Currently I'm using PCM 24 bits as I can't use L24.
There are three ways of streaming audio samples as follows; all have exactly the same audio sample content, and all work equally well.
- L24 = 24 bit audio samples sent in high byte, middle byte, low byte order; the HTTP Content-type header gives bit depth, channel width, sample rate
- AIF = 24 bit audio samples sent in high byte, middle byte, low byte order; the start of the stream has a header giving bit depth, channel width, sample rate
- WAV = 24 bit audio samples sent in low byte, middle byte, high byte order; the start of the stream has a header giving bit depth, channel width, sample rate
I still see information on the display of the Marantz. I suppose that's because tagging information is still send.
Correct. The meta data is sent via the UPnP / DLNA commands. If the file is a Flac, Mp3 etc. the meta data might
also be duplicated as embedded tags within the stream itself. However in all cases the renderer is required to use the former (UPnP) meta data rather than the latter (embedded) meta data, so essentially this duplicate embedded meta data is just a waste of space and time.
I don't understand why I have issues with flac and not with PCM. Or it was just luck that it didn't play songs with "strange" tagging information during my test using PCM 24 bits.
PCM is pure audio samples only. There is nothing else in the stream that might cause a renderer to choke. Each sample is sent one after the other. All are exactly the same size, structure and tempo from beginning to end. So if the renderer is able to play the first sample it requires no special smarts for it to keep on clocking the samples through to the end of the stream. On the other hand, in Flac, Mp3 and other compressed audio formats, the audio samples are squeezed up into blocks using data compression algorithms so some samples are essentially longer (more bytes) than others. This means that the renderer needs to be cleverer and keep track of what is going on. Also if there is embedded meta data in the stream, the renderer needs to have sufficient smarts to detect that those bytes are not audio samples and to discard them.
Is there a way that MC can check for "wrong" tags?
We don't know if your tags are actually "wrong" -- it could be that the tags are just fine, but that the renderer is "broken". But in any case if you send PCM there won't be any duplicate embedded tags, so the problem does not arise.