some ideas:

* custom output format using % encoding
* beat/onset detection (using aubio)
* adaptive threshold, volume pre-scaling (replaygain, ebu128,..)
