Implementation of the "vocoder extension" by SV9OAN, as described in https://github.com/chazapis/pydv, which allows the use of the open source Codec 2 with D-STAR. With the changes here, xlxd/ambed can be used to transcode and bridge the two formats.
The solution implemented uses an additional DExtra listener on a different port (30201 instead of 30001). The new port is to be used by reflectors using the "ORF" prefix (Open ReFlector). Any client connected to an ORF reflector will receive streams encoded with Codec 2. All other D-STAR protocol handlers will still send out data encoded with AMBE. Note that the protocol/port only affects data transmitted by the reflector. The stream vocoder is recognized by all protocol handlers, so a client can still transmit data using any vocoder on any port. The rationale behind this is that DExtra links may be used by repeaters or other reflectors, so it is not really possible to know what their clients support. So, nothing will change when linking a repeater to an XRF reflector, but will do when linking to an ORF one.
Summary of code changes in xlxd:
* Added protocol "DExtra Open", listening on port 30201. The protocol handler is subclassed from DExtra. Minor changes were required to make the DExtra implementation more "abstract" to allow that.
* Codec used no longer a protocol client property. All D-STAR protocol handlers use the header to determine the protocol. DMR protocol handlers just specify the appropriate codec when opening a stream.
* DV frames now also include Codec 2 data.
* Interface with ambed changed to 1 codec in, 2 codecs out.
* Interface with other xlx peers changed to accommodate the enlarged DV frames, so compatible peers will exchange fully transcoded streams.
* Major version changed to 3, to recognize peer compatibility.
Summary of code changes to ambed:
* Added (virtual) interface class for Codec 2.
* Vocodec channels now have 3 interfaces (1 in, 2 out) and are grouped together in triplets. Each group contains all possible permutations of respective interfaces - 3 channels. Only one channel from each group can be used at a given time.
* When an incoming packet is decoded in the interface loop, it is duplicated and sent to two separate voice queues (one for each outgoing codec).
* Interface changed to 1 codec in, 2 codecs out.
* Major version changed to 2, as the interface is incompatible with previous versions.
Antony Chazapis