Update app.py

app.py

@@ -11,23 +11,26 @@ description = "Gradio Demo for fairseq S2S: speech-to-speech translation models.
 article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2107.05604' target='_blank'>Direct speech-to-speech translation with discrete units</a> | <a href='https://github.com/facebookresearch/fairseq/tree/main/examples/speech_to_speech' target='_blank'>Github Repo</a></p>"
 
 examples = [
-  ["enhanced_direct_s2st_units_audios_es-en_set2_source_12478_cv.flac","xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022"],
+  [None,"enhanced_direct_s2st_units_audios_es-en_set2_source_12478_cv.flac","xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022"],
 ]
 
 io1 = gr.Interface.load("huggingface/facebook/xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022", api_key=st.secrets["api_key"])
    
-def inference(text,model):
-   outtext = io1(text)   
-   return outtext 
+def inference(mic,file,model):
+    if mic is not None and file is None:
+        audio = mic
+    elif file is not None and mic is None:
+        audio = file
+    else:
+        return "ERROR: You must and may only select one method, it cannot be empty or select both methods at once."
+    out_audio = io1(audio)   
+    return out_audio 
 
 
 gr.Interface(
-    inference, 
-    inputs=[
-        gr.Audio(source="microphone", type="filepath", optional=True, label="Input"),
-        gr.Audio(source="upload", type="filepath", optional=True, label="Input"),
-        gr.inputs.Dropdown(choices=["xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022"], default="xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022", type="value", label="Model"),
-    ], 
+    inference,
+    [gr.inputs.Audio(source="microphone", type="filepath", optional=True, label="Input"),gr.inputs.Audio(source="upload", type="filepath", optional=True, label="Input"),gr.inputs.Dropdown(choices=["xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022"], type="value", label="model")
+],
     gr.outputs.Audio(label="Output"),
     article=article,
     title=title,