delta

Functions

int main(int argc, char *argv[])

delta [ option ] [ infile ]

-l int
- length of vector \((1 \le M + 1)\)
-m int
- order of vector \((0 \le M)\)
-d double+
- delta coefficients
-D str
- filename of double-type delta coefficients
-r int+
- width of 1st (and 2nd) regression coefficients
-magic double
- magic number
infile str
- double-type static feature vectors
stdout
- double-type static and dynamic feature vectors

The below examples calculate the first and second order dynamic features from 15-dimensional coefficient vectors in data.d.

delta -l 15 -d -0.5 0.0 0.5 -d 1.0 -2.0 1.0 < data.d > data.delta

This is equivalent to

echo -0.5 0.0 0.5 | x2x +ad > delta.win
echo 1.0 -2.0 1.0 | x2x +ad > accel.win
delta -l 15 -D delta.win -D accel.win < data.d > data.delta

If data contains a special number such as an unvoiced symbol in a sequence of fundamental frequencies, use -magic option:

delta -l 15 -D delta.win -magic -1e+10 < data.lf0 > data.lf0.delta

-r option specifies the width of regression coefficients, \(L^{(1)}\) and \(L^{(2)}\). The first and second derivatives are then calculated as follows:

\[\begin{split}\begin{eqnarray} \Delta^{(1)} x_t &=& \frac{\displaystyle\sum_{\tau=-L^{(1)}}^{L^{(1)}} \tau \, x_{t+\tau}} {\displaystyle\sum_{\tau=-L^{(1)}}^{L^{(1)}} \tau^2}, \\ \Delta^{(2)} x_t &=& \frac{\displaystyle\sum_{\tau=-L^{(2)}}^{L^{(2)}} (a_0 \tau^2 - a_1) x_{t+\tau}} {2 \cdot (a_2 a_0 - a_1^2)}, \end{eqnarray}\end{split}\]

where

\[\begin{split}\begin{eqnarray} a_0 &=& \displaystyle\sum_{\tau=-L^{(2)}}^{L^{(2)}} 1, \\ a_1 &=& \displaystyle\sum_{\tau=-L^{(2)}}^{L^{(2)}} \tau^2, \\ a_2 &=& \displaystyle\sum_{\tau=-L^{(2)}}^{L^{(2)}} \tau^4. \end{eqnarray}\end{split}\]

Parameters:

argc – [in] Number of arguments.
argv – [in] Argument vector.

Returns:

0 on success, 1 on failure.