Recherche avancée

Sur d’autres sites (3690)

• How to make media recorder api individual chunks playable by it self

  10 août 2024, par tedGuy

  I'm trying to send individual chunks to the server instead of sending the whole chunks at once. This way, I have Ffmpeg on my Rails server to convert these chunks to HLS and upload them to S3 to stream the video instantly. However, I've encountered an issue the Media Recorder only provides playable chunks for the first segment after that, they are not playable and need to be concatenated to play.

  


  To avoid this, I've taken a different approach where I start a new Media Recorder every 3 seconds so that I get a playable chunk every time. However, this approach has its issues the video glitches a bit due to the delay when I stop and start a new Media Recorder. Is there a way to achieve this with ease ? This is my current status. please help !

  


  const startVideoRecording = async (
    screenStream: MediaStream,
    audioStream: MediaStream
  ) => {
    setStartingRecording(true);

    try {
      const res = await getVideoId();
      videoId.current = res;
    } catch (error) {
      console.log(error);
      return;
    }

    const outputStream = new MediaStream();
    outputStream.addTrack(screenStream.getVideoTracks()[0]);
    outputStream.addTrack(audioStream.getAudioTracks()[0]); // Add audio track

    const mimeTypes = [
      "video/webm;codecs=h264",
      "video/webm;codecs=vp9",
      "video/webm;codecs=vp8",
      "video/webm",
      "video/mp4",
    ];

    let selectedMimeType = "";
    for (const mimeType of mimeTypes) {
      if (MediaRecorder.isTypeSupported(mimeType)) {
        selectedMimeType = mimeType;
        break;
      }
    }

    if (!selectedMimeType) {
      console.error("No supported mime type found");
      return;
    }

    const videoRecorderOptions = {
      mimeType: selectedMimeType,
    };

    let chunkIndex = 0;

    const startNewRecording = () => {
      // Stop the current recorder if it's running
      if (
        videoRecorderRef.current &&
        videoRecorderRef.current.state === "recording"
      ) {
        videoRecorderRef.current.stop();
      }

      // Create a new MediaRecorder instance
      const newVideoRecorder = new MediaRecorder(
        outputStream,
        videoRecorderOptions
      );
      videoRecorderRef.current = newVideoRecorder;

      newVideoRecorder.ondataavailable = async (event) => {
        if (event.data.size > 0) {
          chunkIndex++;
          totalSegments.current++;
          const segmentIndex = totalSegments.current;
          console.log(event.data);
          handleUpload({ segmentIndex, chunk: event.data });
        }
      };

      // Start recording with a 3-second interval
      newVideoRecorder.start();
    };

    // Start the first recording
    startNewRecording();

    // Set up an interval to restart the recording every 3 seconds
    recordingIntervalIdRef.current = setInterval(() => {
      startNewRecording();
    }, 3000);

    setIsRecording(true);
    setStartingRecording(false);
  };


  


• Downloading individual songs from YouTube playlist [closed]

  23 août 2024, par user26961371
  

  • I've created a custom script using yt-dlp.

  


  • The script takes a single argument, which is the actual playlist URL.

  


  • I'm using the following command :

  


  


  yt-dlp --extract-audio --audio-format mp3 --yes-playlist -o "/Users/$USER/Desktop/%(id)s.%(ext)s" --embed-chapters $1


  


  Try :
I've run the command with a valid playlist URL, but it's downloading each song from the playlist into a single MP3 file for each song. I want to download each song as an individual file, not the entire playlist in a single file 15 times.

  


  Expectation :
I expect yt-dlp to download each song from the playlist as a separate MP3 file, rather than combining all songs into a single MP3 file for each song.

  


  Context :
The issue is likely due to the use of the --yes-playlist option, which tells yt-dlp to treat the input as a playlist URL and download all songs in one go.
I've checked the official documentation for yt-dlp, but I couldn't find a solution.

  


• Significantly different scores when calculating VMAF for the whole video vs individual frames [closed]

  24 octobre 2024, par YG1992

  Suppose i have a raw video 'encoded_raw.yuv' and 'original_raw.yuv' on YUV420p. I am getting significantly different results when im calculating VMAF using both methods

  


  Method 1

  


  $ ffmpeg -s widthxheight -pix_fmt yuv420p -i .\encoded_raw.yuv -s widthxheight -pix_fmt yuv420p -i .\original_raw.yuv -lavfi libvmaf=log_path=vmaf_logfile.json:log_fmt=json -f null -


  


  versus Method 2 : extracting the png out of a single frame and computing the VMAF score for the individual frame

  


  $ ffmpeg -video_size widthxheight -i ./encoded_raw.yuv -vf select=eq(n\,1) -vframes 1 -pix_fmt yuv420p encoded_raw_1.png
$ ffmpeg -video_size widthxheight -i ./original_raw.yuv -vf select=eq(n\,1) -vframes 1 -pix_fmt yuv420p original_raw.png
$ ffmpeg -s widthxheight -pix_fmt yuv420p -i .\encoded_raw_1.png -s widthxheight -pix_fmt yuv420p -i .\original_raw.png -lavfi libvmaf=log_path=vmaf_logfile.json:log_fmt=json -f null -`


  


  method 1 vmaf at frame 1 from json = 80

  


  method 2 vmaf = 70

  


  any reason why this is so ?

  


  Note : for moduse this is on topic as this covers
-> software tools commonly used by programmers (ie ffmpeg)