Update app.py

app.py

@@ -81,25 +81,18 @@ def whisper_speech_demo(multilingual_text, speaker_audio):
     segments = parse_multilingual_text(multilingual_text)
     if not segments:
         return None, "No valid language segments found. Please use the format: <lang> text"
-
     pipe = Pipeline()
     speaker_url = speaker_audio if speaker_audio is not None else None
     audio_segments = []
-
     for lang, text in segments:
         text_str = text if isinstance(text, str) else str(text)
         audio_np = generate_segment_audio(text_str, lang, speaker_url, pipe)
         print("Audio segment shape:", audio_np.shape)  # Debug statement
         audio_segments.append(audio_np)
-
     concatenated_audio = concatenate_audio_segments(audio_segments)
     print("Final concatenated audio shape:", concatenated_audio.shape)  # Debug statement
-
-    # Normalize the concatenated audio
     concatenated_audio = concatenated_audio / np.max(np.abs(concatenated_audio))
-
-    # Return the concatenated audio as a NumPy array
-    return concatenated_audio
+    return concatenated_audio[0]
 
 with gr.Blocks() as demo:
     gr.Markdown(title)