Update app.py

app.py

@@ -24,7 +24,7 @@ def update_passage(passage_name):
         
     return sum, "", gr.update(visible=False), gr.update(visible=False), gr.update(visible=False)
         
-def inference(audio):
+def inference(audio, gt: str):
     audio = whisper.load_audio(audio)
     audio_length = audio.shape[-1]/16000
     audio = whisper.pad_or_trim(audio)
@@ -45,16 +45,16 @@ def inference(audio):
     ]) 
     
     error = jiwer.wer(
-        passage.value,
+        gt,
         result.text, 
         truth_transform=transformation, 
         hypothesis_transform=transformation
        )
     # error = jiwer.wer(passage, result.text)
-    we_num = error * len(passage.value.split())
+    we_num = error * len(gt.split())
     # print(f"WER is {we_num}")
     print(result.text)
-    print(passage.value)
+    print(gt)
     
     return f"For a {audio_length} second audio, {we_num} errors were made, resulting in a final time of {audio_length + we_num}.\n\n{result.text}", gr.update(visible=True), gr.update(visible=True), gr.update(visible=True)
 
@@ -229,7 +229,7 @@ with block:
 
 
         
-        btn.click(inference, inputs=[audio], outputs=[text, community_icon, loading_icon, share_button])
+        btn.click(inference, inputs=[audio, passage.value], outputs=[text, community_icon, loading_icon, share_button])
         btn2.click(update_passage, inputs=[passage_name], outputs=[passage, text, community_icon, loading_icon, share_button])
         share_button.click(None, [], [], _js=share_js)