ for the ISO's internal streams, Jellyfin displays ⁠Unknown⁠, ignoring whatever language data tmm5 logged in the ⁠movie.nfo⁠.

[TheDreadPirate]

That Gemini summary lines up with my understanding of how Jellyfin works.

[deibit]

Wouldn’t it make more sense that the whatever info the NFO provides, overrides whatever Jellyfin parses on its own? I mean that’s precisely the point of generating a NFO in the first place, right?

[TheDreadPirate]

The point of NFOs in Jellyfin is for A) Kodi based clients and B) to retrieve ONLINE metadata and then preserve it so that you don't have to reach out to TMDB/TVDB/etc again if you ever have to rebuild Jellyfin. Movie/show/episode summaries, season and episode orders, cast, crew, artist bios, etc.

My understanding is that Jellyfin has always pretty much exclusively used ffprobe for all information about the contents of the FILE and NFOs for information about the show/movie/music the file represents.

This may simply be a hold over from Emby or an explicit design choice, I'm not sure which.

[deibit]

I see

I think it would be a nice addition to include the parsing of the stream information from the NFO, like for example Kodi (attached image from my Kodi library). I haven’t found a stream extractor better than mediainfo, I kind of lost my hope on ffprobe… I understand that switching to mediainfo is a big thing, but having jellyfin reading from the NFO might have been easier (I guess)