rev app

app.py

@@ -1,7 +1,53 @@
 import gradio as gr
+from transformers import pipeline
+import torch
+import spaces
 
-def greet(name):
-    return "Hello " + name + "!!"
+# Initialize model on CPU
+model = pipeline(
+    "automatic-speech-recognition",
+    model="Aekanun/whisper-small-hi",
+    device="cpu"
+)
 
-demo = gr.Interface(fn=greet, inputs="text", outputs="text")
-demo.launch()
+@spaces.GPU
+def transcribe_speech(audio):
+    """Speech transcription with GPU support"""
+    try:
+        if audio is None:
+            return "กรุณาบันทึกเสียงก่อน"
+            
+        # Move model to GPU
+        model.model = model.model.to("cuda")
+        
+        # Make sure input is on the same device as model
+        with torch.cuda.amp.autocast():
+            # Process audio
+            result = model(audio, batch_size=1)
+            
+            # Get text result
+            text = result["text"] if isinstance(result, dict) else result
+            
+        # Move model back to CPU
+        model.model = model.model.to("cpu")
+        torch.cuda.empty_cache()
+        
+        return text
+        
+    except Exception as e:
+        # Make sure model is back on CPU in case of error
+        model.model = model.model.to("cpu")
+        torch.cuda.empty_cache()
+        return f"เกิดข้อผิดพลาด: {str(e)}"
+
+# Create Gradio interface
+demo = gr.Interface(
+    fn=transcribe_speech,
+    inputs=gr.Audio(type="filepath"),  # Simplified Audio component
+    outputs=gr.Textbox(label="ข้อความ"),
+    title="Thai Speech Transcription",
+    description="บันทึกเสียงเพื่อแปลงเป็นข้อความภาษาไทย",
+)
+
+if __name__ == "__main__":
+    demo.queue().launch(server_name="0.0.0.0")