c_cpp 音频混音器：缩混5.1到立体声或重映射通道
Posted
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了c_cpp 音频混音器：缩混5.1到立体声或重映射通道相关的知识，希望对你有一定的参考价值。
#include "common.h"
#include "mixer.h"
#include <cassert> // for assert
#include <cstdint> // for uint32_t
#include <vector> // for std::vector
using std::vector;
const float M = 1.0f; // Mono
const float L = 2.0f; // Left
const float R = 3.0f; // Right
const float C = 4.0f; // Center
const float LS = 5.0f; // Left Surround
const float RS = 6.0f; // Right Surround
const float RLS = 7.0f; // Rear Left Surround
const float RC = 8.0f; // Rear Center
const float RRS = 9.0f; // Rear Right Surround
const float LFE = 10.0f; // Low Frequency Effects
const float INV_SQRT_2 = 0.707106f; // 1/sqrt(2)
const float DOWNMIX_3F2_RESULTS[2][12][5] = {
// 3F2
{
{ INV_SQRT_2*(L+R) + C + 0.5f*(LS+RS) }, // Mono
{ INV_SQRT_2*(L+R) + C + 0.5f*(LS+RS), 0 }, // Mono-LFE
{ L + INV_SQRT_2*(C+LS), R + INV_SQRT_2*(C+RS) }, // Stereo
{ L + INV_SQRT_2*(C+LS), R + INV_SQRT_2*(C+RS), 0 }, // Stereo-LFE
{ L + INV_SQRT_2*LS, R + INV_SQRT_2*RS, C }, // 3F
{ L + INV_SQRT_2*LS, R + INV_SQRT_2*RS, C, 0 }, // 3F-LFE
{ L + C*INV_SQRT_2, R + C*INV_SQRT_2, INV_SQRT_2*(LS+RS) }, // 2F1
{ L + C*INV_SQRT_2, R + C*INV_SQRT_2, 0, INV_SQRT_2*(LS+RS) }, // 2F1-LFE
{ L, R, C, INV_SQRT_2*(LS+RS) }, // 3F1
{ L, R, C, 0, INV_SQRT_2*(LS+RS) }, // 3F1-LFE
{ L + INV_SQRT_2*C, R + INV_SQRT_2*C, LS, RS }, // 2F2
{ L + INV_SQRT_2*C, R + INV_SQRT_2*C, 0, LS, RS } // 2F2-LFE
},
// 3F2-LFE
{
{ INV_SQRT_2*(L+R) + C + 0.5f*(LS+RS) }, // Mono
{ INV_SQRT_2*(L+R) + C + 0.5f*(LS+RS), LFE }, // Mono-LFE
{ L + INV_SQRT_2*(C+LS), R + INV_SQRT_2*(C+RS) }, // Stereo
{ L + INV_SQRT_2*(C+LS), R + INV_SQRT_2*(C+RS), LFE }, // Stereo-LFE
{ L + INV_SQRT_2*LS, R + INV_SQRT_2*RS, C }, // 3F
{ L + INV_SQRT_2*LS, R + INV_SQRT_2*RS, C, LFE }, // 3F-LFE
{ L + C*INV_SQRT_2, R + C*INV_SQRT_2, INV_SQRT_2*(LS+RS) }, // 2F1
{ L + C*INV_SQRT_2, R + C*INV_SQRT_2, LFE, INV_SQRT_2*(LS+RS) }, // 2F1-LFE
{ L, R, C, INV_SQRT_2*(LS+RS) }, // 3F1
{ L, R, C, LFE, INV_SQRT_2*(LS+RS) }, // 3F1-LFE
{ L + INV_SQRT_2*C, R + INV_SQRT_2*C, LS, RS }, // 2F2
{ L + INV_SQRT_2*C, R + INV_SQRT_2*C, LFE, LS, RS } // 2F2-LFE
}
};
typedef struct {
const LAYOUT layout;
float data[CHANNEL_MAX];
} audio_input;
audio_input audio_inputs[SMPTE_MAX] = {
{ SMPTE_DUAL_MONO, { L, R } },
{ SMPTE_DUAL_MONO_LFE, { L, R, LFE } },
{ SMPTE_MONO, { M } },
{ SMPTE_MONO_LFE, { M, LFE } },
{ SMPTE_STEREO, { L, R } },
{ SMPTE_STEREO_LFE, { L, R, LFE } },
{ SMPTE_3F, { L, R, C } },
{ SMPTE_3F_LFE, { L, R, C, LFE } },
{ SMPTE_2F1, { L, R, RC } },
{ SMPTE_2F1_LFE, { L, R, LFE, RC } },
{ SMPTE_3F1, { L, R, C, RC } },
{ SMPTE_3F1_LFE, { L, R, C, LFE, RC } },
{ SMPTE_2F2, { L, R, LS, RS } },
{ SMPTE_2F2_LFE, { L, R, LFE, LS, RS } },
{ SMPTE_3F2, { L, R, C, LS, RS } },
{ SMPTE_3F2_LFE, { L, R, C, LFE, LS, RS } },
{ SMPTE_3F3R_LFE, { L, R, C, LFE, RC, LS, RS } },
{ SMPTE_3F4_LFE, { L, R, C, LFE, RLS, RRS, LS, RS } },
};
void downmix_test(const float* data, LAYOUT in_layout, LAYOUT out_layout) {
LOG("\nDownmix from %s to %s\n", CHANNEL_LAYOUT_MAPS[in_layout].name, CHANNEL_LAYOUT_MAPS[out_layout].name);
const unsigned long inframes = 10;
const unsigned int in_channels = CHANNEL_LAYOUT_MAPS[in_layout].channels;
const unsigned int out_channels = CHANNEL_LAYOUT_MAPS[out_layout].channels;
vector<float> in(in_channels * inframes);
vector<float> out(out_channels * inframes);
// Generate input data
for (unsigned int offset = 0 ; offset < inframes * in_channels ; offset += in_channels) {
for (unsigned int i = 0 ; i < in_channels ; ++i) {
in[offset + i] = data[i];
}
}
if (!downmix_float(in.data(), inframes, out.data(), in_channels, out_channels, in_layout, out_layout)) {
LOG("Invalid downmix\n");
return;
}
uint32_t in_layout_mask = 0;
for (unsigned int i = 0 ; i < in_channels ; ++i) {
in_layout_mask |= 1 << CHANNEL_LAYOUT_MAPS[in_layout].channel_order[i];
}
uint32_t out_layout_mask = 0;
for (unsigned int i = 0 ; i < out_channels ; ++i) {
out_layout_mask |= 1 << CHANNEL_LAYOUT_MAPS[out_layout].channel_order[i];
}
for (unsigned int i = 0 ; i < inframes * out_channels ; ++i) {
unsigned int index = i % out_channels;
// downmix_3f2
if ((in_layout == SMPTE_3F2 || in_layout == SMPTE_3F2_LFE) &&
out_layout >= SMPTE_MONO && out_layout <= SMPTE_2F2_LFE) {
const float* downmix_results = DOWNMIX_3F2_RESULTS[in_layout - SMPTE_3F2][out_layout - SMPTE_MONO];
LOG("[3f2] Expect: %lf, Get: %lf\n", downmix_results[index], out[index]);
assert(out[index] == downmix_results[index]);
continue;
}
// mix_remap
if (out_layout_mask & in_layout_mask) {
uint32_t mask = 1 << CHANNEL_LAYOUT_MAPS[out_layout].channel_order[index];
LOG("[remap channels] Expect: %lf, Get: %lf\n", (mask & in_layout_mask) ? audio_inputs[out_layout].data[index] : 0, out[index]);
assert(out[index] == ((mask & in_layout_mask) ? audio_inputs[out_layout].data[index] : 0));
continue;
}
// downmix_fallback
LOG("[fallback] Expect: %lf, Get: %lf\n", audio_inputs[in_layout].data[index], out[index]);
assert(out[index] == audio_inputs[in_layout].data[index]);
}
}
int main() {
for (int i = 0 ; i < ARRAY_LENGTH(audio_inputs) ; ++i) { // input
for (int j = 0 ; j < ARRAY_LENGTH(CHANNEL_LAYOUT_MAPS) ; ++j) { // output
downmix_test(audio_inputs[i].data, audio_inputs[i].layout, CHANNEL_LAYOUT_MAPS[j].layout);
}
}
return 0;
}
#ifndef MIXER
#define MIXER
/*
* Convert SMPTE 3F2(-LFE) input audio to various output data with different layout.
*
* SMPTE channel layouts:
* --------------------------------------------------
* DUAL-MONO L R
* DUAL-MONO-LFE L R LFE
* MONO M
* MONO-LFE M LFE
* STEREO L R
* STEREO-LFE L R LFE
* 3F L R C
* 3F-LFE L R C LFE
* 2F1 L R S
* 2F1-LFE L R LFE S
* 3F1 L R C S
* 3F1-LFE L R C LFE S
* 2F2 L R LS RS
* 2F2-LFE L R LFE LS RS
* 3F2 L R C LS RS
* 3F2-LFE L R C LFE LS RS
* 3F3R-LFE L R C LFE BC LS RS
* 3F4-LFE L R C LFE RLS RRS LS RS
*/
#ifdef __cplusplus
extern "C" {
#endif
enum CHANNEL {
CHANNEL_MONO = 0,
CHANNEL_LEFT,
CHANNEL_RIGHT,
CHANNEL_CENTER,
CHANNEL_LS,
CHANNEL_RS,
CHANNEL_RLS,
CHANNEL_RCENTER,
CHANNEL_RRS,
CHANNEL_LFE,
CHANNEL_MAX // Max number of supported channels.
};
enum LAYOUT {
SMPTE_DUAL_MONO = 0,
SMPTE_DUAL_MONO_LFE,
SMPTE_MONO,
SMPTE_MONO_LFE,
SMPTE_STEREO,
SMPTE_STEREO_LFE,
SMPTE_3F,
SMPTE_3F_LFE,
SMPTE_2F1,
SMPTE_2F1_LFE,
SMPTE_3F1,
SMPTE_3F1_LFE,
SMPTE_2F2,
SMPTE_2F2_LFE,
SMPTE_3F2,
SMPTE_3F2_LFE,
SMPTE_3F3R_LFE,
SMPTE_3F4_LFE,
SMPTE_MAX // Max number of supported layouts.
};
typedef struct {
const char* name;
const unsigned int channels;
const LAYOUT layout;
const CHANNEL channel_order[CHANNEL_MAX];
} layout_map;
const layout_map CHANNEL_LAYOUT_MAPS[SMPTE_MAX] = {
{ "dual mono", 2, SMPTE_DUAL_MONO, { CHANNEL_LEFT, CHANNEL_RIGHT } },
{ "dual mono lfe", 3, SMPTE_DUAL_MONO_LFE, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_LFE } },
{ "mono", 1, SMPTE_MONO, { CHANNEL_MONO } },
{ "mono lfe", 2, SMPTE_MONO_LFE, { CHANNEL_MONO, CHANNEL_LFE } },
{ "stereo", 2, SMPTE_STEREO, { CHANNEL_LEFT, CHANNEL_RIGHT } },
{ "stereo lfe", 3, SMPTE_STEREO_LFE, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_LFE } },
{ "3f", 3, SMPTE_3F, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_CENTER } },
{ "3f lfe", 4, SMPTE_3F_LFE, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_CENTER, CHANNEL_LFE } },
{ "2f1", 3, SMPTE_2F1, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_RCENTER } },
{ "2f1 lfe", 4, SMPTE_2F1_LFE, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_LFE, CHANNEL_RCENTER } },
{ "3f1", 4, SMPTE_3F1, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_CENTER, CHANNEL_RCENTER } },
{ "3f1 lfe", 5, SMPTE_3F1_LFE, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_CENTER, CHANNEL_LFE, CHANNEL_RCENTER } },
{ "2f2", 4, SMPTE_2F2, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_LS, CHANNEL_RS } },
{ "2f2 lfe", 5, SMPTE_2F2_LFE, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_LFE, CHANNEL_LS, CHANNEL_RS } },
{ "3f2", 5, SMPTE_3F2, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_CENTER, CHANNEL_LS, CHANNEL_RS } },
{ "3f2 lfe", 6, SMPTE_3F2_LFE, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_CENTER, CHANNEL_LFE, CHANNEL_LS, CHANNEL_RS } },
{ "3f3r lfe", 7, SMPTE_3F3R_LFE, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_CENTER, CHANNEL_LFE, CHANNEL_RCENTER, CHANNEL_LS, CHANNEL_RS } },
{ "3f4 lfe", 8, SMPTE_3F4_LFE, { CHANNEL_LEFT, CHANNEL_RIGHT, CHANNEL_CENTER, CHANNEL_LFE, CHANNEL_RLS, CHANNEL_RRS, CHANNEL_LS, CHANNEL_RS } },
};
bool downmix_float(const float* const in, unsigned long inframes, float* out,
unsigned int in_channels, unsigned int out_channels,
LAYOUT in_layout, LAYOUT out_layout);
#ifdef __cplusplus
}
#endif
#endif // MIXER
#include "mixer.h"
#include <cassert> // for assert
#include <cstdint> // for uint32_t
const int CHANNEL_ORDERING[SMPTE_MAX][CHANNEL_MAX] = {
// M | L | R | C | LS | RS | RLS | RC | RRS | LFE
{ -1, 0, 1, -1, -1, -1, -1, -1, -1, -1 }, // DUAL_MONO
{ -1, 0, 1, -1, -1, -1, -1, -1, -1, 2 }, // DUAL_MONO_LFE
{ 0, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // MONO
{ 0, -1, -1, -1, -1, -1, -1, -1, -1, 1 }, // MONO_LFE
{ -1, 0, 1, -1, -1, -1, -1, -1, -1, -1 }, // STEREO
{ -1, 0, 1, -1, -1, -1, -1, -1, -1, 2 }, // STEREO_LFE
{ -1, 0, 1, 2, -1, -1, -1, -1, -1, -1 }, // 3F
{ -1, 0, 1, 2, -1, -1, -1, -1, -1, 3 }, // 3F_LFE
{ -1, 0, 1, -1, -1, -1, -1, 2, -1, -1 }, // 2F1
{ -1, 0, 1, -1, -1, -1, -1, 3, -1, 2 }, // 2F1_LFE
{ -1, 0, 1, 2, -1, -1, -1, 3, -1, -1 }, // 3F1
{ -1, 0, 1, 2, -1, -1, -1, 4, -1, 3 }, // 3F1_LFE
{ -1, 0, 1, -1, 2, 3, -1, -1, -1, -1 }, // 2F2
{ -1, 0, 1, -1, 3, 4, -1, -1, -1, 2 }, // 2F2_LFE
{ -1, 0, 1, 2, 3, 4, -1, -1, -1, -1 }, // 3F2
{ -1, 0, 1, 2, 4, 5, -1, -1, -1, 3 }, // 3F2_LFE
{ -1, 0, 1, 2, 5, 6, -1, 4, -1, 3 }, // 3F3R_LFE
{ -1, 0, 1, 2, 6, 7, 4, -1, 5, 3 }, // 3F4_LFE
};
// The downmix matrix from TABLE 2 in the ITU-R BS.775-3[1] defines a way to
// convert 3F2 input data to 1F, 2F, 3F, 2F1, 3F1, 2F2 output data. We extend it
// to convert 3F2-LFE input data to 1F, 2F, 3F, 2F1, 3F1, 2F2 and their LFEs
// output data.
// [1] https://www.itu.int/dms_pubrec/itu-r/rec/bs/R-REC-BS.775-3-201208-I!!PDF-E.pdf
// Number of converted layouts: 1F, 2F, 3F, 2F1, 3F1, 2F2 and their LFEs.
const int SUPPORTED_LAYOUT_NUM = 12;
// Number of input channel for downmix conversion.
const int INPUT_CHANNEL_NUM = 6; // 3F2-LFE
// Max number of possible output channels.
const int MAX_OUTPUT_CHANNEL_NUM = 5; // 2F2-LFE or 3F1-LFE
const float INV_SQRT_2 = 0.707106f; // 1/sqrt(2)
// Each array contains coefficients that will be multiplied with { L, R, C, LFE, LS, RS } channels respectively.
const float DOWNMIX_MATRIX_3F2_LFE[SUPPORTED_LAYOUT_NUM][MAX_OUTPUT_CHANNEL_NUM][INPUT_CHANNEL_NUM] =
{
// 1F Mono
{
{ INV_SQRT_2, INV_SQRT_2, 1, 0, 0.5, 0.5 } // M
},
// 1F Mono-LFE
{
{ INV_SQRT_2, INV_SQRT_2, 1, 0, 0.5, 0.5 }, // M
{ 0, 0, 0, 1, 0, 0 } // LFE
},
// 2F Stereo
{
{ 1, 0, INV_SQRT_2, 0, INV_SQRT_2, 0 }, // L
{ 0, 1, INV_SQRT_2, 0, 0, INV_SQRT_2 } // R
},
// 2F Stereo-LFE
{
{ 1, 0, INV_SQRT_2, 0, INV_SQRT_2, 0 }, // L
{ 0, 1, INV_SQRT_2, 0, 0, INV_SQRT_2 }, // R
{ 0, 0, 0, 1, 0, 0 } // LFE
},
// 3F
{
{ 1, 0, 0, 0, INV_SQRT_2, 0 }, // L
{ 0, 1, 0, 0, 0, INV_SQRT_2 }, // R
{ 0, 0, 1, 0, 0, 0 } // C
},
// 3F-LFE
{
{ 1, 0, 0, 0, INV_SQRT_2, 0 }, // L
{ 0, 1, 0, 0, 0, INV_SQRT_2 }, // R
{ 0, 0, 1, 0, 0, 0 }, // C
{ 0, 0, 0, 1, 0, 0 } // LFE
},
// 2F1
{
{ 1, 0, INV_SQRT_2, 0, 0, 0 }, // L
{ 0, 1, INV_SQRT_2, 0, 0, 0 }, // R
{ 0, 0, 0, 0, INV_SQRT_2, INV_SQRT_2 } // S
},
// 2F1-LFE
{
{ 1, 0, INV_SQRT_2, 0, 0, 0 }, // L
{ 0, 1, INV_SQRT_2, 0, 0, 0 }, // R
{ 0, 0, 0, 1, 0, 0 }, // LFE
{ 0, 0, 0, 0, INV_SQRT_2, INV_SQRT_2 } // S
},
// 3F1
{
{ 1, 0, 0, 0, 0, 0 }, // L
{ 0, 1, 0, 0, 0, 0 }, // R
{ 0, 0, 1, 0, 0, 0 }, // C
{ 0, 0, 0, 0, INV_SQRT_2, INV_SQRT_2 } // S
},
// 3F1-LFE
{
{ 1, 0, 0, 0, 0, 0 }, // L
{ 0, 1, 0, 0, 0, 0 }, // R
{ 0, 0, 1, 0, 0, 0 }, // C
{ 0, 0, 0, 1, 0, 0 }, // LFE
{ 0, 0, 0, 0, INV_SQRT_2, INV_SQRT_2 } // S
},
// 2F2
{
{ 1, 0, INV_SQRT_2, 0, 0, 0 }, // L
{ 0, 1, INV_SQRT_2, 0, 0, 0 }, // R
{ 0, 0, 0, 0, 1, 0 }, // LS
{ 0, 0, 0, 0, 0, 1 } // RS
},
// 2F2-LFE
{
{ 1, 0, INV_SQRT_2, 0, 0, 0 }, // L
{ 0, 1, INV_SQRT_2, 0, 0, 0 }, // R
{ 0, 0, 0, 1, 0, 0 }, // LFE
{ 0, 0, 0, 0, 1, 0 }, // LS
{ 0, 0, 0, 0, 0, 1 } // RS
}
};
// Convert data from 3F2 or 3F2-LFE to 1F, 2F, 3F, 2F1, 3F1, 2F2 and their LFEs.
template<class T>
bool downmix_3f2(const T* const in, unsigned long inframes, T* out, LAYOUT in_layout, LAYOUT out_layout) {
if ((in_layout != SMPTE_3F2 && in_layout != SMPTE_3F2_LFE) ||
out_layout < SMPTE_MONO || out_layout > SMPTE_2F2_LFE) {
return false;
}
unsigned int in_channels = CHANNEL_LAYOUT_MAPS[in_layout].channels;
unsigned int out_channels = CHANNEL_LAYOUT_MAPS[out_layout].channels;
// Conversion from 3F2 to 2F2-LFE or 3F1-LFE is allowed, so we use '<=' instead of '<'.
assert(out_channels <= in_channels);
long out_index = 0;
auto & downmix_matrix = DOWNMIX_MATRIX_3F2_LFE[out_layout - SMPTE_MONO]; // The matrix is started from mono.
for (unsigned long i = 0; i < inframes * in_channels; i += in_channels) {
for (unsigned int j = 0; j < out_channels; ++j) {
out[out_index + j] = 0; // Clear its value.
for (unsigned int k = 0 ; k < INPUT_CHANNEL_NUM ; ++k) {
// 3F2-LFE has 6 channels: L, R, C, LFE, LS, RS, while 3F2 has only 5
// channels: L, R, C, LS, RS. Thus, we need to append 0 to LFE(index 3)
// to simulate a 3F2-LFE data when input layout is 3F2.
T data = (in_layout == SMPTE_3F2_LFE) ? in[i + k] : (k == 3) ? 0 : in[i + ((k < 3) ? k : k - 1)];
out[out_index + j] += downmix_matrix[j][k] * data;
}
}
out_index += out_channels;
}
return true;
}
/* Map the audio data by channel name. */
template<class T>
bool mix_remap(const T* const in, unsigned long inframes, T* out, LAYOUT in_layout, LAYOUT out_layout) {
assert(in_layout != out_layout);
unsigned int in_channels = CHANNEL_LAYOUT_MAPS[in_layout].channels;
unsigned int out_channels = CHANNEL_LAYOUT_MAPS[out_layout].channels;
uint32_t in_layout_mask = 0;
for (unsigned int i = 0 ; i < in_channels ; ++i) {
in_layout_mask |= 1 << CHANNEL_LAYOUT_MAPS[in_layout].channel_order[i];
}
uint32_t out_layout_mask = 0;
for (unsigned int i = 0 ; i < out_channels ; ++i) {
out_layout_mask |= 1 << CHANNEL_LAYOUT_MAPS[out_layout].channel_order[i];
}
// If there is no matched channel, then do nothing.
if (!(out_layout_mask & in_layout_mask)) {
return false;
}
long out_index = 0;
for (unsigned long i = 0; i < inframes * in_channels; i += in_channels) {
for (unsigned int j = 0; j < out_channels; ++j) {
CHANNEL channel = CHANNEL_LAYOUT_MAPS[out_layout].channel_order[j];
uint32_t channel_mask = 1 << channel;
// out[out_index + j] = (in_layout_mask & channel_mask) ? in[ i + CHANNEL_ORDERING[in_layout][channel] ] : 0;
int channel_index = CHANNEL_ORDERING[in_layout][channel];
if (in_layout_mask & channel_mask) {
assert(channel_index != -1);
out[out_index + j] = in[i + channel_index];
} else {
assert(channel_index == -1);
out[out_index + j] = 0;
}
}
out_index += out_channels;
}
return true;
}
/* Drop the extra channels beyond the provided output channels. */
template<class T>
bool downmix_fallback(const T* const in, unsigned long inframes, T* out, unsigned int in_channels, unsigned int out_channels) {
assert(in_channels >= out_channels);
long out_index = 0;
for (unsigned long i = 0; i < inframes * in_channels; i += in_channels) {
for (unsigned int j = 0; j < out_channels; ++j) {
out[out_index + j] = in[i + j];
}
out_index += out_channels;
}
return true;
}
template<class T>
bool downmix(const T* const in, unsigned long inframes, T* out,
unsigned int in_channels, unsigned int out_channels,
LAYOUT in_layout, LAYOUT out_layout) {
if (in_channels < out_channels || in_layout == out_layout) {
return false;
}
if (CHANNEL_LAYOUT_MAPS[in_layout].channels == in_channels &&
CHANNEL_LAYOUT_MAPS[out_layout].channels == out_channels) {
if (downmix_3f2(in, inframes, out, in_layout, out_layout)) {
return true;
}
if (mix_remap(in, inframes, out, in_layout, out_layout)) {
return true;
}
}
return downmix_fallback(in, inframes, out, in_channels, out_channels);
}
bool
downmix_float(const float* const in, unsigned long inframes, float* out,
unsigned int in_channels, unsigned int out_channels,
LAYOUT in_layout, LAYOUT out_layout)
{
return downmix(in, inframes, out, in_channels, out_channels, in_layout, out_layout);
}
CC=g++
CFLAGS=-Wall -std=c++14 -Wc++11-extensions
all: mixer.o
$(CC) $(CFLAGS) test_mixer.cpp mixer.o -o test_mixer
mixer.o: mixer.cpp
# -c: Only run preprocess, compile, and assemble steps
# -g: Generate source-level debug information
$(CC) $(CFLAGS) -c -g mixer.cpp
clean:
rm test_mixer *.o
/*
* Convert SMPTE 3F2 audio data from 5 channels to 1 ~ 4 channel data with
* different layouts.
*
* SMPTE channel layouts:
* --------------------------------------------------
* DUAL-MONO L R
* DUAL-MONO-LFE L R LFE
* MONO M
* MONO-LFE M LFE
* STEREO L R
* STEREO-LFE L R LFE
* 3F L R C
* 3F-LFE L R C LFE
* 2F1 L R S
* 2F1-LFE L R LFE S
* 3F1 L R C S
* 3F1-LFE L R C LFE S
* 2F2 L R LS RS
* 2F2-LFE L R LFE LS RS
* 3F2 L R C LS RS
* 3F2-LFE L R C LFE LS RS
* 3F3R-LFE L R C LFE BC LS RS
* 3F4-LFE L R C LFE Rls Rrs LS RS
*/
#include <stdio.h> // for basic io.
#include <stdlib.h> // for malloc.
#define DEBUG 0 // Set 1 to log the debugging messages.
#define LOG(...) DEBUG && fprintf(stdout, __VA_ARGS__)
// This only can NOT be used for the dynamically allocated array(pointer).
#define ARRAY_SIZE(x) ((int) (sizeof(x) / sizeof(x[0])))
// For input channel data:
const float L = 1;
const float R = 2;
const float C = 3;
const float LS = 4;
const float RS = 5;
#define INPUT_CHANNEL_NUMS 5 // 3F2 has 5 channels.
enum LAYOUT {
SMPTE_MONO = 0,
SMPTE_STEREO,
SMPTE_3F,
SMPTE_2F1,
SMPTE_3F1,
SMPTE_2F2,
SMPTE_MAX, // Max number of supported layouts.
};
#define MAX_OUTPUT_CHANNEL_NUMS 4 // Max number of channel supported for the layout.
typedef struct {
char* name;
unsigned int channels;
enum LAYOUT layout;
} layout_map;
static const layout_map CHANNEL_LAYOUT_MAPS[] = {
{ "mono", 1, SMPTE_MONO },
{ "stereo", 2, SMPTE_STEREO },
{ "3f", 3, SMPTE_3F },
{ "2f1", 3, SMPTE_2F1 },
{ "3f1", 4, SMPTE_3F1 },
{ "2f2", 4, SMPTE_2F2 },
};
// Constant for downmix coefficients matrix.
const float SQRT_1_2 = 0.70710678118; // 1/sqrt(2)
// The following matric is refered from TABLE 2 in the ITU-R BS.775-3.
// https://www.itu.int/dms_pubrec/itu-r/rec/bs/R-REC-BS.775-3-201208-I!!PDF-E.pdf
//
// Each array contains coefficients that will be multiplied with
// { L, R, C, LS, RS } channels respectively.
const float DOWNMIX_MATRIX[SMPTE_MAX][MAX_OUTPUT_CHANNEL_NUMS][INPUT_CHANNEL_NUMS] =
{
// 1F Mono
{
{ SQRT_1_2, SQRT_1_2, 1, 0.5, 0.5 }, // M
},
// 2F Stereo
{
{ 1, 0, SQRT_1_2, SQRT_1_2, 0 }, // L
{ 0, 1, SQRT_1_2, 0, SQRT_1_2 } // R
},
// 3F
{
{ 1, 0, 0, SQRT_1_2, 0 }, // L
{ 0, 1, 0, 0, SQRT_1_2 }, // R
{ 0, 0, 1, 0, 0 } // C
},
// 2F1
{
{ 1, 0, SQRT_1_2, 0, 0 }, // L
{ 0, 1, SQRT_1_2, 0, 0 }, // R
{ 0, 0, 0, SQRT_1_2, SQRT_1_2 } // S
},
// 3F1
{
{ 1, 0, 0, 0, 0 }, // L
{ 0, 1, 0, 0, 0 }, // R
{ 0, 0, 1, 0, 0 }, // C
{ 0, 0, 0, SQRT_1_2, SQRT_1_2 }, // S
},
// 2F2
{
{ 1, 0, SQRT_1_2, 0, 0 }, // L
{ 0, 1, SQRT_1_2, 0, 0 }, // R
{ 0, 0, 0, 1, 0 }, // LS
{ 0, 0, 0, 0, 1 }, // RS
},
};
void downmix(float* in, float* out, unsigned const int out_ch_num, unsigned const int out_type) {
for (unsigned int i = 0 ; i < out_ch_num ; ++i) {
out[i] = 0; // Clear its value.
for (unsigned int j = 0 ; j < INPUT_CHANNEL_NUMS ; ++j) {
out[i] += DOWNMIX_MATRIX[out_type][i][j] * in[j];
}
}
}
int main() {
float input[INPUT_CHANNEL_NUMS] = { L, R, C, LS, RS }; // 3F2 input data.
// Downmix for each layout.
for (int i = 0 ; i < ARRAY_SIZE(CHANNEL_LAYOUT_MAPS) ; ++i) {
LOG("\n%s\n", CHANNEL_LAYOUT_MAPS[i].name);
float* output = (float*) malloc(sizeof(float) * CHANNEL_LAYOUT_MAPS[i].channels);
downmix(input, output, CHANNEL_LAYOUT_MAPS[i].channels, CHANNEL_LAYOUT_MAPS[i].layout);
for (int j = 0 ; j < CHANNEL_LAYOUT_MAPS[i].channels ; ++j) {
LOG("ch[%d] = %f\n", j, output[j]);
}
free(output);
}
return 0;
}
#ifndef COMMON
#define COMMON
#include <iostream>
#include <unistd.h>
#define DEBUG false // Set true to log the debugging messages.
#define LOG(...) DEBUG && fprintf(stdout, __VA_ARGS__)
template<typename T, size_t N>
constexpr size_t
ARRAY_LENGTH(T(&)[N])
{
return N;
}
#endif /* COMMON */
以上是关于c_cpp 音频混音器：缩混5.1到立体声或重映射通道的主要内容，如果未能解决你的问题，请参考以下文章
音频处理使用 Adobe Audition 录制电脑内部声音 ( 启用电脑立体声混音 | Adobe Audition 中设置音频设备 | Adobe Audition 内录 )
windows编程：在没有支持“立体声混音”的驱动程序的情况下抓取音频