fix download dataset

test.py

@@ -1,6 +1,17 @@
+import re
+
 import requests
 from bs4 import BeautifulSoup
 
+
+def filter_string(input_string: str) -> str:
+    allowed_chars = []
+    for j in "АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯабвгдеёжзийклмнопрстуфхцчшщъыьэюя1234567890 !,.?-":
+        allowed_chars.append(j)
+    input_string = re.sub(r'^\d+.\s+', '', input_string)
+    return ''.join([char for char in input_string if char in allowed_chars])
+
+
 repetition = 0
 response = {}
 soup = BeautifulSoup(
@@ -33,9 +44,11 @@ with open('MyTTSDataset/transcript.txt', 'w') as f:
         try:
             response = requests.get(value)
             if response.status_code == 200:
-                with open(f"MyTTSDataset/wavs/wav{index}.wav", 'wb') as file:
-                    file.write(response.content)
-            f.write(f'wav{index}|{" ".join(key.split()[1:])}\n')
+                key = filter_string(key)
+                if key and len(key.replace(" ", "")) > 3:
+                    with open(f"MyTTSDataset/wavs/wav{index}.wav", 'wb') as file:
+                        file.write(response.content)
+                    f.write(f'wav{index}|{key}\n')
         except requests.exceptions.MissingSchema:
             pass
         except requests.exceptions.InvalidSchema: