The song bonus from Amplification, is being boosted by other amplification abilities, including the song itself. The easiest way to see this it by playing a fixed bonus song (like Charisma buff for example) without amplification, with amplification, and with amplification cast while you already have the amplification effect on. You see your CHA go up each stage where you would only expect to see it go up once. I believe this even stacks onto itself repeatedly. With this exploit, a bard with no singing AA, no Item: Amplification, and no epic can boost their singing songs to the max mod which I believe is 420% (though my testing puts it closer to 417.666%). My guess is that since Amplification is a Singing song, it’s getting a bonus from that. While that makes sense, it doesn’t seem like the right behavior. My understanding was Amplification should add a 90% flat bonus to songs, like the 80% bonus added by our epic. It seems as though the Singing Mastery AAs are also boosting Amplification even with a single cast. Again, not exactly what I expect.
I’m not sure if it matters but this was on a TLP server, Mangler. I’ve had a couple other bards confirm it’s, all of them with different amounts of Singing Mastery, some with “Item: Amplification”, some without. Doesn’t matter if you have an epic or not, except different other bonuses do change what % Amplification does calculate to for the first cast. I think that’s mainly based on the Singing Mastery AA.
Some concrete numbers: VotS = Voice of the Serpent Item (adds .6 (or 60%) to singing) AA = 3 ranks of Singing Mastery Psalm of Purity resists: 70 PR (actually 69.8ish) 1 + .6 AA + .6 VotS = 2.2 * 70 = 154 expected 154 actual (correct) 1 + .6 AA + .6 VotS + .8 epic = 3.0 * 70 = 210 expected 209 actual (correct with the 70 vs 69.8) 1 + .6 AA + .6 VotS + .9 amp = 3.1 * 70 = 217 expected 245 actual (This is too high! Works out to 3.5x multiplier, should be 3.1) 1 + .6 AA + .6 vots + .8 epic + .9 amp = 3.9 * 70 = 273 expected 293 actual (Again, this is too high! Works out to 4.2x multiplier, should be 3.9)