Update app.py

app.py

@@ -61,7 +61,7 @@ async def generate_audio_with_voice_prefix(text_segment, default_voice, rate, pi
     voice4_full = "en-GB-ThomasNeural - en-GB (Male)"
     voice4_short = voice4_full.split(" - ")[0]
     voice4F_full ="en-US-EmmaNeural - en-US (Female)"
-    voice4F_short = voice4F_full.split(" - ")[0]
+    voice4F_short = voice4_full.split(" - ")[0]
     voice5_full = "en-GB-RyanNeural - en-GB (Male)" #Old Man
     voice5_short = voice5_full.split(" - ")[0]
     voice6_full = "en-GB-MaisieNeural - en-GB (Female)"  #Child
@@ -221,10 +221,15 @@ async def transcript_to_speech(transcript_text, voice, rate, pitch):
             
             # Check duration until the next timestamp
             if i + 1 < len(lines):
-                next_start_time = int(lines[i+1].split(',')[0].replace(':', '')) * 1000
-                next_start_time_ms = (next_start_time // 1000000 * 3600000) + ((next_start_time % 1000000) // 10000 * 60000) + ((next_start_time % 1000000) % 10000 // 100) * 1000 + (next_start_time % 1000000) % 10000 % 100
-                duration_to_next = next_start_time_ms - start_time
-                
+                next_start_time_line = lines[i+1]
+                next_start_time_match = re.match(r'(\d{2}):(\d{2}):(\d{2}),(\d{3})\s+.*', next_start_time_line)
+                if next_start_time_match:
+                    next_h, next_m, next_s, next_ms = next_start_time_match.groups()
+                    next_start_time_ms = (int(next_h) * 3600000 + int(next_m) * 60000 + int(next_s) * 1000 + int(next_ms))
+                    duration_to_next = next_start_time_ms - start_time
+                else:
+                    duration_to_next = float('inf')  # Or some other large value
+
                 if current_audio_duration > duration_to_next:
                     # Hold and append audio from subsequent lines
                     j = i + 1
@@ -242,10 +247,15 @@ async def transcript_to_speech(transcript_text, voice, rate, pitch):
                             
                             #check duration to the next timestamp.
                             if j + 1 < len(lines):
-                                next_start_time_2 = int(lines[j+1].split(',')[0].replace(':', '')) * 1000
-                                next_start_time_ms_2 = (next_start_time_2 // 1000000 * 3600000) + ((next_start_time_2 % 1000000) // 10000 * 60000 ) + ((next_start_time_2 % 1000000) % 10000 // 100) * 1000 + (next_start_time_2 % 1000000) % 10000 % 100
-                                duration_to_next_2 = next_start_time_ms_2 - start_time
-                                if current_audio_duration <= duration_to_next_2:
+                                next_start_time_line_2 = lines[j+1]
+                                next_start_time_match_2 = re.match(r'(\d{2}):(\d{2}):(\d{2}),(\d{3})\s+.*', next_start_time_line_2)
+                                if next_start_time_match_2:
+                                    next_h_2, next_m_2, next_s_2, next_ms_2 = next_start_time_match_2.groups()
+                                    next_start_time_ms_2 = (int(next_h_2) * 3600000 + int(next_m_2) * 60000 + int(next_s_2) * 1000 + int(next_ms_2))
+                                    duration_to_next_2 = next_start_time_ms_2 - start_time
+                                    if current_audio_duration <= duration_to_next_2:
+                                        break
+                                else:
                                     break
                             j += 1
                         else: