Wheezes in the respiratory sounds are indicators for a number of diseases. Automatic detection of these signals -which have musical character and one or more dominant frequencies-is very important. In this study investigation of wheeze detection in the respiratory sounds using Hilbert-Huang transform is studied. This method allows analysis of signals which are neither stationary nor linear. Initially Intrinsic Mode Functions are found using Empirical Mode Decomposition. Instantaneous frequencies calculated using Hilbert transform of these signals were found to have distinct values with low variations, and constancy and variation of these instantaneous frequencies were affected by the power of the dominant frequencies in the wheeze.