app_dcase.py: first commit of this file which now simulates calling DCASE baseline model

app_dcase.py

@@ -0,0 +1,53 @@
+
+import gradio as gr
+from msclap import CLAP
+
+clap_model = CLAP(version = 'clapcap', use_cuda=False)
+
+def clap_inference(mic=None, file=None):
+
+    if mic is not None:
+        audio = mic
+    elif file is not None:
+        audio = file
+    else:
+        return "You must either provide a mic recording or a file"
+
+    # Generate captions for the recording
+    captions = clap_model.generate_caption([audio], 
+                                           resample=True, 
+                                           beam_size=5, 
+                                           entry_length=67, 
+                                           temperature=0.01)
+
+    return captions[0]
+
+
+def create_app():
+
+    with gr.Blocks() as demo:
+        gr.Markdown(
+            """
+            # DCASE demo for automatic audio captioning
+            """
+        )
+        gr.Interface(
+            fn=clap_inference,
+            inputs=[
+                gr.Audio(sources="microphone", type="filepath"),
+                gr.Audio(sources="upload", type="filepath"),
+            ],
+            outputs="text",
+        )
+
+    return demo
+
+def main():
+    
+    app = create_app()
+    app.launch(debug=True)
+
+    
+if __name__ == "__main__":
+    main()
+