Extract encoded samples

The MediaExtractorCompat class is a drop-in replacement for the platform's MediaExtractor class and provides identical APIs and functionality. It facilitates extraction of demuxed, typically encoded, media data from a data source.

It separates a container file (like an MP4 or MKV) into its individual tracks, such as video, audio, and subtitles. The extractor then reads the raw, encoded data from these tracks as a sequence of samples (for example, a single compressed video frame or block of audio) before they are sent to a decoder.

Common use cases include:

  • Transcoding or Remuxing: Reading encoded samples from a track to either change the codec (transcoding) or repackage the streams into a new container (remuxing), such as converting an MP4 file to MKV.
  • Selective content extraction: Isolating and saving a single track, such as extracting an audio stream from a video file.
  • Low-level debugging: Inspecting individual samples to debug file corruption, timestamp issues, or other problems.
  • Building custom players: For niche use cases, building a custom player with full control over the media pipeline.

Overview

The following code sample shows how to use MediaExtractorCompat:

Kotlin

fun extractSamples(context: Context, mediaPath: String) {
  val extractor = MediaExtractorCompat(context)
  try {
    // 1. Setup the extractor
    extractor.setDataSource(mediaPath)

    // Find and select available tracks
    for (i in 0 until extractor.trackCount) {
      val format = extractor.getTrackFormat(i)
      extractor.selectTrack(i)
    }

    // 2. Process samples
    val buffer = ByteBuffer.allocate(10 * 1024 * 1024)
    while (true) {
      // Read an encoded sample into the buffer.
      val bytesRead = extractor.readSampleData(buffer, 0)
      if (bytesRead < 0) break

      // Access sample metadata
      val trackIndex = extractor.sampleTrackIndex
      val presentationTimeUs = extractor.sampleTime
      val sampleSize = extractor.sampleSize

      extractor.advance()
    }
  } catch (e: IOException) {
    handleFailure(e)
  } finally {
    // 3. Release the extractor
    extractor.release()
  }
}

ExtractSamples.kt

Java

public void extractSamples(Context context, String mediaPath) {
  MediaExtractorCompat extractor = new MediaExtractorCompat(context);
  try {
    // 1. Setup the extractor
    extractor.setDataSource(mediaPath);

    // Find and select available tracks
    for (int i = 0; i < extractor.getTrackCount(); i++) {
      MediaFormat format = extractor.getTrackFormat(i);
      extractor.selectTrack(i);
    }

    // 2. Process samples
    ByteBuffer buffer = ByteBuffer.allocate(10 * 1024 * 1024);
    while (true) {
      // Read an encoded sample into the buffer.
      int bytesRead = extractor.readSampleData(buffer, 0);
      if (bytesRead < 0) {
        break;
      }

      // Access sample metadata
      int trackIndex = extractor.getSampleTrackIndex();
      long presentationTimeUs = extractor.getSampleTime();
      long sampleSize = extractor.getSampleSize();

      extractor.advance();
    }
  } catch (IOException e) {
    handleFailure(e);
  } finally {
    // 3. Release the extractor
    extractor.release();
  }
}

ExtractSamples.java