Update whisper.py

whisper.py

@@ -107,7 +107,9 @@ def transcribe_pipeline(audio, task):
     return text
 
 def generate(audio_path, use_v2):
-    
+    task = "transcribe
+    temp_mono_path = None 
+
     if use_v2:
         split_stereo_channels(audio_path)
 
@@ -126,7 +128,6 @@ def generate(audio_path, use_v2):
         right_segs = [(seg["timestamp"][0], seg["timestamp"][1], "Speaker 2", post_process_transcription(seg["text"])) for seg in right_result["chunks"]]
 
         merged_transcript = sorted(left_segs + right_segs, key=lambda x: x[0])
-        merged_text = " ".join([seg[3] for seg in merged_transcript])
         
         output = ""
         for start, end, speaker, text in merged_transcript:
@@ -134,14 +135,12 @@ def generate(audio_path, use_v2):
         
     else: 
         audio = AudioSegment.from_wav(audio_path)
-        temp_mono_path = None
 
         if audio.channels != 1: #stereo2mono
             audio = audio.set_channels(1)
             temp_mono_path = "temp_mono.wav"
             audio.export(temp_mono_path, format="wav") 
             audio_path = temp_mono_path 
-        task = "transcribe"
         output = transcribe_pipeline(format_audio(audio_path), task)
         
     clean_output = post_process_transcription(output, max_repeats=1) #check