Switch File Format in Database

Hoping to get clarification on the following issue as I spent a lot of time improving my database. Thanks

Most of my audio files are stored in the FLAC format with a resolution 96/24. I would like to make make all my music Apple and Airplay compatible. My plan is to change the FLAC 96/24 files to ALAC 44/16. My database consists of approx 2000 albums, Many of them are vinyl recordings and the albums could not be identified by Roon automatically. I spent a considerable amount of time on manual matching.

My fear is that if I replace the FLAC files with ALAC files the manual matching work will have to be redone again. What happens if i only change the resolution from 96/24 to 44/16?

Thanks for clarification.

New file formats for the same albums will result in new imports. Your manual identifications and edits will not carry over to the new imports.


Did you do all your metadata edits within Roon? Or with something like mp3tag that directly embeds metadata into the audio files? If the latter, hopefully whatever conversion tool you use will copy the metadata to the target file.

You mention Apple and Airplay. At least for Airplay, Roon knows how to play to Airplay devices and will do the downsampling on the fly during playback. So you don’t really need to convert the files. Some recentish version of iOS also added native FLAC support (though I don’t own any iOS devices, so I don’t know if it has bitrate/samplesize restrictions). So you shouldn’t need to convert the files for the iOS devices, either. And OSX has had FLAC playback capability for ages through various playback tools.

Ok understood. Say i only downconvert the high resolution files FLAC 96/24 to 44/16. what is the step by step process not in order to preserve my library?


I edit tag info mostly with mp3tag. Conversions I do with dbpoweramp which preserves the tag info.

Yes Roon will play FLAC and even downconvert to 44/16 for Airplay. However I would like to avoid manipulation of data as much as possible as there is the possibility of downgrading the SQ.

My ears even tell me that involving DSP results in downgraded SQ. Not sure about all the volume control possibilities.

There is a lot of upsampling, downsampling, crossampling, changing bit depth and volume going on if DSP is involved and Airplay involved.

Is there a common opinion whether all this has an effect on SQ?

This is incorrect. Roon is doing all this processing on the Roon Core and with ~5 seconds of prebuffering to the endpoint. The endpoint is getting what amounts to a WAV file from the Core – it’s raw, uncompressed PCM. There will be no difference in SQ at the endpoint (unless the Core is also the endpoint, in which case, you’re doing it wrong).

If there is a difference in SQ in this case, then it’s a difference in the algo Roon uses to dither down vs. dbpoweramp’s dithering algo. That’s it.

I use XLD to convert FLAC to Apple Lossless and then use iTunes to set tags. I also use an iTunes script to set filenames to be in a consistent structure (Album Artist - Album - Disc Number - Track Number - Song Name. The last step isn’t really needed for Roon import but it means that all of my files are named consistently.

I find that doing this before importing into Roon means that things are usually identified easily - apart from things that I wouldn’t expect to be identified such as live recordings.

I wouldn’t use XLD to downsample - I’d keep the original 96/24 - unless you were particularly bothered about storage space.

@DuckSoup: when you sa you use itunes to create the tags, how doe this work?

I just use iTunes to set Artist, Album Artist, Disc Title, Genre, Year and (if necessary) track names.

To set things for an album just right-click on the album and select '‘Album Info’.
For individual tracks, right-click and select ‘Track Info’.

No MP3 Tag needed then to create permanent tag info.

Does Roon not use the info and update that i provide for the unidentified albums and update their central database?

Not at present, though crowd sourced data has been discussed.

1 Like