Automatic transcription of general audio data: preliminary analyses.