function metadataOutput of protocol AVPlayerItemMetadataOutputPushDelegate is not always being triggered

Code Block swift
import UIKit
import AVFoundation
class ViewController: UIViewController, AVPlayerItemMetadataOutputPushDelegate {
  private var player: AVPlayer!
  private var playerLayer: AVPlayerLayer!
  private var playerItem: AVPlayerItem!
   
  @IBOutlet weak var videoView: UIView!
   func prepareToPlay(url: URL) {
    let asset = AVAsset(url: url)
    playerItem = AVPlayerItem(asset: asset)
    player = AVPlayer(playerItem: playerItem)
     
    playerLayer.player = player
    let metadataOutput = AVPlayerItemMetadataOutput(identifiers: nil)
    metadataOutput.setDelegate(self, queue: DispatchQueue.main)
    playerItem.add(metadataOutput)
    player.play()
  }
   
  override func viewDidLoad() {
    super.viewDidLoad()
    playerLayer = AVPlayerLayer(player: player)
    playerLayer.frame = self.videoView.bounds
    playerLayer.videoGravity = .resizeAspect
    self.videoView.layer.addSublayer(playerLayer)
     
    if let url = URL(string: "https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/bipbop_16x9_variant.m3u8") {
      prepareToPlay(url: url)
    }
  }
  @available(iOS 8.0, *)
  func metadataOutput(_ output: AVPlayerItemMetadataOutput, didOutputTimedMetadataGroups groups: [AVTimedMetadataGroup], from track: AVPlayerItemTrack?) {
    print("metadata, groups : \(groups[0].timeRange)")
    if let item = groups.first?.items.first
    {
      print("item:\(item)")
      item.value(forKeyPath: #keyPath(AVMetadataItem.value))
      print("raw Metadata value: \n \(item.value(forKeyPath: #keyPath(AVMetadataItem.value))!)")
    }
  }

Change to a better way to make sure asset and playeritem is ready to play, then play.

But in these codes, I missed ID3 of first segment which is time 0 but got the following ID3 data from 5s and so on.

It only happen once for about 50 times of tests, possibility is really low. But I still wonder why there is this situation.

But from my own HLS stream , problem still exist.

import UIKit
import AVFoundation

class ViewController: UIViewController, AVPlayerItemMetadataOutputPushDelegate {

  private let player = AVPlayer()
  private var playerItemContext = 0
   
  @IBOutlet weak var videoView: UIView!
  @IBOutlet weak var playAndPauseButton: UIButton!
   
  @IBAction func playAndPause(_ sender: Any) {
    if (player.timeControlStatus == .paused) {
      playAndPauseButton.setImage(UIImage(systemName: "pause.rectangle.fill"), for: .normal)
      player.play()
    } else if (player.timeControlStatus == .playing) {
      playAndPauseButton.setImage(UIImage(systemName: "play.rectangle.fill"), for: .normal)
      player.pause()
    } else {
      print("unexpected timeControl status :\(player.timeControlStatus.rawValue)")
    }
  }
   
  override func observeValue(forKeyPath keyPath: String?,
                of object: Any?,
                change: [NSKeyValueChangeKey : Any]?,
                context: UnsafeMutableRawPointer?) {
    // Only handle observations for the playerItemContext
    guard context == &playerItemContext else {
      super.observeValue(forKeyPath: keyPath,
                of: object,
                change: change,
                context: context)
      return
    }

    if keyPath == #keyPath(AVPlayerItem.status) {
      let status: AVPlayerItem.Status

      // Get the status change from the change dictionary
      if let statusNumber = change?[.newKey] as? NSNumber {
        status = AVPlayerItem.Status(rawValue: statusNumber.intValue)!
      } else {
        status = .unknown
      }

      // Switch over the status
      switch status {
      case .readyToPlay:
        print("ready to play")
        self.playAndPause(self)
        break
      // Player item is ready to play.
      case .failed: break
      // Player item failed. See error.
      case .unknown: break
        // Player item is not yet ready.
      }
    }
  }
   
   
   
  private func initPlayerAsset(with url: URL, completion: ((_ asset: AVAsset) -> Void)?) {
    let asset = AVAsset(url: url)
    asset.loadValuesAsynchronously(forKeys: ["playable"]) {
      completion?(asset)
    }
  }
  func prepareToPlay(url: URL) {
    initPlayerAsset(with: url, completion: { (asset: AVAsset) in
      let playerItem = AVPlayerItem(asset: asset)

      let metadataOutput = AVPlayerItemMetadataOutput(identifiers: nil)
      metadataOutput.setDelegate(self, queue: DispatchQueue.main)
      playerItem.add(metadataOutput)

      playerItem.addObserver(self,
                  forKeyPath: #keyPath(AVPlayerItem.status),
                  options: [.old, .new],
                  context: &self.playerItemContext)
       
      self.player.replaceCurrentItem(with: playerItem)
    })
  }
   
  override func viewDidLoad() {
    super.viewDidLoad()

    let playerLayer = AVPlayerLayer(player: player)
    playerLayer.frame = self.videoView.bounds
    playerLayer.videoGravity = .resizeAspect

    self.videoView.layer.addSublayer(playerLayer)
     
    if let url = URL(string: "https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/bipbop_16x9_variant.m3u8") {
      prepareToPlay(url: url)
    }
  }

  @available(iOS 8.0, *)
  func metadataOutput(_ output: AVPlayerItemMetadataOutput, didOutputTimedMetadataGroups groups: [AVTimedMetadataGroup], from track: AVPlayerItemTrack?) {
    print("metadata, groups : \(groups[0].timeRange)")
    if let item = groups.first?.items.first
    {
      print("item:\(item)")
      item.value(forKeyPath: #keyPath(AVMetadataItem.value))
      print("raw Metadata value: \n \(item.value(forKeyPath: #keyPath(AVMetadataItem.value))!)")
    } else {
      print("MetaData Error")
    }
  }
}

Sample Url

https://ryanyao-demo.s3-ap-northeast-1.amazonaws.com/output/master.m3u8

After weeks of try and verification, found that it I insert metadata track in video segment will not raise this problem. Only when m3u8 file is A/V split and metadata is in audio track, this callback will not be triggered every time. I am wondering if there is the reason that there are audio and video segments when assign AVPlayerItemOutput to player item and the binding may not work as expected for both audio and video