Update app.py

app.py

@@ -2,54 +2,59 @@ import gradio as gr
 from transformers import pipeline
 from gtts import gTTS
 import tempfile
+import os
 
 # Initialize the speech-to-text transcriber
-from transformers import pipeline
 transcriber = pipeline("automatic-speech-recognition", model="jonatasgrosman/wav2vec2-large-xlsr-53-english")
 
-# Initialize the pre-trained question-answering model
-model_name = "AVISHKAARAM/avishkaarak-ekta-hindi"
-qa_model = pipeline("question-answering", model=model_name)
+# Initialize the question-answering model
+qa_model = pipeline("question-answering", model="AVISHKAARAM/avishkaarak-ekta-hindi")
+
 
 def answer_question(context, question=None, audio=None):
-    # Handle audio input
-    if audio is not None:
-        # Convert audio to text using transcriber
-        transcription_result = transcriber(audio)["text"]
-        question_text = transcription_result
-    else:
-        question_text = question
+    try:
+        # If audio is provided, transcribe it
+        if audio:
+            transcription_result = transcriber(audio)["text"]
+            question_text = transcription_result
+        else:
+            question_text = question
+
+        # Generate an answer to the question
+        qa_result = qa_model(question=question_text, context=context)
+        answer = qa_result["answer"]
+
+        # Convert the answer to speech
+        tts = gTTS(text=answer, lang="en")
+        audio_path = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False).name
+        tts.save(audio_path)
 
-    # Generate the answer using the QA model
-    qa_result = qa_model(question=question_text, context=context)
-    answer = qa_result["answer"]
+        return answer, audio_path
 
-    # Convert the answer to speech using gTTS
-    tts = gTTS(text=answer, lang='en')
-    audio_path = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False).name
-    tts.save(audio_path)
+    except Exception as e:
+        return str(e), None
 
-    return answer, audio_path
 
 # Define the Gradio interface
-context_input = gr.Textbox(label="Context")
+context_input = gr.Textbox(label="Context", lines=3)
 question_input = gr.Textbox(label="Question")
-audio_input = gr.Audio(type="filepath", label="Question Audio")
+audio_input = gr.Audio(type="filepath", label="Question (Audio Input)")
 
 output_text = gr.Textbox(label="Answer")
-output_audio = gr.Audio(label="Answer Audio")
+output_audio = gr.Audio(label="Answer (Audio Output)")
 
-inter = gr.Interface(
+interface = gr.Interface(
     fn=answer_question,
     inputs=[context_input, question_input, audio_input],
     outputs=[output_text, output_audio],
-    title="Question Answering",
-    description="Enter a context and a question to get an answer. You can also record audio with the question.",
+    title="Multimodal Question Answering",
+    description="Provide a context and either a text question or an audio question to get an answer.",
     examples=[
         ["The capital of France is Paris.", "What is the capital of France?", None],
         ["OpenAI is famous for developing GPT-3.", "What is OpenAI known for?", None],
-    ]
+    ],
 )
 
-# Launch the Gradio interface
-inter.launch()
+# Launch the Gradio app
+if __name__ == "__main__":
+    interface.launch()