Ignore repeated prompt

faster_whisper/transcribe.py

@@ -1083,6 +1083,7 @@ def generate_segments(
         clip_idx = 0
         seek = seek_clips[clip_idx][0]
         all_tokens = []
+        all_prompt_text = []
         prompt_reset_since = 0
 
         if options.initial_prompt is not None:
@@ -1293,7 +1294,13 @@ def next_words_segment(segments: List[dict]) -> Optional[dict]:
                 if segment["start"] == segment["end"] or not text.strip():
                     continue
 
-                all_tokens.extend(tokens)
+                check_prompt_num = 1
+                if all(
+                    text.strip() != i.strip()
+                    for i in all_prompt_text[-check_prompt_num:]
+                ):
+                    all_tokens.extend(tokens)
+                    all_prompt_text.append(text)
                 idx += 1
 
                 yield Segment(