|RESPITE: The CASA Toolkit Page: Documentation: Block Library Index:HarmonicRemoval|
The HarmonicRemoval block implements an algorithm for removing higher harmonics from spectral slices. This is a useful form of post-processing to apply to summary autocorrelations as it removes potentially distracting peaks that are not due to an F0 of one of the sources present.
The algorithm works by repeatedly subtracting a frequency stretched copy of the spectral slice from itself. The slice is first stretched by a factor of 2 and subtracted, then by a factor of 3 and so on up to a factor of FACTOR (see figure below). A suitable value of FACTOR is usually about 2, 3 or 4. After the multiple subtractions the spectral slice is half-wave rectified to remove any negative components.
|Inputs||Meaning||Sample||1-D frame||2-D frame|
|in1||frames of spectral data||No||Yes||No|
|out1||spectral data after harmonic removal|
|FACTOR||Integer||-||Number of frequency multiples to subtract|