My experience is with the usbradio channel. Not sure if it will happen with simple usb.
The "fix" could be a hack as you say. Although, if the hardware requires it ... then you might say it's a feature ;-)
To me it seems like some pipeline underflows, or overflows, right at the start. My observation is that it happens when ASL tries to key up the radio and talk ... essentially at the same time. It does not happen *every* time. But I heard the effect three times in a short while after starting the repeater up last night. And I think I heard one chopped up CW ID today sometime. Once the audio gets going, it's fine.
People whose radios are slow to open the squelch will never have heard this. They will also never have heard the first letter of the repeater ID on a cold key-up (and probably don't care anyway).
Doing a Google search I am reminded that this fix was for simple usb (I do not know if it is applicable to usbradio - as you know, that driver is not favored in the other flavor of asterisk controller).
Re-reading the text below (which google just found for me), this fix was actually intended for another problem. Although it might also fix the issue I am talking about.
... the simpleusb driver continues > to send frames of silence to the audio adapter at all times; even when the > transmitter (PTT) isn't active. This reduces the impact of the variable > start-up latency issues which simpleusb has classically exhibited.