Speech Enhancement in the STFT Domain