Update app.py

app.py

@@ -5,25 +5,19 @@ from fastapi import FastAPI
 from pydantic import BaseModel
 from faster_whisper import WhisperModel
 import gradio as gr
+import uvicorn # <-- IMPORT THE SERVER
 
 # --- 1. Configuration & Model Loading ---
-# Set cache directories
 os.environ["HF_HOME"] = "/tmp/huggingface_cache"
 os.environ["HF_HUB_CACHE"] = "/tmp/huggingface_cache"
-
-# Load the model once when the application starts
 model = WhisperModel("Systran/faster-whisper-small", device="cpu", compute_type="int8")
 
 # --- 2. FastAPI Application Setup ---
-# Create a FastAPI app instance
 app = FastAPI()
 
-# Define the structure of the incoming API request
 class AudioInput(BaseModel):
-    # Expects a list with one item: a base64 encoded audio data URI
     data: list[str]
 
-# Define the transcription function (this is what the API will use)
 def transcribe_audio(audio_filepath, language):
     if audio_filepath is None:
         return "Error: No audio file provided."
@@ -32,35 +26,30 @@ def transcribe_audio(audio_filepath, language):
     return " ".join(seg.text for seg in segments)
 
 # --- 3. Create the API Endpoint ---
-# This creates an endpoint at the path /predict
 @app.post("/predict")
 async def predict(audio_input: AudioInput):
     # The Gradio API sends data in a list, so we get the first item
     base64_data_uri = audio_input.data[0]
     
-    # Extract the base64 part of the data URI
+    # Handle the null test case from curl
+    if base64_data_uri is None:
+        return {"data": ["Error: No audio file provided."]}
+
     header, encoded_data = base64_data_uri.split(",", 1)
-    
-    # Decode the base64 string into binary audio data
     audio_data = base64.b64decode(encoded_data)
     
-    # Create a temporary file to save the audio, as the model needs a file path
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio_file:
         temp_audio_file.write(audio_data)
         temp_filepath = temp_audio_file.name
         
     try:
-        # Run transcription on the temporary file
-        transcription = transcribe_audio(temp_filepath, "auto") # Using 'auto' for the API
+        transcription = transcribe_audio(temp_filepath, "auto")
     finally:
-        # Clean up and delete the temporary file
         os.remove(temp_filepath)
         
-    # Return the result in the format Gradio expects
     return {"data": [transcription]}
 
 # --- 4. Create the Gradio User Interface ---
-# Note: We are NOT calling iface.launch() here
 iface = gr.Interface(
     fn=transcribe_audio,
     inputs=[
@@ -73,5 +62,8 @@ iface = gr.Interface(
 )
 
 # --- 5. Mount the Gradio UI onto the FastAPI App ---
-# This makes the Gradio UI available at the root path "/"
-app = gr.mount_gradio_app(app, iface, path="/")
+app = gr.mount_gradio_app(app, iface, path="/")
+
+# --- 6. Run the Server (THIS WAS THE MISSING PART) ---
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=int(os.getenv("PORT", 7860)))