improved UI

app.py

@@ -527,10 +527,15 @@ def build_demo():
 
         with gr.Row():
             with gr.Column():
-                audio_input = gr.Audio(
-                    sources=["upload", "microphone"],
+                audio_upload = gr.Audio(
+                    sources=["upload"],
                     type="filepath",
-                    label="Audio Input",
+                    label="Upload Audio",
+                )
+                audio_mic = gr.Audio(
+                    sources=["microphone"],
+                    type="filepath",
+                    label="Record Audio",
                 )
                 generate_btn = gr.Button("Generate", variant="primary")
 
@@ -538,14 +543,15 @@ def build_demo():
                 gallery = gr.Gallery(label="Generated Images")
                 status = gr.Markdown(visible=False)
 
-        def _generate(audio):
+        def _generate(upload, mic):
+            audio = upload if upload is not None else mic
             imgs, msg = generate(audio, DEFAULT_NUM_DISPLAY, DEFAULT_GUIDANCE_SCALE, DEFAULT_NUM_STEPS, DEFAULT_BASE_SEED)
             visible = bool(msg)
             return imgs, gr.update(value=msg, visible=visible)
 
         generate_btn.click(
             fn=_generate,
-            inputs=[audio_input],
+            inputs=[audio_upload, audio_mic],
             outputs=[gallery, status],
         )