MP4V2封装的类库，可将H264和AAC直接打包到MP4容器中，堪称经典

废话少说，直接上代码：

// MP4Encoder.h
#pragma once

#ifndef _MP4V2_H_
#define _MP4V2_H_
#include "mp4v2/mp4v2.h"
#endif

#define MP4ENCODER_ERROR(err) ((MP4EncoderResult)(-(err)))
#define DEFAULT_RECORD_TIME 0U

typedef enum
{
	MP4ENCODER_ENONE = 0,
	MP4ENCODER_E_CREATE_FAIL,
	MP4ENCODER_E_ADD_VIDEO_TRACK,
	MP4ENCODER_E_ADD_AUDIO_TRACK,
	MP4ENCODER_WARN_RECORD_OVER,
	MP4ENCODER_E_WRITE_VIDEO_DATA,
	MP4ENCODER_E_WRITE_AUDIO_DATA,
	MP4ENCODER_E_ALLOC_MEMORY_FAILED,
	MP4ENCODER_E_UNKONOWN
}MP4EncoderResult;

class MP4Encoder
{
public:
	MP4Encoder(void);
	~MP4Encoder(void);
	MP4EncoderResult MP4CreateFile(const char *sFileName,
		unsigned uRecordTime = DEFAULT_RECORD_TIME);
	MP4EncoderResult MP4AddH264Track(const uint8_t *sData, int nSize,
		int nWidth, int nHeight, int nFrameRate = 25);
	MP4EncoderResult MP4AddAACTrack(const uint8_t *sData, int nSize);
	MP4EncoderResult MP4WriteH264Data(uint8_t *sData, int nSize, uint64_t u64PTS);
	MP4EncoderResult MP4WriteAACData(const uint8_t *sData, int nSize,
		uint64_t u64PTS);
	void MP4ReleaseFile();
private:
	unsigned m_uSecond;
	MP4FileHandle m_hFile;
	bool m_bFirstAudio, m_bFirstVideo;
	MP4TrackId m_videoTrack, m_audioTrack;
	uint64_t m_u64AudioPTS, m_u64VideoPTS, m_u64FirstPTS, m_u64LastPTS;
};

实现部分：

// MP4Encoder.cpp
#pragma once
#include "MP4Encoder.h"
#pragma comment(lib, "../lib/libmp4v2.lib")

#define MIN_FRAME_SIZE 32
#define VIDEO_TIME_SCALE 90000
#define AUDIO_TIME_SCALE 8000
#define MOVIE_TIME_SCALE VIDEO_TIME_SCALE
#define PTS2TIME_SCALE(CurPTS, PrevPTS, timeScale) \
	((MP4Duration)((CurPTS - PrevPTS) * 1.0 / (double)(1e+6) * timeScale))
#define INVALID_PTS 0xFFFFFFFFFFFFFFFF
/* Warning: Followings are magic data originally */
#define DEFAULT_VIDEO_TRACK_NUM 3
#define DEFAULT_VIDEO_PROFILE_LEVEL 1
#define DEFAULT_AUDIO_PROFILE_LEVEL 2

MP4Encoder::MP4Encoder(void)
	: m_hFile(MP4_INVALID_FILE_HANDLE)
	, m_bFirstVideo(true)
	, m_bFirstAudio(true)
	, m_uSecond(DEFAULT_RECORD_TIME)
	, m_videoTrack(MP4_INVALID_TRACK_ID)
	, m_audioTrack(MP4_INVALID_TRACK_ID)
	, m_u64VideoPTS(0)
	, m_u64AudioPTS(0)
	, m_u64FirstPTS(INVALID_PTS)
	, m_u64LastPTS(INVALID_PTS)
{
}


MP4Encoder::~MP4Encoder(void)
{
}

MP4EncoderResult MP4Encoder::MP4CreateFile(const char *sFileName,
	unsigned uRecordTime /* = DEFAULT_RECORD_TIME */)
{
	m_hFile = MP4Create(sFileName);
	if (m_hFile == MP4_INVALID_FILE_HANDLE)
		return MP4ENCODER_ERROR(MP4ENCODER_E_CREATE_FAIL);
	if (!MP4SetTimeScale(m_hFile, MOVIE_TIME_SCALE))
		return MP4ENCODER_ERROR(MP4ENCODER_E_CREATE_FAIL);
	m_uSecond = uRecordTime;
	return MP4ENCODER_ENONE;
}

MP4EncoderResult MP4Encoder::MP4AddH264Track(const uint8_t *sData, int nSize,
	int nWidth, int nHeight, int nFrameRate/* = 25 */)
{
	int sps, pps;
	for (sps = 0; sps < nSize;)
		if (sData[sps++] == 0x00 && sData[sps++] == 0x00 && sData[sps++] == 0x00
			&& sData[sps++] == 0x01)
			break;
	for (pps = sps; pps < nSize;)
		if (sData[pps++] == 0x00 && sData[pps++] == 0x00 && sData[pps++] == 0x00
			&& sData[pps++] == 0x01)
			break;
	if (sps >= nSize || pps >= nSize)
		return MP4ENCODER_ERROR(MP4ENCODER_E_ADD_VIDEO_TRACK);

	m_videoTrack = MP4AddH264VideoTrack(m_hFile, VIDEO_TIME_SCALE,
		VIDEO_TIME_SCALE / nFrameRate, nWidth, nHeight,
		sData[sps + 1], sData[sps + 2], sData[sps + 3], DEFAULT_VIDEO_TRACK_NUM);
	if (MP4_INVALID_TRACK_ID == m_videoTrack)
		return MP4ENCODER_ERROR(MP4ENCODER_E_ADD_VIDEO_TRACK);

	MP4SetVideoProfileLevel(m_hFile, DEFAULT_VIDEO_PROFILE_LEVEL);
	MP4AddH264SequenceParameterSet(m_hFile, m_videoTrack, sData + sps,
		pps - sps - 4);
	MP4AddH264PictureParameterSet(m_hFile, m_videoTrack, sData + pps,
		nSize - pps);

	return MP4ENCODER_ENONE;
}

MP4EncoderResult MP4Encoder::MP4AddAACTrack(const uint8_t *sData, int nSize)
{
	m_audioTrack = MP4AddAudioTrack(m_hFile, AUDIO_TIME_SCALE,
		/**
		 * In fact, this is not a magic number. A formula might be:
		 * SampleRate * ChannelNum * 2 / SampleFormat
		 * 8000 * 1 * 2 / 16 (字节对齐，这里是 AV_SAMPLE_FMT_S16)
		 */
		AUDIO_TIME_SCALE / 8, MP4_MPEG4_AUDIO_TYPE);
	if (MP4_INVALID_TRACK_ID == m_audioTrack)
		return MP4ENCODER_ERROR(MP4ENCODER_E_ADD_AUDIO_TRACK);
	MP4SetAudioProfileLevel(m_hFile, DEFAULT_AUDIO_PROFILE_LEVEL);
	if (!MP4SetTrackESConfiguration(m_hFile, m_audioTrack, sData, nSize))
		return MP4ENCODER_ERROR(MP4ENCODER_E_ADD_AUDIO_TRACK);

	return MP4ENCODER_ENONE;
}

MP4EncoderResult MP4Encoder::MP4WriteH264Data(uint8_t *sData, int nSize, uint64_t u64PTS)
{
	if (nSize < MIN_FRAME_SIZE)
		return MP4ENCODER_ENONE;
	bool result = false;
	sData[0] = (nSize - 4) >> 24;
	sData[1] = (nSize - 4) >> 16;
	sData[2] = (nSize - 4) >> 8;
	sData[3] = nSize - 4;
	if (m_bFirstVideo)
	{
		if (m_u64FirstPTS > u64PTS)
			m_u64FirstPTS = u64PTS;
		m_u64VideoPTS = u64PTS;
		m_bFirstVideo = false;
	}
	if ((sData[4] & 0x0F) == 5)
		result = MP4WriteSample(m_hFile, m_videoTrack, sData, nSize,
		PTS2TIME_SCALE(u64PTS, m_u64VideoPTS, VIDEO_TIME_SCALE));
	else
		result = MP4WriteSample(m_hFile, m_videoTrack, sData, nSize,
		PTS2TIME_SCALE(u64PTS, m_u64VideoPTS, VIDEO_TIME_SCALE), 0, false);
	if (!result)
		return MP4ENCODER_ERROR(MP4ENCODER_E_WRITE_VIDEO_DATA);
	m_u64LastPTS = m_u64VideoPTS = u64PTS;
	if (m_uSecond && (m_u64LastPTS - m_u64FirstPTS) / (1e+6) >= m_uSecond)
		return MP4ENCODER_ERROR(MP4ENCODER_WARN_RECORD_OVER);
	return MP4ENCODER_ENONE;
}

MP4EncoderResult MP4Encoder::MP4WriteAACData(const uint8_t *sData, int nSize,
	uint64_t u64PTS)
{
	if (nSize < MIN_FRAME_SIZE)
		return MP4ENCODER_ENONE;
	bool result = false;
	if (m_bFirstAudio)
	{
		if (m_u64FirstPTS > u64PTS)
			m_u64FirstPTS = u64PTS;
		m_u64AudioPTS = u64PTS;
		m_bFirstAudio = false;
	}
	result = MP4WriteSample(m_hFile, m_audioTrack, sData, nSize,
		PTS2TIME_SCALE(u64PTS, m_u64AudioPTS, AUDIO_TIME_SCALE));
	if (!result)
		return MP4ENCODER_ERROR(MP4ENCODER_E_WRITE_AUDIO_DATA);
	m_u64LastPTS = m_u64AudioPTS = u64PTS;
	if (m_uSecond && (m_u64LastPTS - m_u64FirstPTS) / (1e+6) >= m_uSecond)
		return MP4ENCODER_ERROR(MP4ENCODER_WARN_RECORD_OVER);
	return MP4ENCODER_ENONE;
}

void MP4Encoder::MP4ReleaseFile()
{
	if (m_hFile != MP4_INVALID_FILE_HANDLE)
	{
		MP4Close(m_hFile);
		m_hFile = MP4_INVALID_FILE_HANDLE;
	}
}

注意，这个是 PC 版本，Android 版本要修改 MP4WriteSample 中的 render！

但是问题又出现了，官方文档告知我们：如果是写视频数据，那么 MP4WriteSample 中的 render 是是不应该设置为默认值 0 的，因为有可能出现音视频不同步，结果我按照说明传递了参数，在手机上同步的更好，但是文件导出到电脑上播放时就会有音视频不同步的问题，这是不是官方的 bug 呢？期待有人去向官方反馈一下，因此现在我们就设置为默认值了，但是不保证以后是否要更改！