ESD/GNOME via PulseAudio was easy to figure out: just make sure your login/logout/etc. sounds are no bigger than 1 MiB, not 1MB as some claim. You could use “pactl upload-sample ” to test and make sure PulseAudio’d like the sound.
That was easy enough to figure out. Then Skype came along and showed me the weaknesses of PulseAudio and its padsp script.
The issue was that Skype’s sounds would be very choppy, and my voice would be difficult to understand. I knew that PulseAudio had a way of building up a cache of audio before playing it, and that it would auto-adjust, but I figured that that was messing it up somehow. How I would fix an automatically-adjusted system I knew not.
Well, today I just happened to decide to glance at the PerfectSetup wiki page on PulseAudio’s website today. Apparently it’s been updated, because a new guide for Skype was up. And guess what? It works this time!
The bulk of it is, set output to pulse, and set input to some plughw microphone (test each until you hear yourself). You’ll also want to adjust two lines in /etc/pulse/daemon.conf:
default-fragments = 8
default-fragment-size-msec = 5
TeamSpeak was plagued by the same annoying issues, and they appear to be similar to Skype, because it works now after the above workaround.
This was the last OSS program that was causing me problems. Now I can finally say I have no need for the outdated OSS system. Huzzah!