Rename LLMwithvoice.py to app.py

LLMwithvoice.py → app.py

@@ -2,6 +2,7 @@ import requests
 from pydub import AudioSegment
 from io import BytesIO
 import gradio as gr
+import tempfile
 
 # Hugging Face API URLs
 API_URL_ROBERTA = "https://api-inference.huggingface.co/models/deepset/roberta-base-squad2"
@@ -38,10 +39,9 @@ def generate_speech(api_token, answer):
         response.raise_for_status()  # Raise an error for bad responses
         audio = response.content
         audio_segment = AudioSegment.from_file(BytesIO(audio), format="wav")
-        audio_file = BytesIO()
-        audio_segment.export(audio_file, format="wav")
-        audio_file.seek(0)
-        return audio_file
+        temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
+        audio_segment.export(temp_file.name, format="wav")
+        return temp_file.name
     except requests.exceptions.HTTPError as e:
         print(f"HTTP error occurred: {e}")
         return None
@@ -55,5 +55,25 @@ def gradio_interface(api_token, context, prompt):
     if 'error' in answer:
         return answer['error'], None
     answer_text = answer.get('answer', 'No answer found')
-    audio_file = generate_speech(api_token, answer_text)
-    return answer_text, audio_file
+    audio_file_path = generate_speech(api_token, answer_text)
+    return answer_text, audio_file_path
+
+# Define the Gradio interface
+iface = gr.Interface(
+    fn=gradio_interface,
+    inputs=[
+        gr.Textbox(type="password", lines=1, label="Hugging Face API Token", placeholder="Enter your Hugging Face API token here..."),
+        gr.Textbox(lines=2, label="Context", placeholder="Enter the context here..."),
+        gr.Textbox(lines=1, label="Question", placeholder="Enter your question here...")
+    ],
+    outputs=[
+        gr.Textbox(label="Answer"),
+        gr.Audio(label="Answer as Speech", type="filepath")  # Changed to filepath type
+    ],
+    title="Chat with Roberta with Voice",
+    description="Ask questions based on a provided context using the Roberta model and hear the response via text-to-speech."
+)
+
+# Launch the Gradio app
+if __name__ == "__main__":
+    iface.launch()