Update app.py

app.py

@@ -1,3 +1,65 @@
 import gradio as gr
+from transformers import pipeline
+import sounddevice as sd  # For microphone input
+from diffusers import DiffusionPipeline
 
-gr.load("models/MdEndan/tinysketch-fine-tuned").launch()
+# Load the diffuser pipeline with LORA weights
+pipeline = DiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4")
+pipeline.load_lora_weights("MdEndan/tinysketch-fine-tuned")
+
+def generate_image(text):
+    """Converts speech to text, generates an image using diffuser pipeline,
+       and displays the result."""
+
+    # Speech-to-text using a pre-trained pipeline (replace with your choice)
+    speech_pipe = pipeline("automatic-speech-recognition")
+    try:
+        # Record audio from microphone (adjust duration and sample rate if needed)
+        duration = 5  # Record for 5 seconds
+        fs = 16000  # Sample rate
+        print("Speak now...")
+        myrecording = sd.rec(duration * fs, samplerate=fs, channels=1)
+        sd.wait()
+        print("Recording stopped")
+
+        # Convert audio to WAV for compatibility with some pipelines
+        sd.write("recording.wav", myrecording, fs)
+
+        # Transcribe speech
+        with open("recording.wav", "rb") as f:
+            audio_bytes = f.read()
+        speech_output = speech_pipe(audio_bytes, return_tensors="pt")["sequences"]
+        text = speech_output[0].tolist()  # Extract the transcribed text
+
+    except Exception as e:
+        print(f"Error during speech recognition: {e}")
+        text = "Error: Speech recognition failed."
+
+    # Ensure text input is a string
+    if not isinstance(text, str):
+        text = str(text)
+
+    # Generate image using diffuser pipeline
+    try:
+        image = pipeline(text).images[0]
+        return image
+
+    except Exception as e:
+        print(f"Error during image generation: {e}")
+        return None
+
+# Gradio interface with microphone and image display
+interface = gr.Interface(
+    fn=generate_image,
+    inputs=gr.Audio(sources=["microphone"]),
+    outputs=gr.Image(thumbnail=True),
+    title="Speak & Create: Text-to-Image with Microphone Input (LORA)",
+    description="Speak your description and see an image generated using a fine-tuned model!",
+)
+
+# Handle potential errors during Gradio launch
+try:
+    # Request access to the microphone (might require user permission)
+    interface.launch(share=True, capture_audio=True)
+except Exception as e:
+    print(f"Error launching Gradio interface: {e}")