A current Radio World ebook explores trends in codecs. In this excerpt, Anthony Gervasi, Intraplex sales manager for GatesAir, offers insights.
Radio World: What do you consider the most important current or recent trend in how broadcast codecs are designed or deployed?
Tony Gervasi: The addition of SRT to audio codecs, and the ability to embed E2X data within the MPX payload to keep FM/HD1 synchronized on all transport paths.
RW: What are the implications of FM-MPX and microMPX for the way the radio industry chooses and deploys codecs?
Gervasi: There are variations of FM-MPX codecs, some that provide uncompressed MPX with RDS in 1.64 Mbps (rather than 2.2 Mbps or 3.2 Mbps) by using effective data packing. This reduction in bandwidth allows for MPX to be transported via 950 STL as well as LTE and satellite.
MicroMPX using 384k for compressed MPX with RDS allows for transport over slower connections as well as the ability to transport more than one over a traditional STL system.
RW: How do today’s codecs integrate with today’s AoIP networks and infrastructures?
Gervasi: Having the ability to ingest and output AES67, Wheatnet and Livewire allows the codec to show up on the cross-matrix routing software as a native appliance, allowing audio and GPIO to be steered and transported from site to site or from site to many. This also allows audio that is not in an AoIP format to be ingested into the AoIP network and routed accordingly.
RW: How do today’s codecs avoid problems with dropped packets?
Gervasi: A few different methods for UDP paths — one, by deploying redundant streams using path and/or time diversity to assure reliable transport, and two, by deploying FEC with buffering.
For TCP/IP paths, by using Secure Reliable Transport or SRT.
RW: Why is Forward Error Correction important?
Gervasi: FEC can be a double-edged sword. When using a RTP over UDP, adding 25% FEC with the proper amount of buffering can eliminate most minor packet losses due to path changes, micro path interruptions, etc.
If you are on an oversubscribed switch or system, adding FEC can make things worse. When troubleshooting packet losses and drops and the customer is running uncompressed audio at 44.1 (1.58 Mbps) with FEC, I will have the customer change to Opus 192 kbps with no FEC and see if the packet loss or drops lessen or get eliminated.
If the packet loss stops, this tells me that there is a bandwidth choke someplace, which could be a switch or an oversubscribed sub. By adding FEC to the data stream, you are adding additional payload and contributing more to the oversubscription.
