I think I have my threshold set to 92, because anything higher was clipping other people's amps. I always normalize down to 99.99% to prevent clipping, but the audio itself should be top notch. I get all studio productions from a record pool, and I work on them in 48,000khz sample rate, 32-bit float depth, stereo. When I mix them down I use 320kbps LAME MP3 encoding with a 512-point sinc sampler interpolation.
You may be getting bad sound from the samples I used, I had to dig in and get a sample from an acapella for "California Love" that I made myself. There was no acapella made that I ever saw, so I used that. So it may sound grainy on that part, but the rest should be fine, especially since most of it is post-production work that I added to the track.
Certainly sounds like you know what you're doing. I also use the LAME encoder with audiograbber. Nearly all commercial cds are normalised around 96dB to 98 signal level now. So you should really be aiming for around the same level. But your file is only at 91dB.If i compare your remix with my original sourced cd of California Gurls , level wise yours is almost 6dB lower, that is a big difference.