Delta

The Delta block performs temporal differentiation of the input signal.

The differentiated signal, d, is computed from the input signal, c, using the following regression formula:

where d(t) is a delta value at time t computed in terms of the corresponding input values c(t-N) to c(t+N). The value of N is set using the parameter DELTA_WIN, and defaults to 2.

Since computation of the delta values relies on a window of data around the current input frame, some modification is needed at the very beginning and end of the signal. In this implementation for t<N d(t)=d(N) and the signal will be truncated by N frames at the end.

Inputs	Meaning	Sample	1-D frame	2-D frame
in1	input data	Yes	Yes	Yes

Outputs	Meaning
out1	data after differentiation

Parameters	Type	Default	Meaning
DELTA_WIN	Integer	2	Half size of the regression window