Update app.py

app.py

@@ -49,84 +49,91 @@ SOUDFONT_PATH = 'SGM-v2.01-YamahaGrand-Guit-Bass-v2.7.sf2'
 
 #==================================================================================
 
-print('Loading MIDI Loops Small Dataset...')
-print('=' * 70)
-
-midi_loops_dataset = pickle.load(open(hf_hub_download(repo_id='asigalov61/MIDI-Loops', 
-                                                 filename='MIDI_Loops_Processed_Dataset_116909_Loops_CC_BY_NC_SA.pickle',
-                                                 repo_type='dataset'
-                                                ), 
-                                      'rb')
-                                )
-
-print('=' * 70)
-print('Done!')
-print('=' * 70)
-print('Loaded', len(midi_loops_dataset), 'MIDI Loops')
+def split_by_sublist(main_list, sublist):
+
+    n = len(sublist)
+    indices = []
+    result = []
+
+    # Find all starting indices where sublist matches main_list
+    for i in range(len(main_list) - n + 1):
+        if main_list[i:i + n] == sublist:
+            indices.append(i)
+
+    if not indices:
+        # If the sublist is not found, return the entire main_list as one segment
+        return [main_list]
+
+    # Use the indices to split the main_list
+    for idx, start_index in enumerate(indices):
+        if idx < len(indices) - 1:
+            # Slice from current sublist start to the next sublist start
+            end_index = indices[idx + 1]
+            segment = main_list[start_index:end_index]
+        else:
+            # For the last sublist occurrence, include the rest of the list
+            segment = main_list[start_index:]
+        result.append(segment)
+
+    return result
 
 #==================================================================================
 
-def find_matches(src_array, trg_array):
-
-    matches = np.all(src_array == trg_array, axis=1)
-    matching_indices = np.where(matches)[0]
-    
-    return matching_indices.tolist()
-
-#==================================================================================
-
-def find_closest_tuple(tuples_list, src_tuple):
-    
-    def euclidean_distance(t1, t2):
-        return sum((a - b) ** 2 for a, b in zip(t1, t2)) ** 0.5
-
-    closest_tuple = None
-    min_distance = float('inf')
-
-    for t in tuples_list:
-        distance = euclidean_distance(t, src_tuple)
-        if distance < min_distance:
-            min_distance = distance
-            closest_tuple = t
-
-    return closest_tuple
+def fuzzy_find(values, threshold=256):
+    result = set()
+    for i in range(len(values)):
+        for j in range(i+1, len(values)):
+            if abs(values[i] - values[j]) <= threshold:
+                result.add(values[j])
+                result.add(values[i])
+    return sorted(result)
 
 #==================================================================================
 
-def find_best_midx(midi_loops, midxs, trg_midx):
-
-    all_midxs = midxs + [trg_midx]
-    sidxs = [int(midx // 6) for midx in all_midxs]
-
-    times_durs = []
-
-    for sidx in sidxs:
-        
-        score = midi_loops[sidx][2]
-        dtimes = [e[0] for e in score if e[0] != 0]
-        durs = [e[1] for e in score]
-
-        avg_dtime = int(sum(dtimes) / len(dtimes))
-        avg_dur = int(sum(durs) / len(durs))
-
-        mode_dtimes= statistics.mode(dtimes)
-        mode_durs = statistics.mode(durs)
-
-        times_durs.append([avg_dtime, avg_dur, mode_dtimes, mode_durs])
-
-    best_time_dur = find_closest_tuple(times_durs[:-1], times_durs[-1])
-
-    best_midx = midxs[times_durs.index(best_time_dur)]
-
-    return best_midx
+def kmp_search(main_list, sublist):
+    if not sublist:
+        return -1, -1  # Sublist is empty
+    
+    # Preprocess the sublist to get the longest prefix-suffix array
+    lps = [0] * len(sublist)
+    length = 0  # Length of the previous longest prefix suffix
+    i = 1
+    while i < len(sublist):
+        if sublist[i] == sublist[length]:
+            length += 1
+            lps[i] = length
+            i += 1
+        else:
+            if length != 0:
+                length = lps[length - 1]
+            else:
+                lps[i] = 0
+                i += 1
+    
+    # Start the search
+    i = j = 0  # Index for main_list and sublist
+    while i < len(main_list):
+        if main_list[i] == sublist[j]:
+            i += 1
+            j += 1
+        if j == len(sublist):
+            start_index = i - j
+            end_index = i - 1
+            return start_index, end_index
+        elif i < len(main_list) and main_list[i] != sublist[j]:
+            if j != 0:
+                j = lps[j - 1]
+            else:
+                i += 1
+    return -1, -1  # Sublist not found
 
 #==================================================================================
 
-def Mix_Loops(max_num_loops,
-              comp_loops_mult,
-              chords_chunks_len,
-              loops_chords_set_len
-             ):
+def Remix_MIDI(max_num_loops,
+               comp_loops_mult,
+               chords_chunks_len,
+               loops_chords_set_len
+              ):
 
     #===============================================================================
     
@@ -456,6 +463,10 @@ with gr.Blocks() as demo:
             """)
     
     #==================================================================================
+
+    gr.Markdown("## Upload source MIDI or select an example MIDI below")
+    
+    input_midi = gr.File(label="Input MIDI", file_types=[".midi", ".mid", ".kar"])
     
     gr.Markdown("## Mixing options")
     
@@ -473,7 +484,7 @@ with gr.Blocks() as demo:
     output_plot = gr.Plot(label="MIDI score plot")
     output_midi = gr.File(label="MIDI file", file_types=[".mid"])
 
-    mix_btn.click(Mix_Loops, 
+    mix_btn.click(Remix_MIDI, 
                    [
                     max_num_loops,
                     comp_loops_mult,