Hi all, Today as I tried to implement transcoding a videofeed to an icecast audio-only stream I have encountered some behavior that, in my opionion, leaves room for improvement.
The setup used in this scenario consists of * an Open Broadcaster Studio (OBS) instance sending a video/audio feed * an nginx endpoint to receive and re-stream said feed to various hosts (including localhost) * and lastly an AzuraCast (https://github.com/AzuraCast/AzuraCast)instance which employs Icecast2.4 for audio broadcasting. AzuraCast allows streamer/DJ accounts to live-broadcast to any of its radio-stations by connecting through a Broadcasting software of your choice (noteworthy mentions are BUTT or Mixxx) to a dedicated port (e.g. 8005) and a static[1] mountpoint "/" and providing credentials. [1]static as in there is no easy way for me to change this design consideration) The way in which the icecast protocol is currently implemented in ffmpeg, it will check for characters after the trailing slash and assume this as a valid mountpoint. A trailing slash by itself will always be discarded with the following error: "No mountpoint (path) specified!" see: https://ffmpeg.org/doxygen/2.5/icecast_8c_source.html [lines 157-161] so icecast://source:pass@host:port/ is considered invalid by ffmpeg (I cannot think of a quick and dirty workaround either). To my understanding assigning a lone slash as a mountpoint name isn't considered invalid by the IceCast specification and therefore I suggest that it would be an enhancement to lift this current limitation on the way that mountpoints are parsed in an ffmpeg+icecast instruction. kind regards, c0re
signature.asc
Description: This is a digitally signed message part
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
