rlevdur

Functions

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

rlevdur [ option ] [ infile ]

• -m int

  • order of coefficients $(0\le M)$

• infile str

  • double-type linear predictive coefficients

• stdout

  • double-type autocorrelation

The below example converts LPC coefficients in data.lpc to CSM parameters.

rlevdur -m 10 < data.lpc | acr2csm -m 10 > data.csm

Parameters:

• argc – [in] Number of arguments.

• argv – [in] Argument vector.

Returns:

0 on success, 1 on failure.

See also

levdur

class ReverseLevinsonDurbinRecursion

Calculate linear predictive coefficients from autocorrelation.

The input is the $M$-th order LPC coefficients:

$$\begin{array}{cccc}K,& a(1),& \dots ,& a(M),\end{array}$$

where $K$ is the gain, and the output is the $M$-th order autocorrelation:

$$\begin{array}{cccc}r(0),& r(1),& \dots ,& r(M).\end{array}$$

The autocorrelation matrix can be represented as

$$\begin{array}{r}\mathit{R}=\left[\begin{array}{cccc}r(0)& r(1)& \cdots & r(M)\\ r(1)& r(0)& \cdots & r(M-1)\\ ⋮& ⋮& \ddots & ⋮\\ r(M)& r(M-1)& \cdots & r(0)\end{array}\right].\end{array}$$

The autocorrelation is derived by using the matrix decomposition

$${\mathit{R}}^{-1}=\mathit{U}{\mathit{E}}^{-1}{\mathit{U}}^{\mathsf{T}}.$$

The $\mathit{U}$ is the following upper triangular matrix:

$$\begin{array}{r}\mathit{U}=\left[\begin{array}{cccc}{a}^{(0)}(0)& a^{(1)}(1)& \cdots & a^{(M)}(M)\\ 0& a^{(1)}(0)& \cdots & a^{(M)}(M-1)\\ ⋮& ⋮& \ddots & ⋮\\ 0& 0& \cdots & a^{(M)}(0)\end{array}\right],\end{array}$$

where $a^{(i)}(j)$ is the $j$-th coefficient of the $i$-th order prediction filter polynomial. The $\mathit{E}$ is the following diagonal matrix:

$$\begin{array}{r}\mathit{E}=\left[\begin{array}{cccc}{e}^{(0)}(0)& 0& \cdots & 0\\ 0& e^{(1)}(1)& \cdots & 0\\ ⋮& ⋮& \ddots & ⋮\\ 0& 0& \cdots & e^{(M)}(M)\end{array}\right],\end{array}$$

where $e^{(i)}(i)$ is the prediction error from $i$-th order filter. This decomposition allows us the efficient evaluation of the inverse of the autocorrelation matrix.

Public Functions

explicit ReverseLevinsonDurbinRecursion(int num_order)

Parameters:

num_order – [in] Order of coefficients, $M$.

inline int GetNumOrder() const

Returns:

Order of coefficients.

inline bool IsValid() const

Returns:

True if this object is valid.

bool Run(const std::vector<double> &linear_predictive_coefficients, std::vector<double> *autocorrelation, ReverseLevinsonDurbinRecursion::Buffer *buffer) const

Parameters:

• linear_predictive_coefficients – [in] $M$-th order LPC coefficients.

• autocorrelation – [out] $M$-th order autocorrelation.

• buffer – [out] Buffer.

Returns:

True on success, false on failure.

bool Run(std::vector<double> *input_and_output, ReverseLevinsonDurbinRecursion::Buffer *buffer) const

Parameters:

• input_and_output – [inout] $M$-th order coefficients.

• buffer – [out] Buffer.

Returns:

True on success, false on failure.

class Buffer

Buffer for ReverseLevinsonDurbinRecursion class.