利用GPAC生成MP4文件
Posted byxdaz
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了利用GPAC生成MP4文件相关的知识，希望对你有一定的参考价值。
GPAC主要针对学生和内容创作者，代表了一个跨平台的多媒体框架，开发人员可以使用它在 LGPL 许可下制作开源媒体。GPAC多媒体框架兼容范围广泛的流行文件类型，从常见格式(如 AVI、MPEG 和 MOV)到复杂格式(如 MPEG-4 系统或 VRML/X3D)和 360 电影。
一、MP4Writer类
MP4Writer.h文件
#ifndef _MP4WRITER_H_
#define _MP4WRITER_H_
#define GPAC_MP4BOX_MINI
#include "gpac/setup.h"
#define MP4_AUDIO_TYPE_INVALID 0
#define MP4_AUDIO_TYPE_AAC_MAIN 1
#define MP4_AUDIO_TYPE_AAC_LC 2
#define MP4_AUDIO_TYPE_AAC_SSR 3
#define MP4_AUDIO_TYPE_AAC_LD 23
#ifdef __cplusplus
extern "C"
#endif
void* MP4_Init();
s32 MP4_CreatFile(void *pCMP4Writer, char *strFileName);
s32 MP4_InitVideo265(void *pCMP4Writer, u32 TimeScale);
s32 MP4_Write265Sample(void *pCMP4Writer, u8 *pData, u32 Size, u64 TimeStamp);
s32 MP4_InitVideo264(void *pCMP4Writer, u32 TimeScale);
s32 MP4_Write264Sample(void *pCMP4Writer, u8 *pData, u32 Size, u64 TimeStamp);
s32 MP4_InitAudioAAC(void *pCMP4Writer, u8 AudioType, u32 SampleRate, u8 Channel, u32 TimeScale);
s32 MP4_WriteAACSample(void *pCMP4Writer, u8 *pData, u32 Size, u64 TimeStamp);
void MP4_CloseFile(void *pCMP4Writer);
void MP4_Exit(void *pCMP4Writer);
#ifdef __cplusplus
#endif
#endifMP4Writer.cpp文件
#include "MP4Writer.h"
#include <Winsock2.h>
extern "C"
#include "gpac/isomedia.h"
#include "gpac/constants.h"
#include "gpac/internal/media_dev.h"
#define INIT_STATUS 0
#define CONFIG_STATUS 1
#define CONFIG_FINISH 2
static s8 GetSampleRateID(u32 SamplRate)
switch (SamplRate)
case 96000: return 0;
case 88200: return 1;
case 64000: return 2;
case 48000: return 3;
case 44100: return 4;
case 32000: return 5;
case 24000: return 6;
case 22050: return 7;
case 16000: return 8;
case 12000: return 9;
case 11025: return 10;
case 8000 : return 11;
case 7350 : return 12;
default: return -1;
//gf_m4a_get_profile
static u8 GetAACProfile(u8 AudioType, u32 SampleRate, u8 Channel)
switch (AudioType)
case 2: /* AAC LC */
if (Channel <= 2) return (SampleRate <= 24000) ? 0x28 : 0x29; /* LC@L1 or LC@L2 */
if (Channel <= 5) return (SampleRate <= 48000) ? 0x2A : 0x2B; /* LC@L4 or LC@L5 */
return (SampleRate <= 48000) ? 0x50 : 0x51; /* LC@L4 or LC@L5 */
case 5: /* HE-AAC - SBR */
if (Channel <= 2) return (SampleRate <= 24000) ? 0x2C : 0x2D; /* HE@L2 or HE@L3 */
if (Channel <= 5) return (SampleRate <= 48000) ? 0x2E : 0x2F; /* HE@L4 or HE@L5 */
return (SampleRate <= 48000) ? 0x52 : 0x53; /* HE@L6 or HE@L7 */
case 29: /*HE-AACv2 - SBR+PS*/
if (Channel <= 2) return (SampleRate <= 24000) ? 0x30 : 0x31; /* HE-AACv2@L2 or HE-AACv2@L3 */
if (Channel <= 5) return (SampleRate <= 48000) ? 0x32 : 0x33; /* HE-AACv2@L4 or HE-AACv2@L5 */
return (SampleRate <= 48000) ? 0x54 : 0x55; /* HE-AACv2@L6 or HE-AACv2@L7 */
default: /* default to HQ */
if (Channel <= 2) return (SampleRate < 24000) ? 0x0E : 0x0F; /* HQ@L1 or HQ@L2 */
return 0x10; /* HQ@L3 */
static void GetAudiospecificConfig(u8 AudioType, u8 SampleRateID, u8 Channel, u8 *pHigh, u8 *pLow)
u16 Config;
Config = (AudioType & 0x1f);
Config <<= 4;
Config |= SampleRateID & 0x0f;
Config <<= 4;
Config |= Channel & 0x0f;
Config <<= 3;
*pLow = Config & 0xff;
Config >>= 8;
*pHigh = Config & 0xff;
/* 返回的数据包括起始的4个字节0x00000001 */
static u8* FindNalu(u8 *pStart, u32 Size, u8 *pNaluType, u32 *pNaluSize)
u8 *pEnd;
u8 *pCur;
u8 *pOut;
u8 NaluType;
if (4 >= Size)
return NULL;
/* 找第一个0x00000001 */
pCur = pStart;
pEnd = pStart + Size - 4;
while (pCur < pEnd)
if ( (0 == pCur[0]) && (0 == pCur[1]) && (0 == pCur[2]) && (1 == pCur[3]) )
break;
pCur++;
if (pCur >= pEnd)
return NULL;
NaluType = (pCur[4] >> 1) & 0x3f;
*pNaluType = NaluType;
if (1 == NaluType || 19 == NaluType) /* P帧、I帧, 假设每一包里P帧I帧都是最后一个 */
*pNaluSize = Size - (pCur - pStart);
return pCur;
pOut = pCur;
/* 找第二个0x00000001 */
pCur += 5;
while (pCur <= pEnd)
if ( (0 == pCur[0]) && (0 == pCur[1]) && (0 == pCur[2]) && (1 == pCur[3]) )
break;
pCur++;
if (pCur <= pEnd)
*pNaluSize = pCur - pOut;
return pOut;
*pNaluSize = Size - (pOut - pStart);
return pOut;
/* 返回的数据包括起始的4个字节0x00000001 */
static u8* FindNalu264(u8 *pStart, u32 Size, u8 *pNaluType, u32 *pNaluSize)
u8 *pEnd;
u8 *pCur;
u8 *pOut;
u8 NaluType;
if (4 >= Size)
return NULL;
/* 找第一个0x00000001 */
pCur = pStart;
pEnd = pStart + Size - 4;
while (pCur < pEnd)
if ((0 == pCur[0]) && (0 == pCur[1]) && (0 == pCur[2]) && (1 == pCur[3]))
break;
pCur++;
if (pCur >= pEnd)
return NULL;
NaluType = (pCur[4]) & 0x1f;
*pNaluType = NaluType;
if (1 == NaluType || 5 == NaluType) /* P帧、I帧, 假设每一包里P帧I帧都是最后一个 */
*pNaluSize = Size - (pCur - pStart);
return pCur;
pOut = pCur;
/* 找第二个0x00000001 */
pCur += 5;
while (pCur <= pEnd)
if ((0 == pCur[0]) && (0 == pCur[1]) && (0 == pCur[2]) && (1 == pCur[3]))
break;
pCur++;
if (pCur <= pEnd)
*pNaluSize = pCur - pOut;
return pOut;
*pNaluSize = Size - (pOut - pStart);
return pOut;
// aac数据时记得除掉adts头
static int AdtsDemux(unsigned char * data, unsigned int size, unsigned char** raw, int* raw_size)
int ret = 1;
if (size < 7)
ttf_human_trace("adts: demux size too small");
return 0;
unsigned char* p = data;
unsigned char* pend = data + size;
//unsigned char* startp = 0;
while (p < pend)
// decode the ADTS.
// @see aac-iso-13818-7.pdf, page 26
// 6.2 Audio Data Transport Stream, ADTS
// @see https://github.com/ossrs/srs/issues/212#issuecomment-64145885
// byte_alignment()
// adts_fixed_header:
// 12bits syncword,
// 16bits left.
// adts_variable_header:
// 28bits
// 12+16+28=56bits
// adts_error_check:
// 16bits if protection_absent
// 56+16=72bits
// if protection_absent:
// require(7bytes)=56bits
// else
// require(9bytes)=72bits
//startp = p;
// for aac, the frame must be ADTS format.
/*if (p[0] != 0xff || (p[1] & 0xf0) != 0xf0)
ttf_human_trace("adts: not this format.");
return 0;
*/
// syncword 12 bslbf
p++;
// 4bits left.
// adts_fixed_header(), 1.A.2.2.1 Fixed Header of ADTS
// ID 1 bslbf
// layer 2 uimsbf
// protection_absent 1 bslbf
int8_t pav = (*p++ & 0x0f);
//int8_t id = (pav >> 3) & 0x01;
/*int8_t layer = (pav >> 1) & 0x03;*/
int8_t protection_absent = pav & 0x01;
/**
* ID: MPEG identifier, set to '1' if the audio data in the ADTS stream are MPEG-2 AAC (See ISO/IEC 13818-7)
* and set to '0' if the audio data are MPEG-4. See also ISO/IEC 11172-3, subclause 2.4.2.3.
*/
//if (id != 0x01)
// //ttf_human_trace("adts: id must be 1(aac), actual 0(mp4a).");
// // well, some system always use 0, but actually is aac format.
// // for example, houjian vod ts always set the aac id to 0, actually 1.
// // we just ignore it, and alwyas use 1(aac) to demux.
// id = 0x01;
//
//else
// //ttf_human_trace("adts: id must be 1(aac), actual 1(mp4a).");
//
//int16_t sfiv = (*p << 8) | (*(p + 1));
p += 2;
// profile 2 uimsbf
// sampling_frequency_index 4 uimsbf
// private_bit 1 bslbf
// channel_configuration 3 uimsbf
// original/copy 1 bslbf
// home 1 bslbf
//int8_t profile = (sfiv >> 14) & 0x03;
//int8_t sampling_frequency_index = (sfiv >> 10) & 0x0f;
/*int8_t private_bit = (sfiv >> 9) & 0x01;*/
//int8_t channel_configuration = (sfiv >> 6) & 0x07;
/*int8_t original = (sfiv >> 5) & 0x01;*/
/*int8_t home = (sfiv >> 4) & 0x01;*/
//int8_t Emphasis; @remark, Emphasis is removed, @see https://github.com/ossrs/srs/issues/212#issuecomment-64154736
// 4bits left.
// adts_variable_header(), 1.A.2.2.2 Variable Header of ADTS
// copyright_identification_bit 1 bslbf
// copyright_identification_start 1 bslbf
/*int8_t fh_copyright_identification_bit = (fh1 >> 3) & 0x01;*/
/*int8_t fh_copyright_identification_start = (fh1 >> 2) & 0x01;*/
// frame_length 13 bslbf: Length of the frame including headers and error_check in bytes.
// use the left 2bits as the 13 and 12 bit,
// the frame_length is 13bits, so we move 13-2=11.
//int16_t frame_length = (sfiv << 11) & 0x1800;
//int32_t abfv = ((*p) << 16)
// | ((*(p + 1)) << 8)
// | (*(p + 2));
p += 3;
// frame_length 13 bslbf: consume the first 13-2=11bits
// the fh2 is 24bits, so we move right 24-11=13.
//frame_length |= (abfv >> 13) & 0x07ff;
// adts_buffer_fullness 11 bslbf
/*int16_t fh_adts_buffer_fullness = (abfv >> 2) & 0x7ff;*/
// number_of_raw_data_blocks_in_frame 2 uimsbf
/*int16_t number_of_raw_data_blocks_in_frame = abfv & 0x03;*/
// adts_error_check(), 1.A.2.2.3 Error detection
if (!protection_absent)
if (size < 9)
ttf_human_trace("adts: protection_absent disappare.");
return 0;
// crc_check 16 Rpchof
/*int16_t crc_check = */p += 2;
// TODO: check the sampling_frequency_index
// TODO: check the channel_configuration
// raw_data_blocks
/*int adts_header_size = p - startp;
int raw_data_size = frame_length - adts_header_size;
if (raw_data_size > pend - p)
ttf_human_trace("adts: raw data size too little.");
return 0;
*/
//adts_codec_.protection_absent = protection_absent;
//adts_codec_.aac_object = srs_codec_aac_ts2rtmp((SrsAacProfile)profile);
//adts_codec_.sampling_frequency_index = sampling_frequency_index;
//adts_codec_.channel_configuration = channel_configuration;
//adts_codec_.frame_length = frame_length;
@see srs_audio_write_raw_frame().
TODO: FIXME: maybe need to resample audio.
//adts_codec_.sound_format = 10; // AAC
//if (sampling_frequency_index <= 0x0c && sampling_frequency_index > 0x0a)
// adts_codec_.sound_rate = SrsCodecAudioSampleRate5512;
//
//else if (sampling_frequency_index <= 0x0a && sampling_frequency_index > 0x07)
// adts_codec_.sound_rate = SrsCodecAudioSampleRate11025;
//
//else if (sampling_frequency_index <= 0x07 && sampling_frequency_index > 0x04)
// adts_codec_.sound_rate = SrsCodecAudioSampleRate22050;
//
//else if (sampling_frequency_index <= 0x04)
// adts_codec_.sound_rate = SrsCodecAudioSampleRate44100;
//
//else
// adts_codec_.sound_rate = SrsCodecAudioSampleRate44100;
// //srs_warn("adts invalid sample rate for flv, rate=%#x", sampling_frequency_index);
//
//adts_codec_.sound_type = srs_max(0, srs_min(1, channel_configuration - 1));
TODO: FIXME: finger it out the sound size by adts.
//adts_codec_.sound_size = 1; // 0(8bits) or 1(16bits).
// frame data.
*raw = p;
*raw_size = pend - p;
break;
return ret;
class MP4Writer
public:
MP4Writer();
~MP4Writer();
s32 CreatFile(char *strFileName);
s32 Init265(u32 TimeScale);
s32 Write265Sample(u8 *pData, u32 Size, u64 TimeStamp);
s32 Init264(u32 TimeScale);
s32 Write264Sample(u8 *pData, u32 Size, u64 TimeStamp);
s32 InitAAC(u8 AudioType, u32 SampleRate, u8 Channel, u32 TimeScale);
s32 WriteAACSample(u8 *pData, u32 Size, u64 TimeStamp);
void CloseFile();
private:
GF_ISOFile *m_ptFile;
u32 m_265TrackIndex;
u32 m_265StreamIndex;
u8 m_Video265Statue;
u32 m_264TrackIndex;
u32 m_264StreamIndex;
u8 m_Video264Statue;
s64 m_VideoTimeStampStart;
GF_HEVCConfig *m_ptHEVCConfig;
GF_HEVCParamArray* m_tHEVCNaluParam_VPS;
GF_HEVCParamArray* m_tHEVCNaluParam_SPS;
GF_HEVCParamArray* m_tHEVCNaluParam_PPS;
GF_AVCConfigSlot* m_tAVCConfig_VPS;
GF_AVCConfigSlot* m_tAVCConfig_SPS;
GF_AVCConfigSlot* m_tAVCConfig_PPS;
GF_AVCConfigSlot* m_tAVCConfig_SPS264;
GF_AVCConfigSlot* m_tAVCConfig_PPS264;
GF_AVCConfig *m_ptAVCConfig;
HEVCState *m_ptHEVCState;
AVCState *m_ptAVCState;
u32 m_AACTrackIndex;
u32 m_AACStreamIndex;
u8 m_AudioAACStatue;
s64 m_AudioTimeStampStart;
bool findfirsidr_;
void FreeAllMem();
;
MP4Writer::MP4Writer()
m_ptFile = NULL;
m_ptHEVCConfig = NULL;
m_ptAVCConfig = NULL;
m_Video265Statue = INIT_STATUS;
m_AudioAACStatue = INIT_STATUS;
m_Video264Statue = INIT_STATUS;
m_tHEVCNaluParam_VPS = NULL;
m_tHEVCNaluParam_SPS = NULL;
m_tHEVCNaluParam_PPS = NULL;
m_tAVCConfig_VPS = NULL;
m_tAVCConfig_SPS = NULL;
m_tAVCConfig_PPS = NULL;
m_tAVCConfig_SPS264 = NULL;
m_tAVCConfig_PPS264 = NULL;
m_ptHEVCState = NULL;
m_ptAVCState = NULL;
findfirsidr_ = false;
MP4Writer::~MP4Writer()
CloseFile();
s32 MP4Writer::CreatFile(char *strFileName)
if (NULL != m_ptFile)
return -1;
m_ptFile = gf_isom_open(strFileName, GF_ISOM_OPEN_WRITE, NULL);
if (NULL == m_ptFile)
return -1;
gf_isom_set_brand_info(m_ptFile, GF_ISOM_BRAND_MP42, 0);
return 0;
void MP4Writer::CloseFile()
if (m_ptFile)
gf_isom_close(m_ptFile);
m_ptFile = NULL;
FreeAllMem();
void MP4Writer::FreeAllMem()
if (m_ptHEVCConfig)
gf_odf_hevc_cfg_del(m_ptHEVCConfig);
//gf_list_del(m_ptHEVCConfig->param_array);
//free(m_ptHEVCConfig);
m_ptHEVCConfig = NULL;
if (m_ptAVCConfig)
gf_odf_avc_cfg_del(m_ptAVCConfig);
//gf_list_del(m_ptAVCConfig->param_array);
//free(m_ptAVCConfig);
m_ptAVCConfig = NULL;
if (m_ptHEVCState)
free(m_ptHEVCState);
m_ptHEVCState = NULL;
if (m_ptAVCState)
free(m_ptAVCState);
m_ptAVCState = NULL;
m_tHEVCNaluParam_VPS = NULL;
m_tHEVCNaluParam_SPS = NULL;
m_tHEVCNaluParam_PPS = NULL;
m_tAVCConfig_VPS = NULL;
m_tAVCConfig_SPS = NULL;
m_tAVCConfig_PPS = NULL;
m_tAVCConfig_SPS264 = NULL;
m_tAVCConfig_PPS264 = NULL;
m_Video265Statue = INIT_STATUS;
m_AudioAACStatue = INIT_STATUS;
m_Video264Statue = INIT_STATUS;
findfirsidr_ = false;
s32 MP4Writer::Init265(u32 TimeScale)
if (NULL == m_ptFile || INIT_STATUS != m_Video265Statue)
return -1;
m_VideoTimeStampStart = -1;
/* 创建Track */
m_265TrackIndex = gf_isom_new_track(m_ptFile, 0, GF_ISOM_MEDIA_VISUAL, TimeScale);
if (0 == m_265TrackIndex)
return -1;
if (GF_OK != gf_isom_set_track_enabled(m_ptFile, m_265TrackIndex, 1))
return -1;
/* 创建流 */
m_ptHEVCConfig = gf_odf_hevc_cfg_new();
if (NULL == m_ptHEVCConfig)
return -1;
m_ptHEVCConfig->nal_unit_size = 4;
m_ptHEVCConfig->configurationVersion = 1;
if (GF_OK != gf_isom_hevc_config_new(m_ptFile, m_265TrackIndex, m_ptHEVCConfig, NULL, NULL, &m_265StreamIndex))
return -1;
/* 初始化流的配置结构 */
GF_SAFEALLOC(m_tHEVCNaluParam_VPS, GF_HEVCParamArray);
GF_SAFEALLOC(m_tHEVCNaluParam_SPS, GF_HEVCParamArray);
GF_SAFEALLOC(m_tHEVCNaluParam_PPS, GF_HEVCParamArray);
GF_SAFEALLOC(m_tAVCConfig_VPS, GF_AVCConfigSlot);
GF_SAFEALLOC(m_tAVCConfig_SPS, GF_AVCConfigSlot);
GF_SAFEALLOC(m_tAVCConfig_PPS, GF_AVCConfigSlot);
if (GF_OK != gf_list_add(m_ptHEVCConfig->param_array, m_tHEVCNaluParam_VPS))
return -1;
if (GF_OK != gf_list_add(m_ptHEVCConfig->param_array, m_tHEVCNaluParam_SPS))
return -1;
if (GF_OK != gf_list_add(m_ptHEVCConfig->param_array, m_tHEVCNaluParam_PPS))
return -1;
m_tHEVCNaluParam_VPS->nalus = gf_list_new();
if (NULL == m_tHEVCNaluParam_VPS->nalus)
return -1;
m_tHEVCNaluParam_SPS->nalus = gf_list_new();
if (NULL == m_tHEVCNaluParam_SPS->nalus)
return -1;
m_tHEVCNaluParam_PPS->nalus = gf_list_new();
if (NULL == m_tHEVCNaluParam_PPS->nalus)
return -1;
m_tHEVCNaluParam_VPS->type = GF_HEVC_NALU_VID_PARAM;
m_tHEVCNaluParam_SPS->type = GF_HEVC_NALU_SEQ_PARAM;
m_tHEVCNaluParam_PPS->type = GF_HEVC_NALU_PIC_PARAM;
m_tHEVCNaluParam_VPS->array_completeness = 1;
m_tHEVCNaluParam_SPS->array_completeness = 1;
m_tHEVCNaluParam_PPS->array_completeness = 1;
if (GF_OK != gf_list_add(m_tHEVCNaluParam_VPS->nalus, m_tAVCConfig_VPS))
return -1;
if (GF_OK != gf_list_add(m_tHEVCNaluParam_SPS->nalus, m_tAVCConfig_SPS))
return -1;
if (GF_OK != gf_list_add(m_tHEVCNaluParam_PPS->nalus, m_tAVCConfig_PPS))
return -1;
m_ptHEVCState = (HEVCState *)malloc(sizeof(HEVCState));//gf_malloc
if (NULL == m_ptHEVCState)
return -1;
memset(m_ptHEVCState, 0, sizeof(HEVCState));
m_Video265Statue = CONFIG_STATUS;
return 0;
s32 MP4Writer::Write265Sample(u8 *pData, u32 Size, u64 TimeStamp)
u8 *pStart = pData;
u8 NaluType;
u32 NaluSize = 0;
s32 ID;
GF_ISOSample tISOSample;
/* 265还未初始化 */
if (INIT_STATUS == m_Video265Statue)
return -1;
while (1)
pData = FindNalu(pData + NaluSize, Size - (u32)(pData - pStart) - NaluSize, &NaluType, &NaluSize);
if (NULL == pData)
break;
/* 配置完成后只处理IP帧 */
if (CONFIG_FINISH == m_Video265Statue)
if (1 != NaluType && 19 != NaluType) /* P帧 I帧 */
continue;
if (-1 == m_VideoTimeStampStart)
m_VideoTimeStampStart = TimeStamp;
*((u32 *)pData) = htonl(NaluSize - 4); /* 这里的长度不能包括前四个字节的头！ */
tISOSample.data = (char *)pData;
tISOSample.dataLength = NaluSize;
tISOSample.IsRAP = (19 == NaluType)? RAP: RAP_NO;
tISOSample.DTS = TimeStamp - m_VideoTimeStampStart;
tISOSample.CTS_Offset = 0;
tISOSample.nb_pack = 0;
if (GF_OK != gf_isom_add_sample(m_ptFile, m_265TrackIndex, m_265StreamIndex, &tISOSample))
*((u32 *)pData) = htonl(1); /* 恢复0x00000001的头 */
return -1;
*((u32 *)pData) = htonl(1); /* 恢复0x00000001的头 */
/* 配置未完成时只处理vps sps pps头 */
else if (CONFIG_STATUS == m_Video265Statue)
pData += 4;
NaluSize -= 4;
if (32 == NaluType && NULL == m_tAVCConfig_VPS->data) /* VPS */
ID = gf_media_hevc_read_vps((char *)pData , NaluSize, m_ptHEVCState);
m_ptHEVCConfig->avgFrameRate = m_ptHEVCState->vps[ID].rates[0].avg_pic_rate;
m_ptHEVCConfig->temporalIdNested = m_ptHEVCState->vps[ID].temporal_id_nesting;
m_ptHEVCConfig->constantFrameRate = m_ptHEVCState->vps[ID].rates[0].constand_pic_rate_idc;
m_ptHEVCConfig->numTemporalLayers = m_ptHEVCState->vps[ID].max_sub_layers;
m_tAVCConfig_VPS->id = ID;
m_tAVCConfig_VPS->size = (u16)NaluSize;
m_tAVCConfig_VPS->data = (char *)malloc(NaluSize);
if (NULL == m_tAVCConfig_VPS->data)
continue;
memcpy(m_tAVCConfig_VPS->data, pData, NaluSize);
else if (33 == NaluType && NULL == m_tAVCConfig_SPS->data) /* SPS */
ID = gf_media_hevc_read_sps((char *)pData, NaluSize, m_ptHEVCState);
m_ptHEVCConfig->tier_flag = m_ptHEVCState->sps[ID].ptl.tier_flag;
m_ptHEVCConfig->profile_idc = m_ptHEVCState->sps[ID].ptl.profile_idc;
m_ptHEVCConfig->profile_space = m_ptHEVCState->sps[ID].ptl.profile_space;
m_tAVCConfig_SPS->id = ID;
m_tAVCConfig_SPS->size = (u16)NaluSize;
m_tAVCConfig_SPS->data = (char *)malloc(NaluSize);
if (NULL == m_tAVCConfig_SPS->data)
continue;
memcpy(m_tAVCConfig_SPS->data, pData, NaluSize);
gf_isom_set_visual_info(m_ptFile, m_265TrackIndex, m_265StreamIndex, m_ptHEVCState->sps[ID].width, m_ptHEVCState->sps[ID].height);
else if (34 == NaluType && NULL == m_tAVCConfig_PPS->data) /* PPS */
ID = gf_media_hevc_read_pps((char *)pData, NaluSize, m_ptHEVCState);
m_tAVCConfig_PPS->id = ID;
m_tAVCConfig_PPS->size = (u16)NaluSize;
m_tAVCConfig_PPS->data = (char *)malloc(NaluSize);
if (NULL == m_tAVCConfig_PPS->data)
continue;
memcpy(m_tAVCConfig_PPS->data, pData, NaluSize);
else
continue;
if (m_tAVCConfig_VPS->data && m_tAVCConfig_SPS->data && m_tAVCConfig_PPS->data)
gf_isom_hevc_config_update(m_ptFile, m_265TrackIndex, m_265StreamIndex, m_ptHEVCConfig);
m_Video265Statue = CONFIG_FINISH;
if (m_ptHEVCState)
free(m_ptHEVCState);
m_ptHEVCState = NULL;
return 0;
s32 MP4Writer::Init264(u32 TimeScale)
if (NULL == m_ptFile || INIT_STATUS != m_Video265Statue)
return -1;
m_VideoTimeStampStart = -1;
/* 创建Track */
m_264TrackIndex = gf_isom_new_track(m_ptFile, 0, GF_ISOM_MEDIA_VISUAL, TimeScale);
if (0 == m_264TrackIndex)
return -1;
if (GF_OK != gf_isom_set_track_enabled(m_ptFile, m_264TrackIndex, 1))
return -1;
/* 创建流 */
m_ptAVCConfig = gf_odf_avc_cfg_new();
if (NULL == m_ptAVCConfig)
return -1;
//m_ptAVCConfig->nal_unit_size = 4;
m_ptAVCConfig->configurationVersion = 1;
if (GF_OK != gf_isom_avc_config_new(m_ptFile, m_264TrackIndex, m_ptAVCConfig, NULL, NULL, &m_264StreamIndex))
return -1;
/* 初始化流的配置结构 */
GF_SAFEALLOC(m_tAVCConfig_SPS264, GF_AVCConfigSlot);
GF_SAFEALLOC(m_tAVCConfig_PPS264, GF_AVCConfigSlot);
gf_list_add(m_ptAVCConfig->sequenceParameterSets, m_tAVCConfig_SPS264);
gf_list_add(m_ptAVCConfig->pictureParameterSets, m_tAVCConfig_PPS264);
m_ptAVCState = (AVCState *)malloc(sizeof(AVCState));//gf_malloc
if (NULL == m_ptAVCState)
return -1;
memset(m_ptAVCState, 0, sizeof(AVCState));
m_Video264Statue = CONFIG_STATUS;
return 0;
s32 MP4Writer::Write264Sample(u8 *pData, u32 Size, u64 TimeStamp)
u8 *pStart = pData;
u8 NaluType;
u32 NaluSize = 0;
s32 ID;
GF_ISOSample tISOSample;
/* 265还未初始化 */
if (INIT_STATUS == m_Video264Statue)
return -1;
while (1)
pData = FindNalu264(pData + NaluSize, Size - (u32)(pData - pStart) - NaluSize, &NaluType, &NaluSize);
if (NULL == pData)
break;
/* 配置完成后只处理IP帧 */
if (CONFIG_FINISH == m_Video264Statue)
if (!findfirsidr_)
if (5 != NaluType)
continue;
else
findfirsidr_ = true;
if (1 != NaluType && 5 != NaluType) /* P帧 I帧 */
continue;
if (-1 == m_VideoTimeStampStart)
m_VideoTimeStampStart = TimeStamp;
*((u32 *)pData) = htonl(NaluSize - 4); /* 这里的长度不能包括前四个字节的头！ */
tISOSample.data = (char *)pData;
tISOSample.dataLength = NaluSize;
tISOSample.IsRAP = (5 == NaluType) ? RAP : RAP_NO;
tISOSample.DTS = TimeStamp - m_VideoTimeStampStart;
tISOSample.CTS_Offset = 0;
tISOSample.nb_pack = 0;
int ret = gf_isom_add_sample(m_ptFile, m_264TrackIndex, m_264StreamIndex, &tISOSample);
if (GF_OK != ret)
//*((u32 *)pData) = htonl(1); /* 恢复0x00000001的头 */
return -1;
//*((u32 *)pData) = htonl(1); /* 恢复0x00000001的头 */
/* 配置未完成时只处理vps sps pps头 */
else if (CONFIG_STATUS == m_Video264Statue)
pData += 4;
NaluSize -= 4;
if (7 == NaluType && NULL == m_tAVCConfig_SPS264->data) /* SPS */
ID = gf_media_avc_read_sps((char *)pData, NaluSize, m_ptAVCState,0,NULL);
m_ptAVCConfig->AVCProfileIndication = m_ptAVCState->sps[ID].profile_idc;
m_ptAVCConfig->profile_compatibility = m_ptAVCState->sps[ID].prof_compat;
m_ptAVCConfig->AVCLevelIndication = m_ptAVCState->sps[ID].level_idc;
m_tAVCConfig_SPS264->id = ID;
m_tAVCConfig_SPS264->size = (u16)NaluSize;
m_tAVCConfig_SPS264->data = (char *)malloc(NaluSize);
if (NULL == m_tAVCConfig_SPS264->data)
continue;
memcpy(m_tAVCConfig_SPS264->data, pData, NaluSize);
gf_isom_set_visual_info(m_ptFile, m_264TrackIndex, m_264StreamIndex, m_ptAVCState->sps[ID].width, m_ptAVCState->sps[ID].height);
else if (8 == NaluType && NULL == m_tAVCConfig_PPS264->data) /* PPS */
m_tAVCConfig_PPS264->id = ID;
m_tAVCConfig_PPS264->size = (u16)NaluSize;
m_tAVCConfig_PPS264->data = (char *)malloc(NaluSize);
if (NULL == m_tAVCConfig_PPS264->data)
continue;
memcpy(m_tAVCConfig_PPS264->data, pData, NaluSize);
else
continue;
if (m_tAVCConfig_SPS264->data && m_tAVCConfig_PPS264->data)
gf_isom_avc_config_update(m_ptFile, m_264TrackIndex, m_264StreamIndex, m_ptAVCConfig);
m_Video264Statue = CONFIG_FINISH;
if (m_ptAVCState)
free(m_ptAVCState);
m_ptAVCState = NULL;
return 0;
s32 MP4Writer::InitAAC(u8 AudioType, u32 SampleRate, u8 Channel, u32 TimeScale)
GF_ESD *ptStreamDesc;
s8 SampleRateID;
u16 AudioConfig = 0;
u8 AACProfile;
s32 res = 0;
if (NULL == m_ptFile || INIT_STATUS != m_AudioAACStatue)
return -1;
m_AudioTimeStampStart = -1;
/* 创建Track */
m_AACTrackIndex = gf_isom_new_track(m_ptFile, 0, GF_ISOM_MEDIA_AUDIO, TimeScale);
if (0 == m_AACTrackIndex)
return -1;
if (GF_OK != gf_isom_set_track_enabled(m_ptFile, m_AACTrackIndex, 1))
return -1;
/* 创建并配置流 */
SampleRateID = GetSampleRateID(SampleRate);
if (0 > SampleRateID)
return -1;
GetAudioSpecificConfig(AudioType, (u8)SampleRateID, Channel, (u8*)(&AudioConfig), ((u8*)(&AudioConfig))+1);
ptStreamDesc = gf_odf_desc_esd_new(SLPredef_MP4);
ptStreamDesc->slConfig->timestampResolution = TimeScale;
ptStreamDesc->decoderConfig->streamType = GF_STREAM_AUDIO;
//ptStreamDesc->decoderConfig->bufferSizeDB = 20; //这参数干什么的
ptStreamDesc->decoderConfig->objectTypeIndication = 0x40;//ptStreamDesc->decoderConfig->objectTypeIndication = GPAC_OTI_AUDIO_AAC_MPEG4;
ptStreamDesc->decoderConfig->decoderSpecificInfo->dataLength = 2;
ptStreamDesc->decoderConfig->decoderSpecificInfo->data = (char *)&AudioConfig;
ptStreamDesc->ESID = gf_isom_get_track_id(m_ptFile, m_AACTrackIndex);
if (GF_OK != gf_isom_new_mpeg4_description(m_ptFile, m_AACTrackIndex, ptStreamDesc, NULL, NULL, &m_AACStreamIndex))
res = -1;
goto ERR;
if (gf_isom_set_audio_info(m_ptFile, m_AACTrackIndex, m_AACStreamIndex, SampleRate, Channel, 16, GF_IMPORT_AUDIO_SAMPLE_ENTRY_NOT_SET))
res = -1;
goto ERR;
AACProfile = GetAACProfile(AudioType, SampleRate, Channel);
gf_isom_set_pl_indication(m_ptFile, GF_ISOM_PL_AUDIO, AACProfile);
m_AudioAACStatue = CONFIG_FINISH;
ERR:
ptStreamDesc->decoderConfig->decoderSpecificInfo->data = NULL;
gf_odf_desc_del((GF_Descriptor *)ptStreamDesc);
return res;
s32 MP4Writer::WriteAACSample(u8 *pData, u32 Size, u64 TimeStamp)
GF_ISOSample tISOSample;
if (CONFIG_FINISH != m_AudioAACStatue)
return 0;
if (!findfirsidr_)
return 0;
if (-1 == m_AudioTimeStampStart)
m_AudioTimeStampStart = TimeStamp;
tISOSample.IsRAP = RAP;
tISOSample.dataLength = Size;
tISOSample.data = (char *)pData;
tISOSample.DTS = TimeStamp - m_AudioTimeStampStart;
tISOSample.CTS_Offset = 0;
tISOSample.nb_pack = 0;
if (GF_OK != gf_isom_add_sample(m_ptFile, m_AACTrackIndex, m_AACStreamIndex, &tISOSample))
return -1;
return 0;
extern "C"
void* MP4_Init()
return (void *)(new MP4Writer());
s32 MP4_CreatFile(void *pCMP4Writer, char *strFileName)
if (NULL == pCMP4Writer)
return -1;
return ((MP4Writer *)pCMP4Writer)->CreatFile(strFileName);
s32 MP4_InitVideo265(void *pCMP4Writer, u32 TimeScale)
if (NULL == pCMP4Writer)
return -1;
return ((MP4Writer *)pCMP4Writer)->Init265(TimeScale);
s32 MP4_Write265Sample(void *pCMP4Writer, u8 *pData, u32 Size, u64 TimeStamp)
if (NULL == pCMP4Writer)
return -1;
return ((MP4Writer *)pCMP4Writer)->Write265Sample(pData, Size, TimeStamp);
s32 MP4_InitVideo264(void * pCMP4Writer, u32 TimeScale)
if (NULL == pCMP4Writer)
return -1;
return ((MP4Writer *)pCMP4Writer)->Init264(TimeScale);
s32 MP4_Write264Sample(void * pCMP4Writer, u8 * pData, u32 Size, u64 TimeStamp)
if (NULL == pCMP4Writer)
return -1;
return ((MP4Writer *)pCMP4Writer)->Write264Sample(pData, Size, TimeStamp);
s32 MP4_InitAudioAAC(void *pCMP4Writer, u8 AudioType, u32 SampleRate, u8 Channel, u32 TimeScale)
if (NULL == pCMP4Writer)
return -1;
return ((MP4Writer *)pCMP4Writer)->InitAAC(AudioType, SampleRate, Channel, TimeScale);
s32 MP4_WriteAACSample(void *pCMP4Writer, u8 *pData, u32 Size, u64 TimeStamp)
if (NULL == pCMP4Writer)
return -1;
return ((MP4Writer *)pCMP4Writer)->WriteAACSample(pData, Size, TimeStamp);
void MP4_CloseFile(void *pCMP4Writer)
if (NULL == pCMP4Writer)
return;
((MP4Writer *)pCMP4Writer)->CloseFile();
void MP4_Exit(void *pCMP4Writer)
if (NULL == pCMP4Writer)
return;
delete pCMP4Writer;
注意：
1)MP4_InitAudioAAC那个audiotype指的是aac的profile，lc就是2。
GF_M4A_AAC_MAIN = 1,
GF_M4A_AAC_LC = 2,
GF_M4A_AAC_SSR = 3,
GF_M4A_AAC_LTP = 4,
GF_M4A_AAC_SBR = 5,
GF_M4A_AAC_SCALABLE = 6,
GF_M4A_TWINVQ = 7,
GF_M4A_CELP = 8,
GF_M4A_HVXC = 9,
GF_M4A_TTSI = 12,
GF_M4A_MAIN_SYNTHETIC = 13,
GF_M4A_WAVETABLE_SYNTHESIS = 14,
GF_M4A_GENERAL_MIDI = 15,
GF_M4A_ALGO_SYNTH_AUDIO_FX = 16,
GF_M4A_ER_AAC_LC = 17,
GF_M4A_ER_AAC_LTP = 19,
GF_M4A_ER_AAC_SCALABLE = 20,
GF_M4A_ER_TWINVQ = 21,
GF_M4A_ER_BSAC = 22,
GF_M4A_ER_AAC_LD = 23,
GF_M4A_ER_CELP = 24,
GF_M4A_ER_HVXC = 25,
GF_M4A_ER_HILN = 26,
GF_M4A_ER_PARAMETRIC = 27,
GF_M4A_SSC = 28,
GF_M4A_AAC_PS = 29,
GF_M4A_LAYER1 = 32,
GF_M4A_LAYER2 = 33,
GF_M4A_LAYER3 = 34,
GF_M4A_DST = 35,
GF_M4A_ALS = 36
2)填aac数据时记得除掉adts头，使用上面的AdtsDemux函数即可。
二、写h264流数据为mp4文件步骤
void * handleMp4 = NULL;
handleMp4 = MP4_Init();
Mp4_CreateFile(handleMp4,"d:\\\\test.mp4");
MP4_InitVideo264(handleMp4,1000);
//写h264流数据，包含00 00 00 01标记头的数据
MP4_Write264Sample(handleMp4,h264data,datasize,timestamp);//h264data表示264数据，datasize表示数据大小，timestamp表示时间戳，需要转换成基准为1000时间戳(1000\\90000两种)
...
MP4_CloseFile(handleMp4);
MP4_Exit(handleMp4);三、写h265流数据为mp4文件步骤
void * handleMp4 = NULL;
handleMp4 = MP4_Init();
Mp4_CreateFile(handleMp4,"d:\\\\test.mp4");
MP4_InitVideo265(handleMp4,1000);
//写h265流数据，包含00 00 00 01标记头的数据
MP4_Write265Sample(handleMp4,h264data,datasize,timestamp);//h264data表示264数据，datasize表示数据大小，timestamp表示时间戳，需要转换成基准为1000时间戳
...
MP4_CloseFile(handleMp4);
MP4_Exit(handleMp4);四、利用命令行生成mp4文件
GPAC除了 GPAC 核心，完整的软件包还包括一个 GPAC 插件：MPEG-4 BIF(场景解码器)、MPEG-4 ODF(对象描述符解码器)、MPEG-4 LASeR(场景解码器)、MPEG-4 SAF 解复用器、文本 MPEG- 4 加载器(支持未压缩的 MPEG-4 BT 和 XMT、VRML 和 X3D 文本格式)、图像包(支持 PNG、JPEG、BMP、JPEG2000)等。
1)将h265流转换成MP4，执行下面命令行。
./MP4Box -v -add Catus_1920x1080_50_qp32.bin:FMT=HEVC -fps 50 -new output.mp4./MP4Box -add name_of_annexB_bitstream.(bit,bin,265) -fps 50 -new output.mp4
播放h265视频文件信息，执行下面命令。
./MP4Client output.mp42)将h265流转换成ts，执行下面命令行。
./mp42ts -prog=hevc.mp4 -dst-file=test.ts
播放h265视频文件流信息，执行下面命令。
./MP4Client test.ts备注：
1、GPAC模块下载链接
GitHub - gpac/gpac: Modular Multimedia framework for packaging, streaming and playing your favorite content, see http://netflix.gpac.io或Downloads | GPAC
2、编译指导
Build Introduction · gpac/gpac Wiki · GitHub
Windows：GPAC Build Guide for Windows · gpac/gpac Wiki · GitHub
Linux：GPAC Build Guide for Linux · gpac/gpac Wiki · GitHub
以上是关于利用GPAC生成MP4文件的主要内容，如果未能解决你的问题，请参考以下文章