Update app.py

app.py

@@ -62,9 +62,9 @@ def models(text, model="Llama 3 8B Service", seed=42):
     global conversation_history
     seed = int(randomize_seed_fn(seed))
     generator = torch.Generator().manual_seed(seed)  
-    
+
     client = client_fn(model)
-    
+
     if "Llama 3 8B Service" in model:
         messages = [
             {"role": "system", "content": system_instructions1},
@@ -76,21 +76,21 @@ def models(text, model="Llama 3 8B Service", seed=42):
             messages=messages
         )
         assistant_response = completion.choices[0].message.content
-        
+
         # Update conversation history
         conversation_history.append({"role": "user", "content": text})
         conversation_history.append({"role": "assistant", "content": assistant_response})
-        
+
         # Keep only the last 10 messages to avoid token limit issues
         if len(conversation_history) > 20:
             conversation_history = conversation_history[-20:]
-        
+
         return assistant_response
     else:
         # For other models, we'll concatenate the conversation history into a single string
         history_text = "\n".join([f"{'User' if msg['role'] == 'user' else 'Assistant'}: {msg['content']}" for msg in conversation_history])
         formatted_prompt = f"{system_instructions1}\n\nConversation history:\n{history_text}\n\nUser: {text}\nOPTIMUS:"
-        
+
         generate_kwargs = dict(
             max_new_tokens=300,
             seed=seed
@@ -101,15 +101,15 @@ def models(text, model="Llama 3 8B Service", seed=42):
         for response in stream:
             if not response.token.text == "</s>":
                 output += response.token.text
-        
+
         # Update conversation history
         conversation_history.append({"role": "user", "content": text})
         conversation_history.append({"role": "assistant", "content": output})
-        
+
         # Keep only the last 10 messages to avoid token limit issues
         if len(conversation_history) > 20:
             conversation_history = conversation_history[-20:]
-        
+
         return output
 
 async def respond(audio, model, seed):
@@ -141,10 +141,10 @@ def translate_speech(audio_file, target_language):
     """
     if audio_file is None:
         return None
-    
+
     language_code = LANGUAGE_CODES[target_language]
     output_file = "translated_audio.wav"
-    
+
     command = [
         "expressivity_predict",
         audio_file,
@@ -154,7 +154,7 @@ def translate_speech(audio_file, target_language):
         "--gated-model-dir", "models",
         "--output_path", output_file
     ]
-    
+
     subprocess.run(command, check=True)
 
     if os.path.exists(output_file):
@@ -177,7 +177,7 @@ def speech_translation_tab():
 
 with gr.Blocks(css="style.css") as demo:
     description = gr.Markdown("# <center><b>Hello, I am Optimus Prime your personal AI voice assistant</b></center>")
-    
+
     with gr.Tabs() as tabs:
         with gr.TabItem("Voice Assistant") as voice_assistant:
             select = gr.Dropdown([
@@ -198,7 +198,7 @@ with gr.Blocks(css="style.css") as demo:
             value=0,
             visible=False
             )
-            input = gr.Audio(label="User", sources=["microphone"], type="filepath", live=True)
+            input = gr.Audio(label="User", sources=["microphone"], type="filepath")
             output = gr.Audio(label="AI", type="filepath",
                             interactive=False,
                             autoplay=True,
@@ -210,7 +210,7 @@ with gr.Blocks(css="style.css") as demo:
                 outputs=[output],
                 live=True
             )
-        
+
         with gr.TabItem("Speech Translation") as speech_translation:
             input_audio = gr.Audio(label="User", sources=["microphone"], type="filepath")
             target_lang = gr.Dropdown(
@@ -222,7 +222,7 @@ with gr.Blocks(css="style.css") as demo:
                                     interactive=False,
                                     autoplay=True,
                                     elem_classes="audio")
-            
+
             gr.Interface(
                 fn=translate_speech,
                 inputs=[input_audio, target_lang],
@@ -233,5 +233,5 @@ with gr.Blocks(css="style.css") as demo:
     voice_assistant.select(fn=voice_assistant_tab, inputs=None, outputs=description)
     speech_translation.select(fn=speech_translation_tab, inputs=None, outputs=description)
 
-if __name__ == "__main__":
+if name == "main":
     demo.queue(max_size=200).launch()