Instead of completely removing all silences detected over (1 second), maybe we need an option to cut them by half instead? Often trimming out the silences can make the audio recording sound a little more processed and artifical. A 50% reduction on silences might correct this.