Update app.py

app.py

@@ -9,12 +9,8 @@ import jiwer
 
 # ASR part
 from transformers import pipeline
-# p = pipeline("automatic-speech-recognition")
-p = pipeline(
-    "automatic-speech-recognition",
-    model="KevinGeng/whipser_medium_en_PAL300_step25",
-    device=0,
-)
+p = pipeline("automatic-speech-recognition")
+
 # WER part
 transformation = jiwer.Compose([
     jiwer.ToLowerCase(),
@@ -92,7 +88,7 @@ Add WER interface.
 
 iface = gr.Interface(
   fn=calc_mos,
-  inputs=[gr.Audio(type='filepath', label="Audio to evaluate"), 
+  inputs=[gr.Audio(type='microphone', label="Audio to evaluate"), 
           gr.Textbox(placeholder="Input reference here (Don't keep this empty)", label="Reference")],
   outputs=[gr.Textbox(placeholder="Naturalness evaluation, ranged 1 to 5, the higher the better.", label="Predicted MOS"),
            gr.Textbox(placeholder="Hypothesis", label="Hypothesis"),