Splitter Recursively splits audio on non-voice segments, until target audio segments are shorter than a parameter (default 8sec) or cannot be cut on silence Test cd asr_toolkit python -m unittest split/test.py