ffmpeg转码本地文件
Posted wzzkaifa
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ffmpeg转码本地文件相关的知识,希望对你有一定的参考价值。
ffmpeg转码本地文件(一)
实现目标:输入本地文件。实现本地文件转码,里面包括mux层转码,codec层转码,视频格式转换,音频重採样等功能,功能点请看凝视。注意:凝视非常重要。
#ifndef __FFMPEG_H__ #define __FFMPEG_H__ #include "info.h" extern "C" { #include "libavformat/avformat.h" #include "libavformat/avio.h" #include "libavcodec/avcodec.h" #include "libswscale/swscale.h" #include "libavutil/avutil.h" #include "libavutil/mathematics.h" #include "libswresample/swresample.h" #include "libavutil/opt.h" #include "libavutil/channel_layout.h" #include "libavutil/samplefmt.h" #include "libavdevice/avdevice.h" //摄像头所用 #include "libavfilter/avfilter.h" #include "libavutil/error.h" #include "libavutil/mathematics.h" #include "libavutil/time.h" #include "libavutil/fifo.h" #include "libavutil/audio_fifo.h" //这里是做分片时候重採样编码音频用的 #include "inttypes.h" #include "stdint.h" }; #pragma comment(lib,"avformat.lib") #pragma comment(lib,"avcodec.lib") #pragma comment(lib,"avdevice.lib") #pragma comment(lib,"avfilter.lib") #pragma comment(lib,"avutil.lib") #pragma comment(lib,"postproc.lib") #pragma comment(lib,"swresample.lib") #pragma comment(lib,"swscale.lib") //#define INPUTURL "../in_stream/22.flv" //#define INPUTURL "../in_stream/闪电侠.The.Flash.S01E01.中英字幕.HDTVrip.624X352.mp4" //#define INPUTURL "../in_stream/33.ts" //#define INPUTURL "../in_stream/22mp4.mp4" //#define INPUTURL "../in_stream/EYED0081.MOV" //#define INPUTURL "../in_stream/李荣浩 - 李白.mp3" //#define INPUTURL "../in_stream/avier1.mp4" //#define INPUTURL "../in_stream/分歧者2预告片.mp4" //#define INPUTURL "../in_stream/Class8简单介绍.m4v" //#define INPUTURL "../in_stream/22.ts" //#define INPUTURL "../in_stream/44.mp3" //#define INPUTURL "../in_stream/ceshi.mp4" //#define INPUTURL "../in_stream/33.mp4" #define INPUTURL "../in_stream/father.avi" //#define INPUTURL "../in_stream/西海情歌.wav" //#define INPUTURL "../in_stream/Furious_7_2015_International_Trailer_2_5.1-1080p-HDTN.mp4" //#define INPUTURL "../in_stream/Wildlife.wmv" //#define INPUTURL "../in_stream/单身男女2.HD1280超清国语版.mp4" //#define INPUTURL "rtmp://221.228.193.50:1935/live/teststream1" #define OUTPUTURL "../out_stream/father1111.mp4" //#define OUTPUTURL "rtmp://221.228.193.50:1935/live/zwg" //#define OUTPUTURL "rtmp://221.228.193.50:1935/live/zwg" enum AVSampleFormat_t { AV_SAMPLE_FMT_NONE_t = -1, AV_SAMPLE_FMT_U8_t, ///< unsigned 8 bits AV_SAMPLE_FMT_S16_t, ///< signed 16 bits AV_SAMPLE_FMT_S32_t, ///< signed 32 bits AV_SAMPLE_FMT_FLT_t, ///< float AV_SAMPLE_FMT_DBL_t, ///< double AV_SAMPLE_FMT_U8P_t, ///< unsigned 8 bits, planar AV_SAMPLE_FMT_S16P_t, ///< signed 16 bits, planar AV_SAMPLE_FMT_S32P_t, ///< signed 32 bits, planar AV_SAMPLE_FMT_FLTP_t, ///< float, planar AV_SAMPLE_FMT_DBLP_t, ///< double, planar AV_SAMPLE_FMT_NB_t ///< Number of sample formats. DO NOT USE if linking dynamically }; /************************************************************************/ /************************************************************************/ //是否用音频fifo 否则用全局的fifo 两种方法 //#define AUDIO_FIFO /************************************************************************/ /************************************************************************/ //video param extern int m_dwWidth; extern int m_dwHeight; extern double m_dbFrameRate; //帧率 extern AVCodecID video_codecID; extern AVPixelFormat video_pixelfromat; extern int gop_size; extern int max_b_frame; extern int thread_count; //用cpu内核数目 //audio param extern int m_dwChannelCount; //声道 extern AVSampleFormat_t m_dwBitsPerSample; //样本 extern int m_dwFrequency; //採样率 extern AVCodecID audio_codecID; extern int audio_frame_size; extern int m_audiomuxtimebasetrue; //音频的mux层timebse是否正确 extern AVFifoBuffer * m_fifo; //存放pcm数据 extern AVAudioFifo * m_audiofifo; //音频存放pcm数据 extern int64_t m_first_audio_pts; //第一帧的音频pts extern int m_is_first_audio_pts; //是否已经记录第一帧音频时间戳 #define AUDIO_ID 0 //packet 中的ID ,假设先增加音频 pocket 则音频是 0 视频是1。否则相反(影响add_out_stream顺序) #define VIDEO_ID 1 extern int nRet; //状态标志 extern AVFormatContext* icodec; //输入流context extern AVFormatContext* ocodec ; //输出流context extern char szError[256]; //错误字符串 extern AVStream* ovideo_st; extern AVStream* oaudio_st; extern int video_stream_idx; extern int audio_stream_idx; extern AVCodec *audio_codec; extern AVCodec *video_codec; extern AVPacket pkt; extern AVBitStreamFilterContext * vbsf_aac_adtstoasc; //aac->adts to asc过滤器 static struct SwsContext * img_convert_ctx_video = NULL; static int sws_flags = SWS_BICUBIC; //差值算法,双三次 int init_demux(char * Filename,AVFormatContext ** iframe_c); int init_mux(); int uinit_demux(); int uinit_mux(); //for mux AVStream * add_out_stream(AVFormatContext* output_format_context,AVMediaType codec_type_t); //for codec AVStream * add_out_stream2(AVFormatContext* output_format_context,AVMediaType codec_type_t,AVCodec **codec); int init_decode(int stream_type); int init_code(int stream_type); int uinit_decode(int stream_type); int uinit_code(int stream_type); int perform_decode(int stream_type,AVFrame * picture); int perform_code(int stream_type,AVFrame * picture); //用于AVFifoBuffer int perform_code2(int stream_type,AVFrame * picture); //用于AVAudioFifo void perform_yuv_conversion(AVFrame * pinframe,AVFrame * poutframe); SwrContext * init_pcm_resample(AVFrame *in_frame, AVFrame *out_frame); void uinit_pcm_resample(AVFrame * poutframe,SwrContext * swr_ctx); int preform_pcm_resample(SwrContext * pSwrCtx,AVFrame *in_frame, AVFrame *out_frame); int audio_support(AVCodec * pCodec,int *channel,int * playout,int *samplePerSec,AVSampleFormat_t *sample_fmt); int video_support(AVCodec * pCodec,AVPixelFormat * video_pixelfromat); int transcode(); void write_frame(AVFormatContext *ocodec,int ID,AVPacket pkt_t); //这个是依据传过来的buf 和size 写入文件 #endif
#include "ffmpeg.h" int nRet = 0; AVFormatContext* icodec = NULL; AVFormatContext* ocodec = NULL; char szError[256]; AVStream * ovideo_st = NULL; AVStream * oaudio_st = NULL; int video_stream_idx = -1; int audio_stream_idx = -1; AVCodec *audio_codec = NULL; AVCodec *video_codec = NULL; AVPacket pkt; AVBitStreamFilterContext * vbsf_aac_adtstoasc = NULL; //video param int m_dwWidth = 640; int m_dwHeight = 480; double m_dbFrameRate = 23; //帧率 AVCodecID video_codecID = AV_CODEC_ID_H264; AVPixelFormat video_pixelfromat = AV_PIX_FMT_YUV420P; int bit_rate = 400000; int gop_size = 12; int max_b_frame = 2; int thread_count = 2; //audio param int m_dwChannelCount = 2; //声道 int m_dwFrequency = 48000; //採样率 AVSampleFormat_t m_dwBitsPerSample = AV_SAMPLE_FMT_S16_t; //样本 int m_audiomuxtimebasetrue = 1; //音频的mux层timebse是否正确 //aac AVCodecID audio_codecID = AV_CODEC_ID_AAC; int audio_frame_size = 1024; //mp3 //AVCodecID audio_codecID = AV_CODEC_ID_MP3; //int audio_frame_size = 1152; AVFifoBuffer * m_fifo = NULL; AVAudioFifo * m_audiofifo = NULL; int64_t m_first_audio_pts = 0; int m_is_first_audio_pts = 0; int init_demux(char * Filename,AVFormatContext ** iframe_c) { int i = 0; nRet = avformat_open_input(iframe_c, Filename,NULL, NULL); if (nRet != 0) { av_strerror(nRet, szError, 256); printf(szError); printf("\n"); printf("Call avformat_open_input function failed!\n"); return 0; } if (avformat_find_stream_info(*iframe_c,NULL) < 0) { printf("Call av_find_stream_info function failed!\n"); return 0; } //输出视频信息 av_dump_format(*iframe_c, -1, Filename, 0); //加入音频信息到输出context for (i = 0; i < (*iframe_c)->nb_streams; i++) { if ((*iframe_c)->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO) { double FrameRate = (*iframe_c)->streams[i]->r_frame_rate.num /(double)(*iframe_c)->streams[i]->r_frame_rate.den; m_dbFrameRate =(int)(FrameRate + 0.5); video_stream_idx = i; } else if ((*iframe_c)->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) { audio_stream_idx = i; if(icodec->streams[audio_stream_idx]->time_base.den == 0 || icodec->streams[audio_stream_idx]->time_base.num == 0 || icodec->streams[audio_stream_idx]->codec->sample_rate == 0) { m_audiomuxtimebasetrue = 0; } } } return 1; } int init_mux() { int i = 0; int ret = 0; /* allocate the output media context */ avformat_alloc_output_context2(&ocodec, NULL,NULL, OUTPUTURL); if (!ocodec) { return getchar(); } AVOutputFormat* ofmt = NULL; ofmt = ocodec->oformat; /* open the output file, if needed */ if (!(ofmt->flags & AVFMT_NOFILE)) { if (avio_open(&ocodec->pb, OUTPUTURL, AVIO_FLAG_WRITE) < 0) { printf("Could not open ‘%s‘\n", OUTPUTURL); return getchar(); } } //这里加入的时候AUDIO_ID/VIDEO_ID有影响 //加入音频信息到输出context if(audio_stream_idx != -1)//假设存在音频 { ofmt->audio_codec = audio_codecID; //假设是音频须要编解码 if(audio_codecID != icodec->streams[audio_stream_idx]->codec->codec_id || 1 != icodec->streams[audio_stream_idx]->codec->sample_fmt) { oaudio_st = add_out_stream2(ocodec, AVMEDIA_TYPE_AUDIO,&audio_codec); } else { oaudio_st = add_out_stream(ocodec, AVMEDIA_TYPE_AUDIO); } if ((strstr(ocodec->oformat->name, "flv") != NULL) || (strstr(ocodec->oformat->name, "mp4") != NULL) || (strstr(ocodec->oformat->name, "mov") != NULL) || (strstr(ocodec->oformat->name, "3gp") != NULL)) { if (oaudio_st->codec->codec_id == AV_CODEC_ID_AAC) { vbsf_aac_adtstoasc = av_bitstream_filter_init("aac_adtstoasc"); if(vbsf_aac_adtstoasc == NULL) { return -1; } } } } //加入视频信息到输出context if (video_stream_idx != -1)//假设存在视频 { ofmt->video_codec = video_codecID; //假设是视频须要编解码 if(bit_rate != icodec->streams[video_stream_idx]->codec->bit_rate || m_dwWidth != icodec->streams[video_stream_idx]->codec->width || m_dwHeight != icodec->streams[video_stream_idx]->codec->height || video_codecID != icodec->streams[video_stream_idx]->codec->codec_id || m_dbFrameRate != av_q2d(icodec->streams[video_stream_idx]->r_frame_rate)) { ovideo_st = add_out_stream2(ocodec, AVMEDIA_TYPE_VIDEO,&video_codec); } else { ovideo_st = add_out_stream(ocodec,AVMEDIA_TYPE_VIDEO); } } av_dump_format(ocodec, 0, OUTPUTURL, 1); if (video_stream_idx != -1)//假设存在视频 { //假设是视频须要编解码 if(bit_rate != icodec->streams[video_stream_idx]->codec->bit_rate || m_dwWidth != icodec->streams[video_stream_idx]->codec->width || m_dwHeight != icodec->streams[video_stream_idx]->codec->height || video_codecID != icodec->streams[video_stream_idx]->codec->codec_id || m_dbFrameRate != av_q2d(icodec->streams[video_stream_idx]->r_frame_rate)) { //解码初始化 ret = init_decode(VIDEO_ID); //编码初始化 ret = init_code(VIDEO_ID); } } //假设是音频须要编解码 if(audio_stream_idx != -1)//假设存在音频 { if(audio_codecID != icodec->streams[audio_stream_idx]->codec->codec_id || 1 != icodec->streams[audio_stream_idx]->codec->sample_fmt) { //解码初始化 ret = init_decode(AUDIO_ID); //编码初始化 ret = init_code(AUDIO_ID); } } ret = avformat_write_header(ocodec, NULL); if (ret != 0) { printf("Call avformat_write_header function failed.\n"); return 0; } return 1; } int uinit_demux() { /* free the stream */ av_free(icodec); return 1; } int uinit_mux() { int i = 0; nRet = av_write_trailer(ocodec); if (nRet < 0) { av_strerror(nRet, szError, 256); printf(szError); printf("\n"); printf("Call av_write_trailer function failed\n"); } if (vbsf_aac_adtstoasc !=NULL) { av_bitstream_filter_close(vbsf_aac_adtstoasc); vbsf_aac_adtstoasc = NULL; } av_dump_format(ocodec, -1, OUTPUTURL, 1); if (video_stream_idx != -1)//假设存在视频 { //假设是视频须要编解码 if(bit_rate != icodec->streams[video_stream_idx]->codec->bit_rate || m_dwWidth != icodec->streams[video_stream_idx]->codec->width || m_dwHeight != icodec->streams[video_stream_idx]->codec->height || video_codecID != icodec->streams[video_stream_idx]->codec->codec_id || m_dbFrameRate != av_q2d(icodec->streams[video_stream_idx]->r_frame_rate)) { uinit_decode(VIDEO_ID); uinit_code(VIDEO_ID); } } if(audio_stream_idx != -1)//假设存在音频 { //假设是音频须要编解码 if(audio_codecID != icodec->streams[audio_stream_idx]->codec->codec_id || 1 != icodec->streams[audio_stream_idx]->codec->sample_fmt) { uinit_decode(AUDIO_ID); uinit_code(AUDIO_ID); } } /* Free the streams. */ for (i = 0; i < ocodec->nb_streams; i++) { av_freep(&ocodec->streams[i]->codec); av_freep(&ocodec->streams[i]); } if (!(ocodec->oformat->flags & AVFMT_NOFILE)) { /* Close the output file. */ avio_close(ocodec->pb); } av_free(ocodec); return 1; } int init_decode(int stream_type) { AVCodec *pcodec = NULL; AVCodecContext *cctext = NULL; if (stream_type == AUDIO_ID) { cctext = icodec->streams[audio_stream_idx]->codec; pcodec = avcodec_find_decoder(cctext->codec_id); if (!pcodec) { return -1; } } else if (stream_type == VIDEO_ID) { cctext = icodec->streams[video_stream_idx]->codec; pcodec = avcodec_find_decoder(cctext->codec_id); if (!pcodec) { return -1; } } //打开解码器 nRet = avcodec_open2(cctext, pcodec, NULL); if (nRet < 0) { printf("Could not open decoder\n"); return -1; } return 1; } int init_code(int stream_type) { AVCodecContext *cctext = NULL; if (stream_type == AUDIO_ID) { cctext = oaudio_st->codec; //打开编码器 nRet = avcodec_open2(cctext, audio_codec, NULL); if (nRet < 0) { printf("Could not open encoder\n"); return 0; } } else if (stream_type == VIDEO_ID) { cctext = ovideo_st->codec; //打开编码器 nRet = avcodec_open2(cctext, video_codec, NULL); if (nRet < 0) { printf("Could not open encoder\n"); return -1; } } return 1; } int uinit_decode(int stream_type) { AVCodecContext *cctext = NULL; if (stream_type == AUDIO_ID) { cctext = oaudio_st->codec; } else if (stream_type == VIDEO_ID) { cctext = icodec->streams[video_stream_idx]->codec; } avcodec_close(cctext); return 1; } int uinit_code(int stream_type) { AVCodecContext *cctext = NULL; if (stream_type == AUDIO_ID) { cctext = oaudio_st->codec; } else if (stream_type == VIDEO_ID) { cctext = ovideo_st->codec; } avcodec_close(cctext); return 1; } AVStream * add_out_stream(AVFormatContext* output_format_context,AVMediaType codec_type_t) { AVStream * in_stream = NULL; AVStream * output_stream = NULL; AVCodecContext* output_codec_context = NULL; output_stream = avformat_new_stream(output_format_context,NULL); if (!output_stream) { return NULL; } switch (codec_type_t) { case AVMEDIA_TYPE_AUDIO: in_stream = icodec->streams[audio_stream_idx]; break; case AVMEDIA_TYPE_VIDEO: in_stream = icodec->streams[video_stream_idx]; break; default: break; } output_stream->id = output_format_context->nb_streams - 1; output_codec_context = output_stream->codec; output_stream->time_base = in_stream->time_base; int ret = 0; ret = avcodec_copy_context(output_stream->codec, in_stream->codec); if (ret < 0) { printf("Failed to copy context from input to output stream codec context\n"); return NULL; } //这个非常重要。要么纯复用解复用。不做编解码写头会失败, //另或者须要编解码假设不这样,生成的文件没有预览图,还有加入以下的header失败,置0之后会又一次生成extradata output_codec_context->codec_tag = 0; //if(! strcmp( output_format_context-> oformat-> name, "mp4" ) || //!strcmp (output_format_context ->oformat ->name , "mov" ) || //!strcmp (output_format_context ->oformat ->name , "3gp" ) || //!strcmp (output_format_context ->oformat ->name , "flv")) if(AVFMT_GLOBALHEADER & output_format_context->oformat->flags) { output_codec_context->flags |= CODEC_FLAG_GLOBAL_HEADER; } return output_stream; } AVStream * add_out_stream2(AVFormatContext* output_format_context,AVMediaType codec_type_t,AVCodec **codec) { AVCodecContext* output_codec_context = NULL; AVStream * in_stream = NULL; AVStream * output_stream = NULL; AVCodecID codecID; switch (codec_type_t) { case AVMEDIA_TYPE_AUDIO: codecID = audio_codecID; in_stream = icodec->streams[audio_stream_idx]; break; case AVMEDIA_TYPE_VIDEO: codecID = video_codecID; in_stream = icodec->streams[video_stream_idx]; break; default: break; } /* find the encoder */ *codec = avcodec_find_encoder(codecID); if (!(*codec)) { return NULL; } output_stream = avformat_new_stream(output_format_context,*codec); if (!output_stream) { return NULL; } output_stream->id = output_format_context->nb_streams - 1; output_codec_context = output_stream->codec; output_stream->time_base = in_stream->time_base; switch (codec_type_t) { case AVMEDIA_TYPE_AUDIO: output_codec_context->codec_id = audio_codecID; output_codec_context->codec_type = codec_type_t; output_stream->start_time = 0; output_codec_context->sample_rate = icodec->streams[audio_stream_idx]->codec->sample_rate;//m_dwFrequency; if(icodec->streams[audio_stream_idx]->codec->channels > 2) { output_codec_context->channels = m_dwChannelCount; output_codec_context->channel_layout = av_get_default_channel_layout(m_dwChannelCount); } else { output_codec_context->channels = icodec->streams[audio_stream_idx]->codec->channels; if (icodec->streams[audio_stream_idx]->codec->channel_layout == 0) { output_codec_context->channel_layout = av_get_default_channel_layout(icodec->streams[audio_stream_idx]->codec->channels); } else { output_codec_context->channel_layout = icodec->streams[audio_stream_idx]->codec->channel_layout; } } //这个码率有些编码器不支持特别大,比如wav的码率是1411200 比aac大了10倍多 output_codec_context->bit_rate = 128000;//icodec->streams[audio_stream_idx]->codec->bit_rate; output_codec_context->frame_size = audio_frame_size; output_codec_context->sample_fmt = (AVSampleFormat)m_dwBitsPerSample; //样本 output_codec_context->block_align = 0; //查看音频支持的声道,採样率。样本 audio_support(*codec,&output_codec_context->channels, (int *)&output_codec_context->channel_layout, &output_codec_context->sample_rate, (AVSampleFormat_t *)&output_codec_context->sample_fmt); m_dwChannelCount = output_codec_context->channels; m_dwFrequency = output_codec_context->sample_rate; m_dwBitsPerSample = (AVSampleFormat_t)output_codec_context->sample_fmt; break; case AVMEDIA_TYPE_VIDEO: AVRational r_frame_rate_t; r_frame_rate_t.num = 100; r_frame_rate_t.den = (int)(m_dbFrameRate * 100); output_codec_context->time_base = in_stream->codec->time_base; output_stream->time_base = in_stream->time_base; output_stream->r_frame_rate.num = r_frame_rate_t.den; output_stream->r_frame_rate.den = r_frame_rate_t.num; output_codec_context->codec_id = video_codecID; output_codec_context->codec_type = codec_type_t; output_stream->start_time = 0; output_codec_context->pix_fmt = video_pixelfromat; output_codec_context->width = m_dwWidth; output_codec_context->height = m_dwHeight; output_codec_context->bit_rate = bit_rate; output_codec_context->gop_size = gop_size; /* emit one intra frame every twelve frames at most */; output_codec_context->max_b_frames = max_b_frame; //设置B帧最大数 output_codec_context->thread_count = thread_count; //线程数目 output_codec_context->me_range = 16; output_codec_context->max_qdiff = 4; output_codec_context->qmin = 20; //调节清晰度和编码速度 //这个值调节编码后输出数据量越大输出数据量越小。越大编码速度越快。清晰度越差 output_codec_context->qmax = 40; //调节清晰度和编码速度 output_codec_context->qcompress = 0.6; //查看视频支持的yuv格式 video_support(*codec,&output_codec_context->pix_fmt); video_pixelfromat = output_codec_context->pix_fmt; break; default: break; } //这个非常重要,要么纯复用解复用,不做编解码写头会失败, //另或者须要编解码假设不这样。生成的文件没有预览图,还有加入以下的header失败,置0之后会又一次生成extradata output_codec_context->codec_tag = 0; //if(! strcmp( output_format_context-> oformat-> name, "mp4" ) || // !strcmp (output_format_context ->oformat ->name , "mov" ) || // !strcmp (output_format_context ->oformat ->name , "3gp" ) || // !strcmp (output_format_context ->oformat ->name , "flv" )) if(AVFMT_GLOBALHEADER & output_format_context->oformat->flags) { output_codec_context->flags |= CODEC_FLAG_GLOBAL_HEADER; } return output_stream; } int perform_decode(int stream_type,AVFrame * picture) { AVCodecContext *cctext = NULL; int frameFinished = 0 ; if (stream_type == AUDIO_ID) { cctext = icodec->streams[audio_stream_idx]->codec; avcodec_decode_audio4(cctext,picture,&frameFinished,&pkt); if(frameFinished) { return 0; } } else if (stream_type == VIDEO_ID) { cctext = icodec->streams[video_stream_idx]->codec; avcodec_decode_video2(cctext,picture,&frameFinished,&pkt); if(frameFinished) { return 0; } } return 1; } int perform_code(int stream_type,AVFrame * picture) { AVCodecContext *cctext = NULL; AVPacket pkt_t; av_init_packet(&pkt_t); pkt_t.data = NULL; // packet data will be allocated by the encoder pkt_t.size = 0; int frameFinished = 0 ; if (stream_type == AUDIO_ID) { cctext = oaudio_st->codec; //假设进和出的的声道,样本,採样率不同,须要重採样 if(icodec->streams[audio_stream_idx]->codec->sample_fmt != (AVSampleFormat)m_dwBitsPerSample || icodec->streams[audio_stream_idx]->codec->channels != m_dwChannelCount || icodec->streams[audio_stream_idx]->codec->sample_rate != m_dwFrequency) { int64_t pts_t = picture->pts; int duration_t = 0; if(m_audiomuxtimebasetrue == 0) { duration_t = (double)cctext->frame_size * (icodec->streams[audio_stream_idx]->codec->time_base.den /icodec->streams[audio_stream_idx]->codec->time_base.num)/ icodec->streams[audio_stream_idx]->codec->sample_rate; } else { duration_t = (double)cctext->frame_size * (icodec->streams[audio_stream_idx]->time_base.den /icodec->streams[audio_stream_idx]->time_base.num)/ icodec->streams[audio_stream_idx]->codec->sample_rate; } int frame_bytes = cctext->frame_size * av_get_bytes_per_sample(cctext->sample_fmt)* cctext->channels; AVFrame * pFrameResample = avcodec_alloc_frame(); uint8_t * readbuff = new uint8_t[frame_bytes]; if(av_sample_fmt_is_planar(cctext->sample_fmt)) { frame_bytes /= cctext->channels; } while (av_fifo_size(m_fifo) >= frame_bytes) //取出写入的未读的包 { pFrameResample->nb_samples = cctext->frame_size; av_fifo_generic_read(m_fifo, readbuff, frame_bytes, NULL); //这里一定要考虑音频分片的问题 //假设是分片的avcodec_fill_audio_frame传入的buf是单声道的,可是buf_size 是两个声道加一起的数据量 //假设不是分片的avcodec_fill_audio_frame传入的buf是双声道的,buf_size 是两个声道加一起的数据量 if(av_sample_fmt_is_planar(cctext->sample_fmt)) { avcodec_fill_audio_frame(pFrameResample,cctext->channels,cctext->sample_fmt,readbuff,frame_bytes * cctext->channels,1); } else { avcodec_fill_audio_frame(pFrameResample,cctext->channels,cctext->sample_fmt,readbuff,frame_bytes,0); } if(m_is_first_audio_pts == 0) { m_first_audio_pts = pts_t; m_is_first_audio_pts = 1; } pFrameResample->pts = m_first_audio_pts; m_first_audio_pts += duration_t; pFrameResample->pts = av_rescale_q_rnd(pFrameResample->pts, icodec->streams[audio_stream_idx]->codec->time_base, oaudio_st->codec->time_base, AV_ROUND_NEAR_INF); nRet = avcodec_encode_audio2(cctext,&pkt_t,pFrameResample,&frameFinished); if (nRet>=0 && frameFinished) { write_frame(ocodec,AUDIO_ID,pkt_t); av_free_packet(&pkt_t); } } if (readbuff) { delete []readbuff; readbuff = NULL; } if (pFrameResample) { av_free(pFrameResample); pFrameResample = NULL; } } else { nRet = avcodec_encode_audio2(cctext,&pkt_t,picture,&frameFinished); if (nRet>=0 && frameFinished) { write_frame(ocodec,AUDIO_ID,pkt_t); av_free_packet(&pkt_t); } } } else if (stream_type == VIDEO_ID) { cctext = ovideo_st->codec; //ticks_per_frame这个值用的不正确 //if(icodec->streams[video_stream_idx]->codec->ticks_per_frame != 1) //{ // AVRational time_base_video_t; // time_base_video_t.num = icodec->streams[video_stream_idx]->codec->time_base.num; // time_base_video_t.den = icodec->streams[video_stream_idx]->codec->time_base.den /icodec->streams[video_stream_idx]->codec->ticks_per_frame; // picture->pts = av_rescale_q_rnd(picture->pts, time_base_video_t, ovideo_st->codec->time_base, AV_ROUND_NEAR_INF); //} //else //{ // picture->pts = av_rescale_q_rnd(picture->pts, icodec->streams[video_stream_idx]->codec->time_base, ovideo_st->codec->time_base, AV_ROUND_NEAR_INF); //} picture->pts = av_rescale_q_rnd(picture->pts, icodec->streams[video_stream_idx]->codec->time_base, ovideo_st->codec->time_base, AV_ROUND_NEAR_INF); avcodec_encode_video2(cctext,&pkt_t,picture,&frameFinished); picture->pts++; if (frameFinished) { write_frame(ocodec,VIDEO_ID,pkt_t); av_free_packet(&pkt_t); } } return 1; } int perform_code2(int stream_type,AVFrame * picture) { AVCodecContext *cctext = NULL; AVPacket pkt_t; av_init_packet(&pkt_t); pkt_t.data = NULL; // packet data will be allocated by the encoder pkt_t.size = 0; int frameFinished = 0 ; if (stream_type == AUDIO_ID) { cctext = oaudio_st->codec; //假设进和出的的声道。样本,採样率不同,须要重採样 if(icodec->streams[audio_stream_idx]->codec->sample_fmt != (AVSampleFormat)m_dwBitsPerSample || icodec->streams[audio_stream_idx]->codec->channels != m_dwChannelCount || icodec->streams[audio_stream_idx]->codec->sample_rate != m_dwFrequency) { int64_t pts_t = picture->pts; int duration_t = 0; if(m_audiomuxtimebasetrue == 0) { duration_t = (double)cctext->frame_size * (icodec->streams[audio_stream_idx]->codec->time_base.den /icodec->streams[audio_stream_idx]->codec->time_base.num)/ icodec->streams[audio_stream_idx]->codec->sample_rate; } else { duration_t = (double)cctext->frame_size * (icodec->streams[audio_stream_idx]->time_base.den /icodec->streams[audio_stream_idx]->time_base.num)/ icodec->streams[audio_stream_idx]->codec->sample_rate; } AVFrame * pFrameResample = avcodec_alloc_frame(); pFrameResample = av_frame_alloc(); pFrameResample->nb_samples = cctext->frame_size; pFrameResample->channel_layout = cctext->channel_layout; pFrameResample->channels = cctext->channels; pFrameResample->format = cctext->sample_fmt; pFrameResample->sample_rate = cctext->sample_rate; int error = 0; if ((error = av_frame_get_buffer(pFrameResample, 0)) < 0) { av_frame_free(&pFrameResample); return error; } while (av_audio_fifo_size(m_audiofifo) >= pFrameResample->nb_samples) //取出写入的未读的包 { av_audio_fifo_read(m_audiofifo,(void **)pFrameResample->data,pFrameResample->nb_samples); if(m_is_first_audio_pts == 0) { m_first_audio_pts = pts_t; m_is_first_audio_pts = 1; } pFrameResample->pts = m_first_audio_pts; m_first_audio_pts += duration_t; pFrameResample->pts = av_rescale_q_rnd(pFrameResample->pts, icodec->streams[audio_stream_idx]->codec->time_base, oaudio_st->codec->time_base, AV_ROUND_NEAR_INF); nRet = avcodec_encode_audio2(cctext,&pkt_t,pFrameResample,&frameFinished); if (nRet>=0 && frameFinished) { write_frame(ocodec,AUDIO_ID,pkt_t); av_free_packet(&pkt_t); } } if (pFrameResample) { av_frame_free(&pFrameResample); pFrameResample = NULL; } } else { nRet = avcodec_encode_audio2(cctext,&pkt_t,picture,&frameFinished); if (nRet>=0 && frameFinished) { write_frame(ocodec,AUDIO_ID,pkt_t); av_free_packet(&pkt_t); } } } else if (stream_type == VIDEO_ID) { cctext = ovideo_st->codec; if(icodec->streams[video_stream_idx]->codec->ticks_per_frame != 1) { AVRational time_base_video_t; time_base_video_t.num = icodec->streams[video_stream_idx]->codec->time_base.num; time_base_video_t.den = icodec->streams[video_stream_idx]->codec->time_base.den /icodec->streams[video_stream_idx]->codec->ticks_per_frame; picture->pts = av_rescale_q_rnd(picture->pts, time_base_video_t, ovideo_st->codec->time_base, AV_ROUND_NEAR_INF); } else { picture->pts = av_rescale_q_rnd(picture->pts, icodec->streams[video_stream_idx]->codec->time_base, ovideo_st->codec->time_base, AV_ROUND_NEAR_INF); } avcodec_encode_video2(cctext,&pkt_t,picture,&frameFinished); picture->pts++; if (frameFinished) { write_frame(ocodec,VIDEO_ID,pkt_t); av_free_packet(&pkt_t); } } return 1; } void perform_yuv_conversion(AVFrame * pinframe,AVFrame * poutframe) { //设置转换context if (img_convert_ctx_video == NULL) { img_convert_ctx_video = sws_getContext(icodec->streams[video_stream_idx]->codec->width, icodec->streams[video_stream_idx]->codec->height, icodec->streams[video_stream_idx]->codec->pix_fmt, m_dwWidth, m_dwHeight, video_pixelfromat, sws_flags, NULL, NULL, NULL); if (img_convert_ctx_video == NULL) { printf("Cannot initialize the conversion context\n"); } } //開始转换 sws_scale(img_convert_ctx_video, pinframe->data, pinframe->linesize, 0, icodec->streams[video_stream_idx]->codec->height, poutframe->data, poutframe->linesize); poutframe->pkt_pts = pinframe->pkt_pts; poutframe->pkt_dts = pinframe->pkt_dts; //有时pkt_pts和pkt_dts不同,而且pkt_pts是编码前的dts,这里要给avframe传入pkt_dts而不能用pkt_pts //poutframe->pts = poutframe->pkt_pts; poutframe->pts = pinframe->pkt_dts; } SwrContext * init_pcm_resample(AVFrame *in_frame, AVFrame *out_frame) { SwrContext * swr_ctx = NULL; swr_ctx = swr_alloc(); if (!swr_ctx) { printf("swr_alloc error \n"); return NULL; } AVCodecContext * audio_dec_ctx = icodec->streams[audio_stream_idx]->codec; AVSampleFormat sample_fmt; sample_fmt = (AVSampleFormat)m_dwBitsPerSample; //样本 if (audio_dec_ctx->channel_layout == 0) { audio_dec_ctx->channel_layout = av_get_default_channel_layout(icodec->streams[audio_stream_idx]->codec->channels); } /* set options */ av_opt_set_int(swr_ctx, "in_channel_layout", audio_dec_ctx->channel_layout, 0); av_opt_set_int(swr_ctx, "in_sample_rate", audio_dec_ctx->sample_rate, 0); av_opt_set_sample_fmt(swr_ctx, "in_sample_fmt", audio_dec_ctx->sample_fmt, 0); if(icodec->streams[audio_stream_idx]->codec->channels > 2) { av_opt_set_int(swr_ctx, "out_channel_layout", av_get_default_channel_layout(m_dwChannelCount), 0); } else { av_opt_set_int(swr_ctx, "out_channel_layout", audio_dec_ctx->channel_layout, 0); } av_opt_set_int(swr_ctx, "out_sample_rate", audio_dec_ctx->sample_rate, 0); av_opt_set_sample_fmt(swr_ctx, "out_sample_fmt", sample_fmt, 0); swr_init(swr_ctx); int64_t src_nb_samples = in_frame->nb_samples; out_frame->nb_samples = av_rescale_rnd(swr_get_delay(swr_ctx,oaudio_st->codec->sample_rate) + src_nb_samples, oaudio_st->codec->sample_rate, oaudio_st->codec->sample_rate, AV_ROUND_UP); int ret = av_samples_alloc(out_frame->data, &out_frame->linesize[0], icodec->streams[audio_stream_idx]->codec->channels, out_frame->nb_samples,oaudio_st->codec->sample_fmt,1); if (ret < 0) { return NULL; } #ifdef AUDIO_FIFO m_audiofifo = av_audio_fifo_alloc(oaudio_st->codec->sample_fmt, oaudio_st->codec->channels, out_frame->nb_samples); #else //pcm分包初始化 int buffersize = av_samples_get_buffer_size(NULL, oaudio_st->codec->channels, 2048, oaudio_st->codec->sample_fmt, 1); m_fifo = av_fifo_alloc(buffersize); #endif return swr_ctx; } int preform_pcm_resample(SwrContext * pSwrCtx,AVFrame *in_frame, AVFrame *out_frame) { int ret = 0; if (pSwrCtx != NULL) { ret = swr_convert(pSwrCtx, out_frame->data, out_frame->nb_samples, (const uint8_t**)in_frame->data, in_frame->nb_samples); if (ret < 0) { return -1; } //改动分包内存 int buffersize = av_samples_get_buffer_size(&out_frame->linesize[0], oaudio_st->codec->channels, ret, oaudio_st->codec->sample_fmt, 1); #ifdef AUDIO_FIFO int sss = av_audio_fifo_size(m_audiofifo); sss = av_audio_fifo_realloc(m_audiofifo, av_audio_fifo_size(m_audiofifo) + out_frame->nb_samples); av_audio_fifo_write(m_audiofifo,(void **)out_frame->data,out_frame->nb_samples); sss = av_audio_fifo_size(m_audiofifo); #else int sss = av_fifo_size(m_fifo); sss = av_fifo_realloc2(m_fifo, av_fifo_size(m_fifo) + out_frame->linesize[0]); sss = av_fifo_size(m_fifo); av_fifo_generic_write(m_fifo, out_frame->data[0], out_frame->linesize[0], NULL); #endif out_frame->pkt_pts = in_frame->pkt_pts; out_frame->pkt_dts = in_frame->pkt_dts; //有时pkt_pts和pkt_dts不同。而且pkt_pts是编码前的dts,这里要给avframe传入pkt_dts而不能用pkt_pts //out_frame->pts = out_frame->pkt_pts; out_frame->pts = in_frame->pkt_dts; } return 0; } void uinit_pcm_resample(AVFrame * poutframe,SwrContext * swr_ctx) { if (poutframe) { avcodec_free_frame(&poutframe); poutframe = NULL; } if (swr_ctx) { swr_free(&swr_ctx); swr_ctx = NULL; } #ifdef AUDIO_FIFO if(m_audiofifo) { av_audio_fifo_free(m_audiofifo); m_audiofifo = NULL; } #else //析构pcm分包结构 if(m_fifo) { av_fifo_free(m_fifo); m_fifo = NULL; } #endif } int audio_support(AVCodec * pCodec,int *channel,int * playout,int *samplePerSec,AVSampleFormat_t * sample_fmt) { //支持的声道 if(NULL != pCodec->channel_layouts) { uint64_t layout = av_get_default_channel_layout(*channel); if(0 == layout) { return 0; } int i = 0; int j = 0; while(0 != pCodec->channel_layouts[j]) { printf("pCodec->channel_layouts[j] : %d\n",pCodec->channel_layouts[j]); ++j; } while(0 != pCodec->channel_layouts[i]) { if(layout == pCodec->channel_layouts[i]) { break; } ++i; } //未找到 if(0 == pCodec->channel_layouts[i]) { *playout = pCodec->channel_layouts[i-1]; *channel = av_get_channel_layout_nb_channels(*playout); } } //支持的採样率 if(NULL != pCodec->supported_samplerates) { int i = 0; int j = 0; while(0 != pCodec->supported_samplerates[j]) { printf("pCodec->supported_samplerates[j] : %d\n",pCodec->supported_samplerates[j]); ++j; } while(0 != pCodec->supported_samplerates[i]) { if(*samplePerSec == pCodec->supported_samplerates[i]) { break; } ++i; } //未找到 if(0 == pCodec->supported_samplerates[i]) { *samplePerSec = pCodec->supported_samplerates[i-1]; } } //支持的样本 if(NULL != pCodec->sample_fmts) { int i = 0; int j = 0; while(-1 != pCodec->sample_fmts[j]) { printf("pCodec->sample_fmts[j] : %d\n",pCodec->sample_fmts[j]); ++j; } while(-1 != pCodec->sample_fmts[i]) { if(*sample_fmt == pCodec->sample_fmts[i]) { break; } ++i; } //未找到 if(-1 == pCodec->sample_fmts[i]) { *sample_fmt = (AVSampleFormat_t)pCodec->sample_fmts[i-1]; } } return 1; } int video_support(AVCodec * pCodec,AVPixelFormat * video_pixelfromat) { //支持的yuv格式 if(NULL != pCodec->pix_fmts) { int i = 0; int j = 0; while(0 != pCodec->pix_fmts[j]) { printf("pCodec->pix_fmts[j] : %d\n",pCodec->pix_fmts[j]); ++j; } while(0 != pCodec->pix_fmts[i]) { if(*video_pixelfromat == pCodec->pix_fmts[i]) { break; } ++i; } //未找到 if(-1 == pCodec->pix_fmts[i]) { *video_pixelfromat = pCodec->pix_fmts[i-1]; } } return 1; } void write_frame(AVFormatContext *ocodec,int ID,AVPacket pkt_t) { int64_t pts = 0, dts = 0; int nRet = -1; if(ID == VIDEO_ID) { AVPacket videopacket_t; av_init_packet(&videopacket_t); videopacket_t.pts = av_rescale_q_rnd(pkt_t.pts, icodec->streams[video_stream_idx]->time_base, ovideo_st->time_base, AV_ROUND_NEAR_INF); videopacket_t.dts = av_rescale_q_rnd(pkt_t.dts, icodec->streams[video_stream_idx]->time_base, ovideo_st->time_base, AV_ROUND_NEAR_INF); videopacket_t.duration = av_rescale_q(pkt_t.duration,icodec->streams[video_stream_idx]->time_base, ovideo_st->time_base); videopacket_t.flags = pkt_t.flags; videopacket_t.stream_index = VIDEO_ID; //这里add_out_stream顺序有影响 videopacket_t.data = pkt_t.data; videopacket_t.size = pkt_t.size; nRet = av_interleaved_write_frame(ocodec, &videopacket_t); if (nRet != 0) { printf("error av_interleaved_write_frame _ video\n"); } printf("video\n"); } else if(ID == AUDIO_ID) { AVPacket audiopacket_t; av_init_packet(&audiopacket_t); if(m_audiomuxtimebasetrue == 0) { audiopacket_t.pts = av_rescale_q_rnd(pkt_t.pts, icodec->streams[audio_stream_idx]->codec->time_base, oaudio_st->time_base, AV_ROUND_NEAR_INF); audiopacket_t.dts = av_rescale_q_rnd(pkt_t.dts, icodec->streams[audio_stream_idx]->codec->time_base, oaudio_st->time_base, AV_ROUND_NEAR_INF); audiopacket_t.duration = av_rescale_q(pkt_t.duration,icodec->streams[audio_stream_idx]->codec->time_base, oaudio_st->time_base); } else { audiopacket_t.pts = av_rescale_q_rnd(pkt_t.pts, icodec->streams[audio_stream_idx]->time_base, oaudio_st->time_base, AV_ROUND_NEAR_INF); audiopacket_t.dts = av_rescale_q_rnd(pkt_t.dts, icodec->streams[audio_stream_idx]->time_base, oaudio_st->time_base, AV_ROUND_NEAR_INF); audiopacket_t.duration = av_rescale_q(pkt_t.duration,icodec->streams[audio_stream_idx]->time_base, oaudio_st->time_base); } audiopacket_t.flags = pkt_t.flags; audiopacket_t.stream_index = AUDIO_ID; //这里add_out_stream顺序有影响 audiopacket_t.data = pkt_t.data; audiopacket_t.size = pkt_t.size; //加入过滤器 if(! strcmp( ocodec->oformat-> name, "mp4" ) || !strcmp (ocodec ->oformat ->name , "mov" ) || !strcmp (ocodec ->oformat ->name , "3gp" ) || !strcmp (ocodec ->oformat ->name , "flv" )) { if (oaudio_st->codec->codec_id == AV_CODEC_ID_AAC) { if (vbsf_aac_adtstoasc != NULL) { AVPacket filteredPacket = audiopacket_t; int a = av_bitstream_filter_filter(vbsf_aac_adtstoasc, oaudio_st->codec, NULL,&filteredPacket.data, &filteredPacket.size, audiopacket_t.data, audiopacket_t.size, audiopacket_t.flags & AV_PKT_FLAG_KEY); if (a > 0) { av_free_packet(&audiopacket_t); filteredPacket.destruct = av_destruct_packet; audiopacket_t = filteredPacket; } else if (a == 0) { audiopacket_t = filteredPacket; } else if (a < 0) { fprintf(stderr, "%s failed for stream %d, codec %s", vbsf_aac_adtstoasc->filter->name,audiopacket_t.stream_index,oaudio_st->codec->codec ? oaudio_st->codec->codec->name : "copy"); av_free_packet(&audiopacket_t); } } } } nRet = av_interleaved_write_frame(ocodec, &audiopacket_t); if (nRet != 0) { printf("error av_interleaved_write_frame _ audio\n"); } printf("audio\n"); } } int transcode() { AVFrame *pinframe = NULL; AVFrame * pout_video_frame = NULL; AVFrame * pout_audio_frame = NULL; SwrContext * swr_ctx = NULL; int dst_nb_samples = -1; int resampled_data_size = 0; //分配一个AVFrame并设置默认值 pinframe = avcodec_alloc_frame(); pout_video_frame = avcodec_alloc_frame(); pout_audio_frame = avcodec_alloc_frame(); pinframe->pts = 0; pout_video_frame->pts = 0; pout_audio_frame->pts = 0; if (pinframe == NULL) { printf("avcodec_alloc_frame pinframe error\n"); return 0; } //video if (pout_video_frame == NULL) { printf("avcodec_alloc_frame pout_video_frame error\n"); return 0; } int Out_size = avpicture_get_size(video_pixelfromat, m_dwWidth,m_dwHeight); uint8_t * pOutput_buf =( uint8_t *)malloc(Out_size * 3 * sizeof(char)); //最大分配的空间,能满足yuv的各种格式 avpicture_fill((AVPicture *)pout_video_frame, (unsigned char *)pOutput_buf, video_pixelfromat,m_dwWidth, m_dwHeight); //内存关联 //audio if (pout_audio_frame == NULL) { printf("avcodec_alloc_frame pout_audio_frame error\n"); return 0; } avcodec_get_frame_defaults(pout_audio_frame); //開始解包 while (1) { av_init_packet(&pkt); if (av_read_frame(icodec, &pkt) < 0) { break; } //视频 if(pkt.stream_index == video_stream_idx) { //假设是视频须要编解码 if(bit_rate != icodec->streams[video_stream_idx]->codec->bit_rate || m_dwWidth != icodec->streams[video_stream_idx]->codec->width || m_dwHeight != icodec->streams[video_stream_idx]->codec->height || video_codecID != icodec->streams[video_stream_idx]->codec->codec_id || m_dbFrameRate != av_q2d(icodec->streams[video_stream_idx]->r_frame_rate)) { nRet = perform_decode(VIDEO_ID,pinframe); if (nRet == 0) { perform_yuv_conversion(pinframe,pout_video_frame); #ifdef AUDIO_FIFO nRet = perform_code2(VIDEO_ID,pout_video_frame); #else nRet = perform_code(VIDEO_ID,pout_video_frame); #endif } } else { write_frame(ocodec,VIDEO_ID,pkt); } } //音频 else if (pkt.stream_index == audio_stream_idx) { //假设是音频须要编解码 if(audio_codecID != icodec->streams[audio_stream_idx]->codec->codec_id || 1 != icodec->streams[audio_stream_idx]->codec->sample_fmt) { nRet = perform_decode(AUDIO_ID,pinframe); if (nRet == 0) { //假设进和出的的声道,样本。採样率不同,须要重採样 if(icodec->streams[audio_stream_idx]->codec->sample_fmt != (AVSampleFormat)m_dwBitsPerSample || icodec->streams[audio_stream_idx]->codec->channels != m_dwChannelCount || icodec->streams[audio_stream_idx]->codec->sample_rate != m_dwFrequency) { if (swr_ctx == NULL) { swr_ctx = init_pcm_resample(pinframe,pout_audio_frame); } preform_pcm_resample(swr_ctx,pinframe,pout_audio_frame); #ifdef AUDIO_FIFO perform_code2(AUDIO_ID,pout_audio_frame); #else perform_code(AUDIO_ID,pout_audio_frame); #endif } else { pinframe->pts = pinframe->pkt_pts; #ifdef AUDIO_FIFO perform_code2(AUDIO_ID,pinframe); #else perform_code(AUDIO_ID,pinframe); #endif } } } else { write_frame(ocodec,AUDIO_ID,pkt); } } } if (pinframe) { avcodec_free_frame(&pinframe); pinframe = NULL; } if (pout_video_frame) { avcodec_free_frame(&pout_video_frame); pout_video_frame = NULL; } uinit_pcm_resample(pout_audio_frame,swr_ctx); return 1; }
实现效果:
源代码地址:http://download.csdn.net/detail/zhuweigangzwg/9456782
交流请加QQ群:62054820
QQ:379969650.
以上是关于ffmpeg转码本地文件的主要内容,如果未能解决你的问题,请参考以下文章
使用 FFmpeg 在 iOS 上编写时将 fMP4 转码为 HLS
音视频开发8. 使用ffmpeg 将pcm转码aac实践(C++)