Initial commit with FastAPI + Gradio app

app.py

@@ -14,20 +14,8 @@ verification = SpeakerRecognition.from_hparams(
 app = FastAPI()
 
 # Function to calculate similarity score
-def get_similarity(file1_data, file2_data):
+def get_similarity(file1_path, file2_path):
     try:
-        # Create temporary files for the uploaded audio
-        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmpfile1, \
-             tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmpfile2:
-            
-            # Write audio data to the temporary files
-            tmpfile1.write(file1_data)
-            tmpfile2.write(file2_data)
-
-            # Get the file paths
-            file1_path = tmpfile1.name
-            file2_path = tmpfile2.name
-
         # Use `verify_files` to compare the audio files
         score, prediction = verification.verify_files(file1_path, file2_path)
 
@@ -47,12 +35,20 @@ async def compare_voices_api(file1: UploadFile = File(...), file2: UploadFile =
     Compare two audio files and return the similarity score and prediction.
     """
     try:
-        # Read the uploaded file data
-        file1_data = await file1.read()
-        file2_data = await file2.read()
+        # Create temporary files for the uploaded audio
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmpfile1, \
+             tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmpfile2:
+            
+            # Write audio data to the temporary files
+            tmpfile1.write(await file1.read())
+            tmpfile2.write(await file2.read())
+
+            # Get the file paths
+            file1_path = tmpfile1.name
+            file2_path = tmpfile2.name
 
-        # Call the get_similarity function with file data
-        result = get_similarity(file1_data, file2_data)
+        # Call the get_similarity function with file paths
+        result = get_similarity(file1_path, file2_path)
         
         return result
 
@@ -64,8 +60,8 @@ def gradio_interface():
     return gr.Interface(
         fn=compare_voices_api,  # FastAPI function is wrapped here
         inputs=[
-            gr.Audio(type="file", label="First Audio File"),  # Audio file input
-            gr.Audio(type="file", label="Second Audio File")  # Audio file input
+            gr.Audio(type="filepath", label="First Audio File"),  # Audio file input
+            gr.Audio(type="filepath", label="Second Audio File")  # Audio file input
         ],
         outputs="json",  # Output as JSON
         live=False  # No live interface, just the API
@@ -75,8 +71,8 @@ def gradio_interface():
 @app.on_event("startup")
 async def startup():
     gr.Interface(fn=compare_voices_api, inputs=[
-        gr.Audio(type="file", label="First Audio File"),  # Audio file input
-        gr.Audio(type="file", label="Second Audio File")  # Audio file input
+        gr.Audio(type="filepath", label="First Audio File"),  # Audio file input
+        gr.Audio(type="filepath", label="Second Audio File")  # Audio file input
     ], outputs="json", live=False).launch(share=True, inline=True)
 
 # Running the FastAPI app with Gradio