Don't forget that the power supply can introduce its own noise and resulting distortion. As it's powering an active device stright in the signal path, I personally wouldn't use the circuit shown in the article, as LT1085/1033 are not particluarly 'quiet' and the smoothing there won't give a flat DC waveform. Good enough, but certainly not 'very low noise'.