1、根据RTSP的Announce请求中的sdp信息和RTP包中的信息初始化音视频编解码器

o_video_stream = avformat_new_stream(o_fmt_ctx, NULL);
    {
        AVCodecContext *c;
        c = o_video_stream->codec;
        c->codec_id = AV_CODEC_ID_H264;
        c->codec_type = AVMEDIA_TYPE_VIDEO;
        c->time_base.num = 1;
        c->time_base.den = 180000;
        c->pkt_timebase = {1, 90000};
        c->channels = 0;
        c->width = 480;
        c->height = 640;
        c->coded_width = 480;
        c->coded_height = 640;
        c->pix_fmt = AV_PIX_FMT_YUV420P;
        if (c->flags & AVFMT_GLOBALHEADER)
            c->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;
    }

    o_audio_stream = avformat_new_stream(o_fmt_ctx, NULL);
    {
        AVCodecContext *c;
        c = o_audio_stream->codec;
        c->audio_service_type = AV_AUDIO_SERVICE_TYPE_MAIN;
        c->sample_rate = 8000;
        c->sample_fmt = AV_SAMPLE_FMT_FLTP;
        c->codec_id = AV_CODEC_ID_AAC;
        c->codec_type = AVMEDIA_TYPE_AUDIO;
        c->time_base.num = 1;
        c->time_base.den = 8000;
        c->channels = 1;
        if (c->flags & AVFMT_GLOBALHEADER)
            c->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;
    }

2、将RTP包中的音视频数据提取出来

*pOutLen = 0;
    if (InLen < RTP_HEAD_LENGTH) {
        return false;
    }

    unsigned char *src = (unsigned char*)bufferIn + RTP_HEAD_LENGTH;
    unsigned char fu_indicator = src[0];
    unsigned char fu_header = src[1];
    unsigned char nalu_type = fu_indicator & 0x1f;
    

    if (nalu_type == 0x1C) {  //FU-A
        unsigned char start_flag = fu_header & 0x80;
        unsigned char end_flag = fu_header & 0x40;
        unsigned char nalu_header = (fu_indicator & 0xe0) | (fu_header & 0x1f);
        if (start_flag != 0) {
            memcpy(*pBufferOut, src - 3, InLen - RTP_HEAD_LENGTH + 3);
            *((int*)(*pBufferOut)) = 0x01000000;
            *((char*)(*pBufferOut + 4)) = nalu_header;
            *pOutLen = InLen - RTP_HEAD_LENGTH + 3;
        }
        else if (end_flag != 0) {
            memcpy(*pBufferOut, src + 2, InLen - RTP_HEAD_LENGTH - 2);
            *pOutLen = InLen - RTP_HEAD_LENGTH - 2;
        }
        else {
            memcpy(*pBufferOut, src + 2, InLen - RTP_HEAD_LENGTH - 2);
            *pOutLen = InLen - RTP_HEAD_LENGTH - 2;
        }
    }
    else if (nalu_type == 0x18) {  //STAP-A
        unsigned int srcOffset = 1 ,bufOffset = 0;
        while (((InLen - RTP_HEAD_LENGTH) - srcOffset) > 2)
        {
            *((int*)(*pBufferOut + bufOffset)) = 0x01000000;
            bufOffset += 4;

            unsigned int size = 0;
            size |= *(src + srcOffset) << 8;
            size |= *(src + srcOffset +1);

            srcOffset += 2;
            memcpy(*pBufferOut + bufOffset, src+srcOffset, size);
            bufOffset += size;
            srcOffset += size;
        }
        *pOutLen = bufOffset;
    }
    else if (nalu_type == 0x1) {
        *pOutLen = InLen - RTP_HEAD_LENGTH + 4;
        memcpy(*pBufferOut, src - 4, *pOutLen);
        *((int *)(*pBufferOut)) = 0x01000000;
    }
    else {
        printf("Unsupport nalu type!\n");
    }

    bool bFinishFrame;
    unsigned char *bufTemp = (unsigned char *)bufferIn;
    if (bufTemp[1] & 0x80) {
        bFinishFrame = true;
    }
    else {
        bFinishFrame = false;
    }
    
    return bFinishFrame;