Improve progress reporting for audio separation

Refactored process_video_with_progress to handle structured progress updates using value and description, and updated separate_tracks to yield progress as dictionary events with normalized float values. Enhanced clarity and granularity of progress feedback throughout the audio separation workflow.

app.py

@@ -199,7 +199,8 @@ def process_video_with_progress(
     status_lines = []
 
     def on_progress(message):
-        status_lines.append(message)
+        if isinstance(message, str):
+            status_lines.append(message)
 
     try:
         if cookies_upload is not None:            
@@ -227,7 +228,7 @@ def process_video_with_progress(
         progress(0.4, desc="Separating tracks")
         yield "", "\n".join(status_lines)
 
-        separation = separate_tracks(audio_path, job_id, progress_callback=on_progress)
+        separation = separate_tracks(audio_path, job_id)
         while True:
             try:
                 event_type, payload = next(separation)
@@ -235,7 +236,7 @@ def process_video_with_progress(
                 break
 
             if event_type == "progress":
-                status_lines.append(str(payload))
+                progress(payload["value"], desc=payload["desc"])
                 yield "", "\n".join(status_lines)
             elif event_type == "result":
                 drums, vocals, guitar, bass, other, piano, music, full = payload

modules/yt_audio_get_tracks.py

@@ -163,12 +163,13 @@ def separate_tracks(input_wav, job_id, progress_callback=None):
     if not os.path.exists(input_wav):
         raise FileNotFoundError(f"{input_wav} does not exist")
 
-    def _emit(msg):
-        _emit_progress(progress_callback, msg)
-        return ("progress", msg)
+    def _emit_progress_event(value, desc):
+        if progress_callback is not None:
+            progress_callback(desc)
+        return ("progress", {"value": value, "desc": desc})
 
-    yield _emit("Validating input audio...")
-    yield _emit("Starting Demucs separation (htdemucs_6s)...")
+    yield _emit_progress_event(0.0, "Validating input audio...")
+    yield _emit_progress_event(0.05, "Starting Demucs separation (htdemucs_6s)...")
 
     output_dir = 'separated'
 
@@ -178,6 +179,7 @@ def separate_tracks(input_wav, job_id, progress_callback=None):
     ], stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True, bufsize=1)
 
     progress_pattern = re.compile(r'(\d+)%\|')
+    last_percent = None
 
     # Read progress in real time
     for line in proc.stdout or []:
@@ -186,15 +188,20 @@ def separate_tracks(input_wav, job_id, progress_callback=None):
             match = progress_pattern.search(line)
             if match:
                 percent = int(match.group(1))
-                yield _emit(f"Demucs progress: {percent}%")
+                if percent != last_percent:
+                    last_percent = percent
+                    yield _emit_progress_event(
+                        0.1 + (percent / 100.0) * 0.7,
+                        f"Demucs progress: {percent}%",
+                    )
             elif "Separating track" in line:
-                yield _emit("Demucs: Starting separation...")
+                yield _emit_progress_event(0.1, "Demucs: Starting separation...")
 
     proc.wait()
     if proc.returncode != 0:
         raise RuntimeError("Demucs failed")
 
-    yield _emit("Demucs separation complete. Loading stems...")
+    yield _emit_progress_event(0.82, "Demucs separation complete. Loading stems...")
 
     base = os.path.join(output_dir, 'htdemucs_6s', job_id)
 
@@ -205,20 +212,20 @@ def separate_tracks(input_wav, job_id, progress_callback=None):
     piano = f'{base}/piano.mp3'
     other = f'{base}/other.mp3'
 
-    yield _emit("Creating music stem (bass + other)...")
+    yield _emit_progress_event(0.88, "Creating music stem (bass + other)...")
     music = AudioSegment.from_mp3(bass).overlay(AudioSegment.from_mp3(other))
     music_path = os.path.join(base, 'music.mp3')
     music.export(music_path, format="mp3")
 
     # Full mix export (rest of your code)
     full_path = os.path.join(base, 'full.mp3')
-    yield _emit("Exporting full mix...")
+    yield _emit_progress_event(0.94, "Exporting full mix...")
     src = AudioSegment.from_file(input_wav)
     src.export(full_path, format="mp3")
 
-    yield _emit("Cleaning up...")
+    yield _emit_progress_event(0.98, "Cleaning up...")
     os.remove(input_wav)
-    yield _emit("Separation complete.")
+    yield _emit_progress_event(1.0, "Separation complete.")
 
     yield ("result", (drums, vocals, guitar, bass, other, piano, music_path, full_path))