Recherche avancée

Sur d’autres sites (6158)

• ffprobe (ffmpeg) seek in audio wave file

  26 avril 2019, par loretoparisi

  I’m using ffprobe to seek to position on mp3 files, and it works ok. This is how I do in my ffprobe node.js wrapper :

  seek = function (fpath, seconds) {
  
        var self = this;
  
        return new Promise((resolve, reject) => {
  
          //ffprobe -i /myfile.mp3 -show_frames -show_entries frame=pkt_pos -of default=noprint_wrappers=1:nokey=1 -hide_banner -loglevel panic -read_intervals 20%+#1
  
          var loglevel = self.logger.isDebug() ? 'debug' : 'panic';
  
          const args = [
  
            '-hide_banner',
  
            '-loglevel', loglevel,
  
            '-show_frames',//Display information about each frame
  
            '-show_entries', 'frame=pkt_pos',// Display only information about byte position
  
            '-of', 'default=noprint_wrappers=1:nokey=1',//Don't want to print the key and the section header and footer
  
            '-read_intervals', seconds + '%+#1', //Read only 1 packet after seeking to position 01:23
  
            '-print_format', 'json',
  
            '-v', 'quiet',
  
            '-i', fpath
  
          ];
  
          const opts = {
  
            cwd: self._options.tempDir
  
          };
  
          const cb = (error, stdout) => {
  
            if (error)
  
              return reject(error);
  
            try {
  
              const outputObj = JSON.parse(stdout);
  
              return resolve(outputObj);
  
            } catch (ex) {
  
              self.logger.error("seek failed %s", ex);
  
              return reject(ex);
  
            }
  
          };
  
          cp.execFile('ffprobe', args, opts, cb)
  
            .on('error', reject);
  
        });
  
      }//seek

  So to seek to a specified startTime position in seconds you do

  seek(path,startTime)

  and then you will get the equivalent in bytes looking at the packet of the first frames in frames like :

  position  = function() {
  
        if(this.raw.frames) {
  
          if(this.raw.frames.length>0) {// get first stream ref
  
            if(this.raw.frames[0].pkt_pos) {// get first stream ref
  
              return this.raw.frames[0].pkt_pos;
  
            }
  
          }
  
        }
  
        return '';
  
      }//position

  So, when doing seek on a mp3 file it works ok and the range query is

  rangeBytes bytes=80685-8000685 {
  
    "url": "",
  
    "path": "",
  
    "raw": {
  
      "frames": [
  
        {
  
          "pkt_pos": "80685"
  
        }
  
      ]
  
    }
  
  } {
  
    "url": "",
  
    "path": "",
  
    "raw": {
  
      "frames": [
  
        {
  
          "pkt_pos": "8000685"
  
        }
  
      ]
  
    }
  
  }

  the problem is that when I use this on a wave file, the resulting range query for the wave file :

  rangeBytes bytes=768058-76800058 {
  
    "url": "",
  
    "path": "",
  
    "raw": {
  
      "frames": [
  
        {
  
          "pkt_pos": "768058"
  
        }
  
      ]
  
    }
  
  } {
  
    "url": "",
  
    "path": "",
  
    "raw": {
  
      "frames": [
  
        {
  
          "pkt_pos": "76800058"
  
        }
  
      ]
  
    }
  
  }

  will not work in the standard HTML5 audio player.

  NOTE
  The ffprobe command line command that I’m using with the wrapper above was :

  ffprobe -i /myfile.mp3 -show_frames -show_entries frame=pkt_pos -of default=noprint_wrappers=1:nokey=1 -hide_banner -loglevel panic -read_intervals 20%+#1

  my guess is if this applies to both mp3 and wave files.

• FFMPEG/libavfilter drawtext scaling without affecting video [duplicate]

  23 avril 2019, par Captain Jack

  This question already has an answer here :

  • Overlaying text on video with required angle using FFMPEG
    
    2 answers
    

  How can I scale drawtext without affecting the input video ?

  Here’s my attempt - this filter is applied in C but is equivalent to -filter_complex in command line.

  [in]drawtext=text='Test Text': fontcolor=white: fontsize=w/40: x=w/20: y=h*16/18:shadowx=1:shadowy=1,scale=iw/2:ih/3[out]

  The above does scale text but it also affects video as it’s part of the drawtext input.

  I tried things like :

  nullsrc=s=iwxih[ns];
  
  [ns]drawtext=text='Test Text': fontcolor=white: fontsize=w/40: x=w/20: y=h*16/18:shadowx=1:shadowy=1,scale=iw/2:ih/3[text];
  
  [in][text]overlay=0:0[out]

  But it doesn’t seem to like nullsrc and it hangs... or I am doing something wrong.

  Any ideas ?

• Sox mute white noise silences keeping the lenght of the audio file [on hold]

  1er mars 2019, par Lya1981

  I have an audio which is 200 seconds long.
  I run the following command, which removes silences within the threshold specified :

  sox in.wav out.wav silence 1 0.1 1% -1 0.5 1%

  Once those silences are removed, the audio becomes 100 seconds long, which means I am losing my original time stamps.

  I need to basically blank out / mute those silences (they are variations of white noise), leaving equivalent gaps in their place in order to keep the original length and timestamps within the audio.

  Is there any way to do it with exactly the params above, just not trimming it but blanking it out ? They produce the perfect outcome but I really need to keep the time stamps...
  Thank you in advance !