Namespace MOSAIC.Models.SignalProcessing
Classes
Crop
Crops (slices) incoming vectors or matrices to a sub-range along the depth and/or channel axis.
CrossCorrelation
Two-channel cross-correlation block. Consumes a Matrix<double> [rows x 2] (col0 = ch1, col1 = ch2) — or, as a fallback, an interleaved Vector<double> (even = ch1, odd = ch2). Pipeline per processed batch: z-normalize → 10-sample moving maximum → subtract a 500-sample forward mean → cross-correlate. Output lags ascend from -MaxLag to +MaxLag (0 selects the full range); the result has L2 norm 10, scaled to an L2 norm of 10 rather than expressed as a Pearson coefficient. Params: MaxLag (0), BufferLen (1000), ProcessEveryN packets (5). Named Lags parameters enable the selectable-channel/filter pipeline and publish [peak lag, sample rate / lag, reserved zero]; see CrossCorrelation.Configuration.cs.
DepthFilter
Digital filter block that applies filtering along the depth (samples) dimension of each frame.
FFT
Computes the FFT magnitude spectrum for each channel using FftFlat (MIT, Ooura).
Filter
Dynamic digital filter block with real-time parameter adjustment.
Function
Applies element-wise mathematical functions to input vectors.
Negate
Flips the sign of every value in the incoming vector. The smallest possible block: one input, one output, no parameters, no card view.
Resampler
Sample-time resampler with zero-order hold and stable output snapshots.
SlidingWindow
Sliding window over a stream of row samples, publishing a windowed
MathNet.Numerics.LinearAlgebra.Matrix<T> of BufferSize × Channels every Stride rows.
Enums
FFT.OutputMode
Spectrum scaling mode.
FFT.WindowType
Window function applied before the FFT.
FilterImplementation
Filter implementation type.
FilterType
Filter type enumeration.
SlidingWindow.WindowType
Window function type.