The silver mica caps noticeably lower the noise floor and increase dynamics by extending the effective operating frequency range of the electrolytic and reduce internal vibration of the electrolytic.
Were you able to measure a reduction of the noise floor?
I ask because the noise in the filament supply is 120Hz. The 80,000uF of capacitance per channel has an AC impedance of 0.016 ohms at 120Hz (pretty close to a dead short!). A 0.056uF cap has about 24,000 ohms of AC impedance at 120Hz. The impedance at 120Hz will drop by about 0.001 ohm when you add those bypass caps.
(Still, it's not surprising that it sounds different)