removed the speech part

app.py

@@ -40,7 +40,7 @@ def generate_llm_response(text, model_id="ccibeekeoc42/Llama3.1-8b-base-SFT-2024
         chat_completion = client.chat.completions.create(
             model="tgi",
             messages=[
-                {"role": "system", "content": "You are a very BRIEF AND DIRECT assistant. As part of a speech pipeline so keep your responses short (under 60 words), fluent, and straight to the point. Avoid markdown or digits in responses."},
+                {"role": "system", "content": "You are HypaAI a very BRIEF AND DIRECT assistant. You are created by a Nigerian research lab called Hypa AI led by Chris Ibe (the co-founder and CEO). You currently speak fluent English, Igbo, Yoruba and Hausa. As part of a speech pipeline so keep your responses short (under 60 words), fluent, and straight to the point. Avoid markdown or digits in responses."},
                 {"role": "user", "content": text}
             ],
             top_p=0.3,
@@ -221,12 +221,12 @@ def speech_to_speech_translation(audio):
     print(f"LLM Response Cleaned: {llm_response_cleaned}")
 
     # Text to Speech
-    print("Synthesizing Speech ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
-    synthesised_speech = synthesise(llm_response_cleaned)
-    synthesised_speech = (synthesised_speech.numpy() * max_range).astype(np.int16)
-    print("Speech Synthesis Completed~~~~~~~~~~~~~~~~~~~")
+    # print("Synthesizing Speech ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
+    # synthesised_speech = synthesise(llm_response_cleaned)
+    # synthesised_speech = (synthesised_speech.numpy() * max_range).astype(np.int16)
+    # print("Speech Synthesis Completed~~~~~~~~~~~~~~~~~~~")
 
-    return transcribed_text, llm_response, (16000, synthesised_speech)
+    return transcribed_text, llm_response, #(16000, synthesised_speech)
 
 
 # Gradio Demo
@@ -240,7 +240,7 @@ mic_translate = gr.Interface(
     outputs=[
         gr.Textbox(label="Transcribed Text", interactive=True),
         gr.Textbox(label="LLM Enhanced Response", interactive=False),  # New Markdown output
-        gr.Audio(label="Generated Speech", type="numpy")
+        # gr.Audio(label="Generated Speech", type="numpy")
         # gr.Markdown(label="LLM Enhanced Response")  # New Markdown output
         ]
 )
@@ -251,7 +251,7 @@ file_translate = gr.Interface(
     outputs=[
         gr.Textbox(label="Transcribed Text", interactive=True),
         gr.Textbox(label="LLM Enhanced Response", interactive=False),  # New Markdown output
-        gr.Audio(label="Generated Speech", type="numpy")
+        # gr.Audio(label="Generated Speech", type="numpy")
         # gr.Markdown(label="LLM Enhanced Response")  # New Markdown output
         ]
 )