後期程序步驟，先後順序。。。

1. filter noise; fade in / fade out to remove noise; high pass filter / remove low-end 

2. EQ sweeping to moderate obvous intrusive noise; 

3. compression (do not remove more than 4-6 dB); add 3-4 dB back after compression tightens things up. 

fast attack (say 0.5 ms) gets engaged immediately (crashing the transient) and reduces volume above threshhold. it feels like the vocal is pushed back. the initial pronunciation of a word is not as dynamic.

4 EQ again to add to clarity and presence (high freq) or warmth (low freq).

5. mix and export