diff --git a/mediaplayer/src/jvmMain/kotlin/io/github/kdroidfilter/composemediaplayer/mac/MacVideoPlayerState.kt b/mediaplayer/src/jvmMain/kotlin/io/github/kdroidfilter/composemediaplayer/mac/MacVideoPlayerState.kt
index 035445b6..f98d3575 100644
--- a/mediaplayer/src/jvmMain/kotlin/io/github/kdroidfilter/composemediaplayer/mac/MacVideoPlayerState.kt
+++ b/mediaplayer/src/jvmMain/kotlin/io/github/kdroidfilter/composemediaplayer/mac/MacVideoPlayerState.kt
@@ -62,6 +62,9 @@ internal val macLogger = TaggedLogger("MacVideoPlayerState")
  */
 private const val POSITION_POLL_INTERVAL_MS = 200L
 
+// Polls (250 ms apart) without a video size before media with an audio track counts as audio-only.
+private const val AUDIO_ONLY_MIN_POLLS = 8
+
 /**
  * macOS implementation of the video player state.
  *
@@ -418,15 +421,19 @@ class MacVideoPlayerState : VideoPlayerState {
                     try {
                         val w = MacNativeBridge.nGetFrameWidth(ptr)
                         val h = MacNativeBridge.nGetFrameHeight(ptr)
-                        if (w <= 0 || h <= 0) {
+                        // Audio-only media (an mp3, an m4a) has no video size but still plays: only a
+                        // source with neither a picture nor any audio or duration is an error.
+                        val isAudioOnly = (w <= 0 || h <= 0) &&
+                            (resolvedDuration > 0.0 || MacNativeBridge.nGetAudioChannels(ptr) > 0)
+                        if ((w <= 0 || h <= 0) && !isAudioOnly) {
                             setError("Failed to retrieve video size")
                             return@withLock
                         }
-                        videoWidth = w
-                        videoHeight = h
+                        videoWidth = w.coerceAtLeast(0)
+                        videoHeight = h.coerceAtLeast(0)
 
                         // Scale output to match the display surface (saves memory for 4K+ video).
-                        if (surfaceWidth > 0 && surfaceHeight > 0) {
+                        if (!isAudioOnly && surfaceWidth > 0 && surfaceHeight > 0) {
                             videoReaderMutex.withLock {
                                 MacNativeBridge.nSetOutputSize(ptr, surfaceWidth, surfaceHeight)
                                 val sw = MacNativeBridge.nGetFrameWidth(ptr)
@@ -461,6 +468,8 @@ class MacVideoPlayerState : VideoPlayerState {
                             MacNativeBridge.nPlay(ptr)
                         }
                         _isPlaying = startPlayback
+                        // No frame will clear the loading state of audio-only media.
+                        if (isAudioOnly) isLoading = false
 
                         videoJob = startVideoPipeline()
                     } catch (e: Exception) {
@@ -496,6 +505,13 @@ class MacVideoPlayerState : VideoPlayerState {
                 macLogger.d { "Dimensions validated (w=$width, h=$height) after $attempt attempts" }
                 return
             }
+            // Audio-only media never gets dimensions: once the asset reports an audio track and a
+            // while has passed without a video size, stop waiting (a late video size is still
+            // picked up by the frame pipeline, which keeps reading).
+            if (attempt >= AUDIO_ONLY_MIN_POLLS && MacNativeBridge.nGetAudioChannels(ptr) > 0) {
+                macLogger.d { "No video size after $attempt attempts but an audio track: treating as audio-only" }
+                return
+            }
             delay(250)
         }
         macLogger.e { "Unable to retrieve valid dimensions after $maxAttempts attempts" }
diff --git a/mediaplayer/src/jvmMain/native/macos/NativeVideoPlayer.swift b/mediaplayer/src/jvmMain/native/macos/NativeVideoPlayer.swift
index 36d1318a..5877ebb5 100644
--- a/mediaplayer/src/jvmMain/native/macos/NativeVideoPlayer.swift
+++ b/mediaplayer/src/jvmMain/native/macos/NativeVideoPlayer.swift
@@ -726,16 +726,25 @@ class MacVideoPlayer {
         // Retrieve the video track to obtain the actual dimensions
         asset.loadTracks(withMediaType: .video) { [self] tracks, error in
             guard let videoTrack = tracks?.first, error == nil else {
-                print(
-                    "Error loading video tracks: \(error?.localizedDescription ?? "Unknown")"
-                )
                 // For HLS streams without video track info yet, use default dimensions
                 if isHLSStream {
+                    print("No video track info yet for the HLS stream: \(error?.localizedDescription ?? "none")")
                     frameWidth = 1920
                     frameHeight = 1080
                     nativeVideoWidth = frameWidth
                     nativeVideoHeight = frameHeight
                     setupVideoOutputAndPlayer(with: asset)
+                } else if error == nil {
+                    // Audio-only media (an MP3, an M4A): no video track, but it plays. The player is set up with
+                    // no video output; the Kotlin side reports a 0x0 size as audio-only.
+                    print("No video track: playing as audio-only")
+                    frameWidth = 0
+                    frameHeight = 0
+                    nativeVideoWidth = 0
+                    nativeVideoHeight = 0
+                    setupVideoOutputAndPlayer(with: asset, hasVideo: false)
+                } else {
+                    print("Error loading video tracks: \(error?.localizedDescription ?? "Unknown")")
                 }
                 return
             }
@@ -845,15 +854,20 @@ class MacVideoPlayer {
     }
 
     // Helper method to setup video output and player
-    private func setupVideoOutputAndPlayer(with asset: AVAsset, videoComposition: AVVideoComposition? = nil) {
-        // Create attributes for the CVPixelBuffer (BGRA format) with IOSurface for better performance
-        let pixelBufferAttributes: [String: Any] = [
-            kCVPixelBufferPixelFormatTypeKey as String: kCVPixelFormatType_32BGRA,
-            kCVPixelBufferWidthKey as String: frameWidth,
-            kCVPixelBufferHeightKey as String: frameHeight,
-            kCVPixelBufferIOSurfacePropertiesKey as String: [:],
-        ]
-        videoOutput = AVPlayerItemVideoOutput(pixelBufferAttributes: pixelBufferAttributes)
+    private func setupVideoOutputAndPlayer(with asset: AVAsset, videoComposition: AVVideoComposition? = nil, hasVideo: Bool = true) {
+        if hasVideo {
+            // Create attributes for the CVPixelBuffer (BGRA format) with IOSurface for better performance
+            let pixelBufferAttributes: [String: Any] = [
+                kCVPixelBufferPixelFormatTypeKey as String: kCVPixelFormatType_32BGRA,
+                kCVPixelBufferWidthKey as String: frameWidth,
+                kCVPixelBufferHeightKey as String: frameHeight,
+                kCVPixelBufferIOSurfacePropertiesKey as String: [:],
+            ]
+            videoOutput = AVPlayerItemVideoOutput(pixelBufferAttributes: pixelBufferAttributes)
+        } else {
+            // Audio-only: no frames to read (captureFrame and captureInitialFrame need an output and return early).
+            videoOutput = nil
+        }
 
         let item = AVPlayerItem(asset: asset)