Experimental method for OPUS CODEC packet loss concealment - morse code audio over mumble

[Chuck Vaughn]

https://youtu.be/m2P1-18sDhU

A/B test  using - with PLC and without PLC

of note...is...that there may be some familiar sounds from listening to your own Packet Loss(without PLC in this example)

on iCW on your own Mumble client...ie...

- full volume audio spikes from dropped packets...

-  rubber band "twang" sounds  

et al...

SETUP: using Gstreamer's IDENTITY MODULE https://gstreamer.freedesktop.org/documentation/coreelements/identity.html?gi-language=c to simulate frequent packet loss...using the OPUS CODEC at 144k bitrate, and 2.5 ms frame-rate... a CW REGENERATION method was used to rebuild the lost packets over ip, using the app - TRIGGER MIDI MONO https://lsp-plug.in/?page=manuals&section=trigger_midi_mono CW AUDIO output sending by the CW KEYBOARD KEYER - the AD5DZ CW KEYBOARD https://github.com/recri/keyer The ad5dz's cw keyboard's compensation setting was used to lower the element length of both the dit and the dah by the same amount of milliseconds...then it was increased on the TRIGGER MIDI MONO app to regain what was lost from the setting on the CW KEYER at 2.5 ms/opus codec frame-rate....it keeps a packet that is lost down to 2.5 milliseconds...which is then hidden/recovered by using TRIGGER MIDI MONO settings(as shown on video) to rebuild the CW lost packets by increasing the "holding" "ON/release TIME" of the TRIGGER MIDI MONO REGEN PROCESS - making up for the lost time created by the packet loss...and adjusted to sounds like "normal" CW dits & dahs at the speed of the sent CW from the source...