Tune chord detection - less aggressive filtering

- hop_length: 2048 → 1024 (finer ~46ms resolution)
- smoothing: 21 → 11 frames (~0.5s instead of ~2s)
- confidence: 0.20 → 0.15 (catch more changes)
- min_duration: 1.0 → 0.5 seconds (allow half-bar changes)

Should detect 30-80 chords per song instead of 5 or 993.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

modules/chords.py

@@ -270,13 +270,13 @@ def match_chord_template_with_confidence(chroma_frame, templates, focus='harmony
     return best_chord, best_score
 
 
-def extract_chords(audio_path, min_duration=1.0):
+def extract_chords(audio_path, min_duration=0.5):
     """
     Extract chords from audio file with musical timing.
 
     Args:
         audio_path: Path to audio file
-        min_duration: Minimum chord duration in seconds (default 1.0s = reasonable for most music)
+        min_duration: Minimum chord duration in seconds (default 0.5s for half-bar changes)
 
     Returns:
         List of (timestamp, chord_name) tuples
@@ -288,15 +288,15 @@ def extract_chords(audio_path, min_duration=1.0):
         # Load audio
         y, sr = librosa.load(audio_path, sr=22050, duration=None)
 
-        # Use larger hop for coarser time resolution (fewer false changes)
-        # hop_length=2048 at 22050Hz = ~93ms per frame
-        hop_length = 2048
+        # Use moderate hop for reasonable time resolution
+        # hop_length=1024 at 22050Hz = ~46ms per frame
+        hop_length = 1024
 
         # Extract chroma features
         chroma = librosa.feature.chroma_cqt(y=y, sr=sr, hop_length=hop_length)
 
-        # Heavy smoothing to reduce noise - median filter across 21 frames (~2 seconds)
-        chroma = scipy.ndimage.median_filter(chroma, size=(1, 21))
+        # Moderate smoothing - median filter across 11 frames (~0.5 seconds)
+        chroma = scipy.ndimage.median_filter(chroma, size=(1, 11))
 
         templates = create_chord_templates()
 
@@ -312,7 +312,7 @@ def extract_chords(audio_path, min_duration=1.0):
         consolidated = []
         current_chord = None
         current_start = 0.0
-        confidence_threshold = 0.20  # Balanced threshold - not too noisy, not too strict
+        confidence_threshold = 0.15  # Lower threshold to catch more chord changes
 
         for time, chord, conf in raw_chords:
             if conf >= confidence_threshold: