05 - FFmpeg 提取 PCM 音频裸数据
----------------------------------------------------------------- PCM介绍 ----------------------------------------------------------------
PCM(Pulse Code Modulation)，脉冲编码调制，是一种用数字表示采样模拟信号的方法。
核心过程:采样-->量化-->编码
----------------------------------------------------------------- PCM关键要素 -----------------------------------------------------------------
·采样率(sampleRate):每秒中采集样本的个数，如8KHz，表示每秒采样8000次。
奈奎斯特定理，明确:按比声音最高频率高2倍以上的频率对声音进行采样;
人耳能接受的频率范围为20Hz～20kHz，故采样率一般为44.1KHz较好，采样率越高，质量越高，但存储空间增大。
·量化格式(sampleFormat) : ffmpeg支持的量化格式: ffmpeg -formats | grep PCM
声道数"(channel):单声道(mono)、双声道(stereo)
----------------------------------------------------------------- PCM数据格式 -----------------------------------------------------------------
存储格式:
> 双声道音频文件，采样数据按LRLR方式存储，存储的时候与字节序有关。
> 单声道音频文件，采样数据按时间先后依次存入(有时也会用LRLR方式存储，但另一个声道数据为0)。
· 存储格式分为Packed和Planner两种，对于双通道音频，Packed为两个声道的数据交错存储;Planner 为两个声道数据分开存储:
> Packed: LRLRLR
> Planner: LLLRRR
----------------------------------------------------------------- PCM计算 -----------------------------------------------------------------
· 大小计算:以CD的音质为例:量化格式为16比特(2字节)，采样率为44100，声道数为2
比特率为:44100*16*2=1378.125kbps
1分钟音频大小:1378.125 * 60/8/1024=10.09MB
· ffmpeg提取pcm数据命令:
ffmpeg -i break.aac -ar 48000 -ac 2 -f s16le out.pcm
· ffplay播放pcm数据:
ffplay -ar 48000 -ac 2 -f s16le out.pcm
方法一
int decodeAudioInterface(AVCodecContext *decoderCtx, AVPacket *packet, AVFrame *frame, FILE *dest_fp)
{
int ret = avcodec_send_packet(decoderCtx, packet);
if (ret < 0)
{
av_log(NULL, AV_LOG_ERROR, "send packet to decoder failed:%s\n", av_err2str(ret));
return -1;
}
while (ret >= 0)
{
ret = avcodec_receive_frame(decoderCtx, frame);
if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF)
{
av_log(NULL, AV_LOG_WARNING, "[decodeAudioInterface] -- AVERROR(EAGAIN) || AVERROR_EOF \n", av_err2str(ret));
return 0;
}
else if (ret < 0)
{
av_log(NULL, AV_LOG_ERROR, "decode packet failed:%s\n", av_err2str(ret));
}
// frame -- fltp
int dataSize = av_get_bytes_per_sample(decoderCtx->sample_fmt);
if (dataSize < 0)
{
av_log(NULL, AV_LOG_ERROR, "get bytes failed!\n");
return -1;
}
for (int i = 0; i < frame->nb_samples; i++)
{
for (int channel = 0; channel < decoderCtx->channels; channel++)
{
fwrite(frame->data[channel] + dataSize * i, 1, dataSize, dest_fp);
}
}
}
return 0;
}
int decodeAudio(const char *inFileName, const char *outFileName)
{
/********************************************************************/
FILE *dest_fp = fopen(outFileName, "wb+");
if (dest_fp == NULL)
{
av_log(NULL, AV_LOG_ERROR, "open outfile %s failed!\n", outFileName);
goto end;
}
/********************************************************************/
AVFormatContext *inFmtCtx = NULL;
int ret = avformat_open_input(&inFmtCtx, inFileName, NULL, NULL);
if (ret != 0)
{
av_log(NULL, AV_LOG_ERROR, "open input file failed:%s\n", av_err2str(ret));
return -1;
}
ret = avformat_find_stream_info(inFmtCtx, NULL);
if (ret < 0)
{
av_log(NULL, AV_LOG_ERROR, "find stream info failed:%s\n", av_err2str(ret));
goto end;
}
int audioIndex = av_find_best_stream(inFmtCtx, AVMEDIA_TYPE_AUDIO, -1, -1, NULL, 0);
if (audioIndex < 0)
{
av_log(NULL, AV_LOG_ERROR, "find bast stream failed:%s\n", av_err2str(audioIndex));
goto end;
}
AVCodecContext *decoderCtx = avcodec_alloc_context3(NULL);
if (decoderCtx == NULL)
{
av_log(NULL, AV_LOG_ERROR, "avcodec alloc context failed\n");
ret = -1;
goto end;
}
// 拷贝编码参数
avcodec_parameters_to_context(decoderCtx, inFmtCtx->streams[audioIndex]->codecpar);
AVCodec *decoder = avcodec_find_decoder(decoderCtx->codec_id);
if (decoder == NULL)
{
av_log(NULL, AV_LOG_ERROR, "find decoder %d failed!\n", decoderCtx->codec_id);
}
ret = avcodec_open2(decoderCtx, decoder, NULL);
if (ret < 0)
{
av_log(NULL, AV_LOG_ERROR, "open decoder failed:%s\n", av_err2str(ret));
goto end;
}
AVPacket packet;
av_init_packet(&packet);
AVFrame *frame = av_frame_alloc();
int frameSize = av_samples_get_buffer_size(NULL, decoderCtx->channels, frame->nb_samples, decoderCtx->sample_fmt, 1);
uint8_t *frameBuffer = av_malloc(frameSize);
avcodec_fill_audio_frame(frame, decoderCtx->channels, decoderCtx->sample_fmt, frameBuffer, frameSize, 1);
while (av_read_frame(inFmtCtx, &packet) >= 0)
{
if (packet.stream_index == audioIndex)
{
decodeAudioInterface(decoderCtx, &packet, frame, dest_fp);
}
av_packet_unref(&packet);
}
decodeAudioInterface(decoderCtx, NULL, frame, dest_fp);
end:
if (inFmtCtx)
{
avformat_close_input(&inFmtCtx);
}
if (decoderCtx)
{
avcodec_free_context(&decoderCtx);
}
if (frame)
{
av_frame_free(&frame);
}
if (frameBuffer)
{
av_freep(frameBuffer);
}
if (dest_fp)
{
fclose(dest_fp);
}
return 0;
}
====================================================================================
方法二、
#define AUDIO_INBUF_SIZE 20480
#define AUDIO_REFLT_THRESH 4096
int decodeAudioInterface(AVCodecContext *decoderCtx, AVPacket *packet, AVFrame *frame, FILE *dest_fp)
{
int ret = avcodec_send_packet(decoderCtx, packet);
if (ret == AVERROR(EAGAIN))
{
av_log(NULL, AV_LOG_WARNING, "[decodeAudioInterface] -- AVERROR(EAGAIN) \n");
}
else if (ret < 0)
{
av_log(NULL, AV_LOG_ERROR, "send packet to decoder failed: %s\n", av_err2str(ret));
return -1;
}
while (ret >= 0)
{
// 对于frame avcodec_receive_frame 内部每次都先调用
ret = avcodec_receive_frame(decoderCtx, frame);
if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF)
{
return;
}
else if (ret < 0)
{
av_log(NULL, AV_LOG_ERROR, "decode packet failed:%s\n", av_err2str(ret));
}
// 获取单个sample 占用的字节
size_t dataSize = av_get_bytes_per_sample(decoderCtx->sample_fmt);
if (dataSize < 0)
{
/*This should not occur,checking just for paranoia*/
av_log(NULL, AV_LOG_ERROR, "get bytes failed!\n");
return -1;
}
av_log(NULL, AV_LOG_INFO,"采样率: %uHZ, 通道：%u, 编码格式：%u \n", frame->sample_rate,frame->channels, frame->format);
for (int i = 0; i < frame->nb_samples; i++)
{
for (int channel = 0; channel < decoderCtx->channels; channel++) // 交错的方式写入，float的格式输出
{
fwrite(frame->data[channel] + dataSize * i, 1, dataSize, dest_fp);
}
}
}
}
// ffplay -ar 48000 -ac 2 -f f32le outTest.pcm
int decodeAudio(const char *inFileName, const char *outFileName)
{
/************************************************************************************/
// 打开输入文件
FILE *inFile = fopen(inFileName, "rb");
if (!inFile)
{
av_log(NULL, AV_LOG_ERROR, "Could not open:%s\n", inFileName);
goto _end;
}
// 打开输出文件
FILE *outFile = fopen(outFileName, "wb");
if (!outFile)
{
av_log(NULL, AV_LOG_ERROR, "Could not open:%s\n", inFileName);
goto _end;
}
uint8_t inbuf[AUDIO_INBUF_SIZE + AV_INPUT_BUFFER_PADDING_SIZE];
uint8_t *data = inbuf;
size_t dataSize = fread(inbuf, 1, AUDIO_INBUF_SIZE, inFile);
/************************************************************************************/
enum AVCodecID AudioCodecID = AV_CODEC_ID_AAC;
if (strstr(inFileName, "aac") != NULL)
{
AudioCodecID = AV_CODEC_ID_AAC;
}
else if (strstr(inFileName, "mp3") != NULL)
{
AudioCodecID = AV_CODEC_ID_MP3;
}
else
{
av_log(NULL, AV_LOG_WARNING, "default codec id:%d\n", AudioCodecID);
}
// 查找解码器
const AVCodec *decoder;
decoder = avcodec_find_decoder(AudioCodecID); // AV_CODEC_ID_AAC
if (!decoder)
{
av_log(NULL, AV_LOG_ERROR, "decoder not found\n");
goto _end;
}
// 获得裸流的解析器 -- 根据制定的解码器ID初始化相应裸流的解析器
AVCodecParserContext *parserCtx = av_parser_init(decoder->id);
if (!parserCtx)
{
av_log(NULL, AV_LOG_ERROR, "parserCtx not found\n");
goto _end;
}
AVCodecContext *decoderCtx = NULL;
decoderCtx = avcodec_alloc_context3(decoder);
if (!decoderCtx)
{
av_log(NULL, AV_LOG_ERROR, "Conld not allocate audio codec context\n");
goto _end;
}
// 将解码器和解码器上下文进行关联
int ret = avcodec_open2(decoderCtx, decoder, NULL);
if (ret < 0)
{
av_log(NULL, AV_LOG_ERROR, "Could not open codec\n");
goto _end;
}
AVFrame *decodeFrame = av_frame_alloc();
if (decodeFrame == NULL)
{
av_log(NULL, AV_LOG_ERROR, "Could not allocate audio decodeFrame\n");
goto _end;
}
AVPacket *packet = NULL;
packet = av_packet_alloc();
while (dataSize > 0)
{
ret = av_parser_parse2(parserCtx, decoderCtx, &packet->data, &packet->size, data, dataSize, AV_NOPTS_VALUE, AV_NOPTS_VALUE, 0);
if (ret < 0)
{
av_log(NULL, AV_LOG_ERROR, "Error while parsing\n");
goto _end;
}
data += ret; // 跳过已经解析的数据
dataSize -= ret; // 对应的缓存大小也做相应的减小
if (packet->size)
decodeAudioInterface(decoderCtx, packet, decodeFrame, outFile);
if (dataSize < AUDIO_REFLT_THRESH) // 如果数据少了则再次读取
{
memmove(inbuf, data, dataSize); // 把之前剩的数据拷贝到 buffer 的其实位置
data = inbuf;
// 读取数据 长度： AUDIO_INBUF_SIZE - dataSize
int len = fread(data + dataSize, 1, AUDIO_INBUF_SIZE - dataSize, inFile);
if (len > 0)
dataSize += len;
}
}
_end:
if (outFile)
fclose(outFile);
if (inFile)
fclose(inFile);
if (decoderCtx)
avcodec_free_context(&decoderCtx);
if (parserCtx)
av_parser_close(parserCtx);
if (decodeFrame)
av_frame_free(&decodeFrame);
if (packet)
av_packet_free(&packet);
return 0;
}原文地址：https://blog.csdn.net/weixin_44977283/article/details/140349472
免责声明：本站文章内容转载自网络资源，如本站内容侵犯了原著者的合法权益，可联系本站删除。更多内容请关注自学内容网（zxcms.com）！