I have to convert big wav files using a speech codec. There are a lot of silences and I have to cut away them to take final size as small as possible. Im testing Speex with DTX (in addition to VAD/VBR): it works good but, as documented:
QUOTE
we cannot just stop writing to the file, only 5 bits are used for such frames (corresponding to 250 bps)
, so it can't cut out silences.

Maybe I need to use another tool (better if command line) that detects silences into wav, cut them out then rebuild wav so than I can give it to Speex (or another codec).

Any suggestions?