Serviio

I think the behaviour of this function needs tweaking slightly, particularly to benefit low CPU power servers like NAS units. At the moment, remuxing audio is always classified as a transcode, but it perhaps shouldn't be. The logic should instead check whether the audio is actually being transcoded. i.e. don't downmix if "acodec=copy".

Consider for example playing an mkv containing h264 and aac5.1 on a renderer that plays these streams but only in m2ts. The device profile will feature a transcode line to remux the container, but the audio stream will be copied not re-encoded. Currently the command line that's being fed to FFmpeg is to downmix which, in the case of aac5.1 > aac 2 channel will fail (due to an FFmpeg problem). Since I'm serving from a NAS I wouldn't really want it set on preserve original, because this seems to break DTS remuxing (another FFmpeg limitation) but it would also cost a lot of additional CPU power for those occasions that I really am transcoding audio.

Are there really renderers that can't downmix themselves (i.e. that would need everything forcing to 2 channel)?

if you don't specify targetACodec or it's the same as the source it will just remux the stream (-acodec copy). The downmixing setting is not taking into consideration in this case. Is that what you meant?

Yep, that's it. I thought it was always hard setting it though with -ac 2. I'll have to check again. How come I have the problem with remuxing 5.1ch aac then unless I put it on Keep Original?

Not sure, try the ffmpeg command manually

False alarm. When I tried again, I saw the affected film was in the list twice and as I selected it it became one and then sorted out the problem. So it was either a database rescan in progress or something, or a Sony firmware glitch I think.