transcribe test

app.py

@@ -11,7 +11,7 @@ def prepare_input(input_file, progress=gr.Progress()):
     ffmpeg.input(input_file).audio.output(
         output_file, format="wav").run()
     progress(0.4, desc="Acquiring diarization")
-    start_diarization(output_file, progress)
+    start_diarization(output_file)
     progress(0.6, desc="Transcribing audio")
     return start_transcribe(progress)
 

diarization.py

@@ -12,7 +12,7 @@ device = torch.device("cuda")
 pipeline.to(device)
 
 
-def start_diarization(input_file, progress: gr.Progress):
+def start_diarization(input_file):
     diarization = pipeline(input_file)
 
     sample_groups = []

transcribe.py

@@ -15,11 +15,15 @@ def start_transcribe(progress):
         # Transcribe and save temp file
         audiof = f"{speaker}.wav"
         print(f"Loading {audiof}")
-        result = model.transcribe(
+        segments, _ = model.transcribe(
             audio=audiof, language='id', word_timestamps=True)
-        with open(f"{speaker}.json", "w") as text_file:
-            json.dump(result, text_file, indent=4)
-        return result['text']
+        segments_list = list(segments)
+        print("SEGMENT LIST: " + str(segments_list))
+        text = segments_list.join(' ')
+        print("TEXT: " + str(text))
+        # with open(f"{speaker}.json", "w") as text_file:
+        #     json.dump(text, text_file, indent=4)
+        # return result['text']
 
 
 def load_groups_json():