Update app.py

app.py

@@ -321,23 +321,23 @@ async def inference_via_llm_api(input_text, min_new_tokens=2, max_new_tokens=64)
             return "The system got some error during vLLM generation. Please try it again."
 
 async def transcribe_and_speak(audio, source_lang, target_lang, youtube_url=None, target_speaker=None, progress_tracker=None):
-    global transcription_update, translation_update, audio_update, acc_cosy_audio
+    global transcription_update, translation_update, audio_update, acc_cosy_audio,audio_update_event
     transcription_update = {"content": "", "new": False}
     translation_update = {"content": "", "new": False}
     audio_update = {"content": None, "new": False}
     acc_cosy_audio =None
     video_path = None
-    
+    audio_update_event.clear()
     #progress = gr.Progress();
     
     #progress(0.1, "started:")
     if youtube_url:
         audio = await download_youtube_audio(youtube_url)
         if audio is None:
-            return "Failed to download YouTube audio.", None, None, video_path
+            return "Failed to download YouTube audio.", None, None, video_path,(22050, accumulated_audio)
         audio, video_path = audio 
     if not audio:
-        return "Please provide an audio input or a valid YouTube URL.", None, None, video_path
+        return "Please provide an audio input or a valid YouTube URL.", None, None, video_path,(22050, accumulated_audio)
 
     # ASR
     #progress(0.2, "ASR started:")
@@ -358,7 +358,7 @@ async def transcribe_and_speak(audio, source_lang, target_lang, youtube_url=None
                     transcription_update["content"] = transcription
                     transcription_update["new"] = True
             else:
-                return "ASR failed", None, None, video_path
+                return "ASR failed", None, None, video_path,(22050, accumulated_audio)
     #progress(0.4, "ASR done:")
     
 
@@ -455,7 +455,7 @@ async def transcribe_and_speak(audio, source_lang, target_lang, youtube_url=None
     if accumulated_audio is not None:
         return transcription, translated_text, audio_update["content"], video_path, (sample_rate,accumulated_audio)
     else:
-        return transcription, translated_text, "TTS failed", video_path, accumulated_audio
+        return transcription, translated_text, "TTS failed", video_path, (sample_rate, accumulated_audio)
 
 """
 async def run_speech_translation(audio, source_lang, target_lang, youtube_url, target_speaker):
@@ -553,7 +553,7 @@ with gr.Blocks() as demo:
         
             
         temp_video_path = None
-        transcription, translated_text, audio_chunksr, temp_video_path,accumulated_aud_buf = await transcribe_and_speak(audio, source_lang, target_lang, youtube_url, target_speaker)
+        transcription, translated_text, audio_chunksr, temp_video_path, accumulated_aud_buf = await transcribe_and_speak(audio, source_lang, target_lang, youtube_url, target_speaker)
         
         yield (1, 
                gr.update(interactive=True), 
@@ -724,6 +724,6 @@ with gr.Blocks() as demo:
 
 demo.queue()
 
-demo.launch(auth=(os.getenv("DEV_USER"), os.getenv("DEV_PWD")))
-#asyncio.run(demo.launch(auth=(os.getenv("DEV_USER"), os.getenv("DEV_PWD"))))
+#demo.launch(auth=(os.getenv("DEV_USER"), os.getenv("DEV_PWD")))
+asyncio.run(demo.launch(auth=(os.getenv("DEV_USER"), os.getenv("DEV_PWD"))))