Best subsequence selection of heart sound recording based on degree of sound periodicity

Noise often appears in parts of heart sound recordings, which may also be longer than necessary for subsequent analysis. Human intervention is needed to select the subsequence with the best quality. Presented is an automated scheme for best subsequence selection to obviate such human intervention. The scheme is based on the finding that heart sound signals with less random noise contamination have a greater degree of periodicity. Both simulations and experiments validate the scheme.