Add file upload feature

app.py

@@ -32,7 +32,7 @@ OUTPUT_FAIL = (
 """
 )
 
-THRESHOLD = 0.70
+THRESHOLD = 0.80
 
 model_name = "nvidia/speakerverification_en_titanet_large"
 model = EncDecSpeakerLabelModel.from_pretrained(model_name).to(device)
@@ -64,8 +64,11 @@ inputs = [
     gr.inputs.Audio(source="microphone", type="filepath", optional=True, label="Speaker #1"),
     gr.inputs.Audio(source="microphone", type="filepath", optional=True, label="Speaker #2"),
 ]
-output = gr.outputs.HTML(label="")
 
+upload_inputs = [
+    gr.inputs.Audio(source="upload", type="filepath", optional=True, label="Speaker #1"),
+    gr.inputs.Audio(source="upload", type="filepath", optional=True, label="Speaker #2"),
+]
 
 description = (
     "This demonstration will analyze two recordings of speech and ascertain whether they have been spoken by the same individual.\n"
@@ -85,10 +88,24 @@ examples = [
     ["data/id10270_5r0dWxy17C8-00002.wav", "data/id10271_1gtz-CUIygI-00002.wav"],
 ]
 
-interface = gr.Interface(
+microphone_interface = gr.Interface(
     fn=compare_samples,
     inputs=inputs,
-    outputs=output,
+    outputs=gr.outputs.HTML(label=""),
+    title="Speaker Verification with TitaNet Embeddings",
+    description=description,
+    article=article,
+    layout="horizontal",
+    theme="huggingface",
+    allow_flagging=False,
+    live=False,
+    examples=examples,
+)
+
+upload_interface = gr.Interface(
+    fn=compare_samples,
+    inputs=upload_inputs,
+    outputs=gr.outputs.HTML(label=""),
     title="Speaker Verification with TitaNet Embeddings",
     description=description,
     article=article,
@@ -98,4 +115,7 @@ interface = gr.Interface(
     live=False,
     examples=examples,
 )
-interface.launch(enable_queue=True)
+
+demo = gr.TabbedInterface([microphone_interface, upload_interface], ["Microphone", "Upload File"])
+
+demo.launch(enable_queue=True)