Transcribed long audio vocals . We need to split them into short speech