Algebraic separation applied to concurrent vowel separation and ECG signal separation

Separation of a mixture of periodic or quasi-periodic signals into its constituent components is an important problem that finds many applications. In this paper, the algebraic separation approach is applied to the problem of separating mixtures of real and synthetic vowels and mixtures of ECG signals. The algorithm is able to effectively accomplish component separation via the restoration options even when there is significant harmonic overlap.