The sample was very noisy in LF and HF ranges. Very loud and distorted too. Ears killer. 1. I reduced the true peaks to -12dB to hear clear what is it about. Usually, the professional audio processing is made at -12dB. 2. I used the first 0.5 sec of your sample, a pattern of the heavy noise, in iZotope Spectral De-noise learning. The reduction level was found by trial-and-error with "output noise only" to not lose the music data. The goal was to push the residual noise beyond the audible range, at -80 dB. 3. After the sample cleaning, I used A high pass filter at 20 Hz order 4 (-24 dB per octave) to remove the rumble. 4. I used a low pass filter at 20 KHz order 1 (-6 dB per octave) to remove the HF noise, following the pink noise curve. 5. I rebalanced the bass, mid and treble bands with "Oxford EQ" VST (steps 3 and 4 too). Max 1 dB changes at a time, better 0.5 dB. This is the best EQ i found so far, following the Steve (our host) suggestion. 6. Finally I boosted the sound to -18 LUFS. I dont use the -0.01dB BS method like in the over-clipped cheap CDs.