Voice recognition and dictation software in Linux

I need a standalone application that runs independently under linux, that has a GUI for KDE plasma 5.2 (can be simple), that, once started: 1) has a toggle switch, (preferably a configurable keyboard shortcut or key) so when it is toggled on, it starts recording my voice 2) can be toggled off, of course 3) when dictation is over, it leverages on Mozilla DeepSpeech (could also be Google Voice Services, but this costs, and I want it to be as economical as possible) to analyze the speech recorded. *Preferably, it should support continuous speech, and not wait until recording is over, to start transcribing 4) after transcription, it sends the recognized text to an active window of my choice (for example, if I am typing in Writer, or if I am writing an email in Firefox).... in the end, wherever the active cursor is ;-) 5) Main usage case: I dictate medical visit notes. This is much faster than typing. As a reference, this is more or less how JamezQ's Palaver used to work (it relied on Google, and required constant internet access... it also was a little fiddly) [login to view URL] 6) User should be able to update the speech corpus (Mozilla Common Voice) and change recognized language