tklb::ConvolverBruteTpl< T > Class Template Reference

Brute force convolver using element wise multiplication. More...

#include <TConvolverBrute.hpp>

Collaboration diagram for tklb::ConvolverBruteTpl< T >:

Public Member Functions
	ConvolverBruteTpl ()=default
template<typename T2 >
void	load (const AudioBufferTpl< T2 > &ir, const uint blockSize)
	Load a impulse response and prepare the convolution. More...
template<typename T2 >
void	process (const AudioBufferTpl< T2 > &in, AudioBufferTpl< T2 > &out)
	Do the convolution. More...

Private Types
using	uchar = unsigned char
using	uint = unsigned int
using	Buffer = AudioBufferTpl< T >
using	Size = typename Buffer::Size

Private Attributes
Buffer	mIr

Detailed Description

template<typename T>
class tklb::ConvolverBruteTpl< T >

Brute force convolver using element wise multiplication.

Super slow especially for longer impulse responses

Definition at line 15 of file TConvolverBrute.hpp.

Member Typedef Documentation

Buffer

template<typename T >

using tklb::ConvolverBruteTpl< T >::Buffer = AudioBufferTpl<T>

private

Definition at line 18 of file TConvolverBrute.hpp.

Size

template<typename T >

using tklb::ConvolverBruteTpl< T >::Size = typename Buffer::Size

private

Definition at line 19 of file TConvolverBrute.hpp.

uchar

template<typename T >

using tklb::ConvolverBruteTpl< T >::uchar = unsigned char

private

Definition at line 16 of file TConvolverBrute.hpp.

uint

template<typename T >

using tklb::ConvolverBruteTpl< T >::uint = unsigned int

private

Definition at line 17 of file TConvolverBrute.hpp.

Constructor & Destructor Documentation

ConvolverBruteTpl()

template<typename T >

tklb::ConvolverBruteTpl< T >::ConvolverBruteTpl ( )

default

Member Function Documentation

load()

template<typename T >

template<typename T2 >

void tklb::ConvolverBruteTpl< T >::load ( const AudioBufferTpl< T2 > &  ir, const uint  blockSize  )

inline

Load a impulse response and prepare the convolution.

Parameters

buffer	The ir buffer.
channel	Which channel to use from the AudioBuffer
blockSize	Size of blocks ir will be divided in

Definition at line 32 of file TConvolverBrute.hpp.

                                                                      {
            // trim silence, since longer IRs increase CPU usage considerably
            Size irLength = ir.size();
            if (irLength == 0) { return; }
            const T2 silence = 0.0001;
 
            for (Size i = irLength - 1; 0 < i; i--) {
                for (uchar c = 0; c < ir.channels(); c++) {
                    if (silence < fabs(ir[c][i])) {
                        // if any of the channels are over the
                        // silence threshold we use the current
                        // size and bail
                        irLength = i + 1;
                        goto endLoop;
                    }
                }
            }
        endLoop:
 
            mIr.resize(irLength, ir.channels());
            mIr.set(ir);
        }

Here is the call graph for this function:

process()

template<typename T >

template<typename T2 >

void tklb::ConvolverBruteTpl< T >::process ( const AudioBufferTpl< T2 > &  in, AudioBufferTpl< T2 > &  out  )

inline

Do the convolution.

Parameters

in	Input signal, can be mono
out	Output buffer, needs to have enough space allocated

Definition at line 61 of file TConvolverBrute.hpp.

                                                                            {
            const Size nf = mIr.size();
            const Size ng = in.validSize();
            const Size n = out.size();
 
            for (uchar c = 0; c < out.channels(); c++) {
                // eg the input is mono, but the IR stereo
                // the result will still be stereo
                const uchar inChannel = c % in.channels();
 
                // eg the ir is mono, it'll be used
                // for all channels
                const uchar irChannel = c % mIr.channels();
                // TODO tklb check for underflows
                for(Size i = 0; i < n; i++) {
                    const Size jmn = (i >= ng - 1) ? (i - (ng - 1)) : 0;
                    const Size jmx = (i <  nf - 1) ?  i : (nf - 1);
                    for(Size j = jmn; j <= jmx; j++) {
                        // nested loop goes brr
                        out[c][i] += (mIr[irChannel][j] * in[inChannel][i - j]);
                    }
                }
            }
        }

Here is the call graph for this function:

Member Data Documentation

mIr

template<typename T >

Buffer tklb::ConvolverBruteTpl< T >::mIr

private

Definition at line 20 of file TConvolverBrute.hpp.

---

The documentation for this class was generated from the following file:

• TConvolverBrute.hpp