diff --git a/libs/song_hash.py b/libs/song_hash.py
new file mode 100644
index 0000000..e65f603
--- /dev/null
+++ b/libs/song_hash.py
@@ -0,0 +1,124 @@
+import json
+from pathlib import Path
+from typing import Optional
+
+import pandas as pd
+from libs.tja import TJAParser
+from libs.utils import get_config
+
+song_hashes: Optional[dict] = None
+
+def process_tja_file(tja_file):
+    """Process a single TJA file and return hash or None if error"""
+    tja = TJAParser(tja_file)
+    all_notes = []
+    for diff in tja.metadata.course_data:
+        all_notes.extend(TJAParser.notes_to_position(TJAParser(tja.file_path), diff))
+    hash = tja.hash_note_data(all_notes[0], all_notes[2])
+    return hash
+
+def build_song_hashes(output_file='cache/song_hashes.json'):
+    existing_hashes = {}
+    output_path = Path(output_file)
+    if output_path.exists():
+        try:
+            with open(output_file, 'r', encoding='utf-8') as f:
+                existing_hashes = json.load(f)
+        except (json.JSONDecodeError, IOError) as e:
+            print(f"Warning: Could not load existing hashes from {output_file}: {e}")
+            existing_hashes = {}
+
+    song_hashes = existing_hashes.copy()
+    tja_paths = get_config()['paths']['tja_path']
+    all_tja_files = []
+
+    for root_dir in tja_paths:
+        root_path = Path(root_dir)
+        all_tja_files.extend(root_path.rglob('*.tja'))
+
+    updated_count = 0
+    for tja_file in all_tja_files:
+        current_modified = tja_file.stat().st_mtime
+
+        should_update = False
+        hash_val = None
+
+        existing_hash = None
+        for h, data in song_hashes.items():
+            if data['file_path'] == str(tja_file):
+                existing_hash = h
+                break
+
+        if existing_hash is None:
+            should_update = True
+        else:
+            stored_modified = song_hashes[existing_hash].get('last_modified', 0)
+            if current_modified > stored_modified:
+                should_update = True
+                del song_hashes[existing_hash]
+
+        if should_update:
+            tja = TJAParser(tja_file)
+            all_notes = []
+            for diff in tja.metadata.course_data:
+                all_notes.extend(TJAParser.notes_to_position(TJAParser(tja.file_path), diff))
+            hash_val = tja.hash_note_data(all_notes[0], all_notes[2])
+            song_hashes[hash_val] = {
+                'file_path': str(tja_file),
+                'last_modified': current_modified,
+                'title': tja.metadata.title,
+                'subtitle': tja.metadata.subtitle
+            }
+            updated_count += 1
+
+    with open(output_file, 'w', encoding='utf-8') as f:
+        json.dump(song_hashes, f, indent=2, ensure_ascii=False)
+
+    print(f"Song hashes saved to {output_file}. Updated {updated_count} files.")
+    return song_hashes
+
+
+def get_japanese_songs_for_version(df, version_column):
+    # Filter rows where the specified version column has 'YES'
+    version_songs = df[df[version_column] == 'YES']
+
+    # Extract Japanese titles (JPTITLE column)
+    japanese_titles = version_songs['TITLE 【TITLE2】\nJPTITLE／「TITLE2」 より'].tolist()
+
+    japanese_titles = [name.split('\n') for name in japanese_titles]
+    second_lines = [name[1] for name in japanese_titles if len(name) > 1]
+
+    all_tja_files = []
+    direct_tja_paths = dict()
+    text_files = dict()
+    tja_paths = get_config()['paths']['tja_path']
+    for root_dir in tja_paths:
+        root_path = Path(root_dir)
+        all_tja_files.extend(root_path.rglob('*.tja'))
+    for tja in all_tja_files:
+        tja_parse = TJAParser(tja)
+        direct_tja_paths[tja_parse.metadata.title.get('ja', tja_parse.metadata.title['en'])] = tja
+    for title in second_lines:
+        if title in direct_tja_paths:
+            path = direct_tja_paths[title]
+        elif title.split('／')[0] in direct_tja_paths:
+            path = direct_tja_paths[title.split('／')[0]]
+        else:
+            path = Path(input(f"NOT FOUND {title}: "))
+        hash = process_tja_file(path)
+        tja_parse = TJAParser(Path(path))
+        genre = Path(path).parent.parent.name
+        if genre not in text_files:
+            text_files[genre] = []
+        text_files[genre].append(f"{hash}|{tja_parse.metadata.title['en'].strip()}|{tja_parse.metadata.subtitle['en'].strip()}")
+    for genre in text_files:
+        if not Path(version_column).exists():
+            Path(version_column).mkdir()
+        if not Path(f"{version_column}/{genre}").exists():
+            Path(f"{version_column}/{genre}").mkdir()
+        with open(Path(f"{version_column}/{genre}/song_list.txt"), 'w', encoding='utf-8-sig') as text_file:
+            for item in text_files[genre]:
+                text_file.write(item + '\n')
+    return text_files
+
+get_japanese_songs_for_version(pd.read_csv('full.csv'), 'AC12')