This feat.req. is similar to AnotherAndrew's http://nch.invisionzone.com/index.php?/topic/20627-feature-requests-preserve-lead-ins-and-fade-outs/. I was volunteering for Lakota Language Consortium to split words recorded by dictor for New Lakota Dictionary Online. Split Audio File Splitter is the best, perhaps the only utility to split the recorded text into words. It has only one flaw: when the recording is not ideal, there's no optimal Threshold/Duration combination: --if you have Threshold too low then the S.A.F.S. starts to split clicks, sighs, other noises, beside real words. --if you have threshold too high then SAFS crop initial/final s/sh/p, consonants, especially in consonant clusters (sp-, sht-, pt- etc.) Sometimes SAFS does both - misidentifies clicks as words and cuts off leading/trailing consonants. The way out, as I see it, is addition of one or two additional parameter(s): leading/trailing offset, in msec, (100, 300, etc.) that is, to move back the start and move forward the end of the sound by nnn msec to the identified boundaries of a sound. This would add s-, -sh, etc. into the splitted audios for sure. I had to do many splitting manually for sp-, pt- etc. words.