利用GPAC生成MP4文件

关注

VIP免费原创最新推荐文章于 2026-02-19 05:24:44 发布 1.7k 阅读 AI 写同款· GEO 优化 ›

GPAC实战：MP4Box带你轻松解决MP4封装、修复与拼接难题 MP4作为一种常见的多媒体封装格式，其内部由ftyp、moov、mdat等box结构组成，其中moov保存时间戳与样本索引，一旦损坏或timescale设置不当，就会出现播放器打不开、mp4文件时间长度不对等问题。GPAC是一个开源多媒体框架，其命令行工具MP4Box能对MP4进行精细化的封装、拆封与修复。与ffmpeg相比，MP4Box更擅长处理容器结构问题，例如将m3u8分片转为MP4、修复时长偏差、重建moov索引等。本文从MP4基础结构出发，系统讲解用MP4Box完成裸流封装、多段拼接、损坏修复的实阅读详情

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
 
 
 
#endif

MP4Writer.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)=