Recherche avancée

Sur d’autres sites (7715)

• How to encode resampled PCM-audio to AAC using ffmpeg-API when input pcm samples count not equal 1024

  26 septembre 2016, par Aleksei2414904

  I am working on capturing and streaming audio to RTMP server at a moment. I work under MacOS (in Xcode), so for capturing audio sample-buffer I use AVFoundation-framework. But for encoding and streaming I need to use ffmpeg-API and libfaac encoder. So output format must be AAC (for supporting stream playback on iOS-devices).

  And I faced with such problem : audio-capturing device (in my case logitech camera) gives me sample-buffer with 512 LPCM samples, and I can select input sample-rate from 16000, 24000, 36000 or 48000 Hz. When I give these 512 samples to AAC-encoder (configured for appropriate sample-rate), I hear a slow and jerking audio (seems as like pice of silence after each frame).

  I figured out (maybe I am wrong), that libfaac encoder accepts audio frames only with 1024 samples. When I set input samplerate to 24000 and resample input sample-buffer to 48000 before encoding, I obtain 1024 resampled samples. After encoding these 1024 sampels to AAC, I hear proper sound on output. But my web-cam produce 512 samples in buffer for any input samplerate, when output sample-rate must be 48000 Hz. So I need to do resampling in any case, and I will not obtain exactly 1024 samples in buffer after resampling.

  Is there a way to solve this problem within ffmpeg-API functionality ?

  I would be grateful for any help.

  PS :
  I guess that I can accumulate resampled buffers until count of samples become 1024, and then encode it, but this is stream so there will be troubles with resulting timestamps and with other input devices, and such solution is not suitable.

  The current issue came out of the problem described in [question] : How to fill audio AVFrame (ffmpeg) with the data obtained from CMSampleBufferRef (AVFoundation) ?

  Here is a code with audio-codec configs (there also was video stream but video work fine) :

      /*global variables*/
  
      static AVFrame *aframe;
  
      static AVFrame *frame;
  
      AVOutputFormat *fmt; 
  
      AVFormatContext *oc; 
  
      AVStream *audio_st, *video_st;
  
  Init ()
  
  {
  
      AVCodec *audio_codec, *video_codec;
  
      int ret;
  
  
  
      avcodec_register_all();  
  
      av_register_all();
  
      avformat_network_init();
  
      avformat_alloc_output_context2(&oc, NULL, "flv", filename);
  
      fmt = oc->oformat;
  
      oc->oformat->video_codec = AV_CODEC_ID_H264;
  
      oc->oformat->audio_codec = AV_CODEC_ID_AAC;
  
      video_st = NULL;
  
      audio_st = NULL;
  
      if (fmt->video_codec != AV_CODEC_ID_NONE) 
  
        { //…  /*init video codec*/}
  
      if (fmt->audio_codec != AV_CODEC_ID_NONE) {
  
      audio_codec= avcodec_find_encoder(fmt->audio_codec);
  
  
  
      if (!(audio_codec)) {
  
          fprintf(stderr, "Could not find encoder for '%s'\n",
  
                  avcodec_get_name(fmt->audio_codec));
  
          exit(1);
  
      }
  
      audio_st= avformat_new_stream(oc, audio_codec);
  
      if (!audio_st) {
  
          fprintf(stderr, "Could not allocate stream\n");
  
          exit(1);
  
      }
  
      audio_st->id = oc->nb_streams-1;
  
  
  
      //AAC:
  
      audio_st->codec->sample_fmt  = AV_SAMPLE_FMT_S16;
  
      audio_st->codec->bit_rate    = 32000;
  
      audio_st->codec->sample_rate = 48000;
  
      audio_st->codec->profile=FF_PROFILE_AAC_LOW;
  
      audio_st->time_base = (AVRational){1, audio_st->codec->sample_rate };
  
      audio_st->codec->channels    = 1;
  
      audio_st->codec->channel_layout = AV_CH_LAYOUT_MONO;      
  
  
  
  
  
      if (oc->oformat->flags & AVFMT_GLOBALHEADER)
  
          audio_st->codec->flags |= CODEC_FLAG_GLOBAL_HEADER;
  
      }
  
  
  
      if (video_st)
  
      {
  
      //   …
  
      /*prepare video*/
  
      }
  
      if (audio_st)
  
      {
  
      aframe = avcodec_alloc_frame();
  
      if (!aframe) {
  
          fprintf(stderr, "Could not allocate audio frame\n");
  
          exit(1);
  
      }
  
      AVCodecContext *c;
  
      int ret;
  
  
  
      c = audio_st->codec;
  
  
  
  
  
      ret = avcodec_open2(c, audio_codec, 0);
  
      if (ret < 0) {
  
          fprintf(stderr, "Could not open audio codec: %s\n", av_err2str(ret));
  
          exit(1);
  
      }
  
  
  
      //…
  
  }

  And resampling and encoding audio :

  if (mType == kCMMediaType_Audio)
  
  {
  
      CMSampleTimingInfo timing_info;
  
      CMSampleBufferGetSampleTimingInfo(sampleBuffer, 0, &timing_info);
  
      double  pts=0;
  
      double  dts=0;
  
      AVCodecContext *c;
  
      AVPacket pkt = { 0 }; // data and size must be 0;
  
      int got_packet, ret;
  
       av_init_packet(&pkt);
  
      c = audio_st->codec;
  
        CMItemCount numSamples = CMSampleBufferGetNumSamples(sampleBuffer);
  
  
  
      NSUInteger channelIndex = 0;
  
  
  
      CMBlockBufferRef audioBlockBuffer = CMSampleBufferGetDataBuffer(sampleBuffer);
  
      size_t audioBlockBufferOffset = (channelIndex * numSamples * sizeof(SInt16));
  
      size_t lengthAtOffset = 0;
  
      size_t totalLength = 0;
  
      SInt16 *samples = NULL;
  
      CMBlockBufferGetDataPointer(audioBlockBuffer, audioBlockBufferOffset, &lengthAtOffset, &totalLength, (char **)(&samples));
  
  
  
      const AudioStreamBasicDescription *audioDescription = CMAudioFormatDescriptionGetStreamBasicDescription(CMSampleBufferGetFormatDescription(sampleBuffer));
  
  
  
      SwrContext *swr = swr_alloc();
  
  
  
      int in_smprt = (int)audioDescription->mSampleRate;
  
      av_opt_set_int(swr, "in_channel_layout",  AV_CH_LAYOUT_MONO, 0);
  
  
  
      av_opt_set_int(swr, "out_channel_layout", audio_st->codec->channel_layout,  0);
  
  
  
      av_opt_set_int(swr, "in_channel_count", audioDescription->mChannelsPerFrame,  0);
  
      av_opt_set_int(swr, "out_channel_count", audio_st->codec->channels,  0);
  
  
  
      av_opt_set_int(swr, "out_channel_layout", audio_st->codec->channel_layout,  0);
  
      av_opt_set_int(swr, "in_sample_rate",     audioDescription->mSampleRate,0);
  
  
  
      av_opt_set_int(swr, "out_sample_rate",    audio_st->codec->sample_rate,0);
  
  
  
      av_opt_set_sample_fmt(swr, "in_sample_fmt",  AV_SAMPLE_FMT_S16, 0);
  
  
  
      av_opt_set_sample_fmt(swr, "out_sample_fmt", audio_st->codec->sample_fmt,  0);
  
  
  
      swr_init(swr);
  
      uint8_t **input = NULL;
  
      int src_linesize;
  
      int in_samples = (int)numSamples;
  
      ret = av_samples_alloc_array_and_samples(&input, &src_linesize, audioDescription->mChannelsPerFrame,
  
                                               in_samples, AV_SAMPLE_FMT_S16P, 0);
  
  
  
  
  
      *input=(uint8_t*)samples;
  
      uint8_t *output=NULL;
  
  
  
  
  
      int out_samples = av_rescale_rnd(swr_get_delay(swr, in_smprt) +in_samples, (int)audio_st->codec->sample_rate, in_smprt, AV_ROUND_UP);
  
  
  
      av_samples_alloc(&output, NULL, audio_st->codec->channels, out_samples, audio_st->codec->sample_fmt, 0);
  
      in_samples = (int)numSamples;
  
      out_samples = swr_convert(swr, &output, out_samples, (const uint8_t **)input, in_samples);
  
  
  
  
  
      aframe->nb_samples =(int) out_samples;
  
  
  
  
  
      ret = avcodec_fill_audio_frame(aframe, audio_st->codec->channels, audio_st->codec->sample_fmt,
  
                               (uint8_t *)output,
  
                               (int) out_samples *
  
                               av_get_bytes_per_sample(audio_st->codec->sample_fmt) *
  
                               audio_st->codec->channels, 1);
  
  
  
      aframe->channel_layout = audio_st->codec->channel_layout;
  
      aframe->channels=audio_st->codec->channels;
  
      aframe->sample_rate= audio_st->codec->sample_rate;
  
  
  
      if (timing_info.presentationTimeStamp.timescale!=0)
  
          pts=(double) timing_info.presentationTimeStamp.value/timing_info.presentationTimeStamp.timescale;
  
  
  
      aframe->pts=pts*audio_st->time_base.den;
  
      aframe->pts = av_rescale_q(aframe->pts, audio_st->time_base, audio_st->codec->time_base);
  
  
  
      ret = avcodec_encode_audio2(c, &pkt, aframe, &got_packet);
  
  
  
      if (ret < 0) {
  
          fprintf(stderr, "Error encoding audio frame: %s\n", av_err2str(ret));
  
          exit(1);
  
      }
  
      swr_free(&swr);
  
      if (got_packet)
  
      {
  
          pkt.stream_index = audio_st->index;
  
  
  
          pkt.pts = av_rescale_q(pkt.pts, audio_st->codec->time_base, audio_st->time_base);
  
          pkt.dts = av_rescale_q(pkt.dts, audio_st->codec->time_base, audio_st->time_base);
  
  
  
          // Write the compressed frame to the media file.
  
         ret = av_interleaved_write_frame(oc, &pkt);
  
         if (ret != 0) {
  
              fprintf(stderr, "Error while writing audio frame: %s\n",
  
                      av_err2str(ret));
  
              exit(1);
  
          }
  
  
  
  }

• Revision 7fd5d8e6a4 : Fix horz loopfilter loops If count was greater than 1 the src pointer would be

  18 juillet 2013, par Frank Galligan

  Changed Paths :
   Modify /vp9/common/arm/neon/vp9_loopfilter_neon.asm

  
  
  Fix horz loopfilter loops

  If count was greater than 1 the src pointer would be off on
  the second loop.

  Change-Id : I8e09037e68dc4ae92076a8067f7b6dacbbef8263

• Revision 5af6bf6c43 : Speedup loopfilter neon code. Try and cut down the cycle count by rearranging t

  19 juillet 2013, par Frank Galligan

  Changed Paths :
   Modify /vp9/common/arm/neon/vp9_loopfilter_neon.asm

  
  
  Speedup loopfilter neon code.

  Try and cut down the cycle count by rearranging the instructions
  so there are less stalls.

  Change-Id : Ic1383335ee0f05e656477d9ee9c179ec231285d5