The effects on speech detection of low sample frequency audio data