Neutron Forum

The problem is cause by the malformed values of R128_ALBUM_GAIN and R128_TRACK_GAIN values which come from the tags and which do not conform to RFC7845 standard (http://tools.ietf.org/html/rfc7845) for encoding Replay Gain into Opus tags. I checked sample files and it appears that besides tags there is Replay Gain info in the header of the Opus file. Since next update Neutron will respect Opus header's Replay Gain values and ignore tags if header's info is present, it will solve the problem.

The workaround for now is to remove or update R128_ALBUM_GAIN and R128_TRACK_GAIN values with Mp3Tag utility.

Playing with Foobar2000 revealed a bug in Neutron's Opus Replay Gain if opus file has [output_gain] parameter set, Neutron is replacing it with R128 tag from the metadata and it is not correct. Will be fixed in the next update.

There may be some difference if Foobar200 Replay Gain scanner is used as it will apply +5 dB correction which is a difference between R128 -23 dB LUFS and ReplayGain -18 dB LUFS.

Such +5 dB correction can not be detected in order to apply it in Neutron as Foobar2000 is writing into R128 tags (R128_ALBUM_GAIN/R128_TRACK_GAIN) and thus there is a problem. If Neutron decides to hardcode and apply +5 dB to all R128 tags of the Opus files it will break Replay Gain if they were analyzed with other tools which did not apply +5 dB correction.

So decision is to not apply +5 dB correction unless there is some workaround or better solution, or may be Neutron's Reply Gain settings need an option to convert R128 to ReplayGain that will cause +5 dB correction.

Files are made with desktop foobar2000 (1.5.3) from lossless source. I believe they know what they doing. And of course you know too. I hope that you see the universal solution, because that would be a pain to fix everything as I have a large OPUS archive for Neutron.

dmitrykos wrote: ↑

Mon Jul 20, 2020 6:43 pm

So decision is to not apply +5 dB correction unless there is some workaround or better solution, or may be Neutron's Reply Gain settings need an option to convert R128 to ReplayGain that will cause +5 dB correction.

That may be left for user to decide in settings or default to auto for program to decide it some way, for example, compare the RG and header gain values. This looks as some kind of foobar2000 "standard".

Should you wonder what LUFS means ...